LMK05318RGZR_技术文档

Support &

Community

Product

Folder

Order

Now

Tools &

Software

Technical

Documents

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

LMK05318 具有两个频域的超低抖动网络同步器时钟

1 特性

•

工业温度范围：-40°C 至 +85°C

1

•

一个数字锁相环 (DPLL)，具有：

2 应用

–

无中断切换：±50ps 相位瞬态

•

SyncE (G.8262)、SONET/SDH（Stratum 3/3E、

具有快速锁定功能的可编程环路带宽

G.813、GR-1244、GR-253）、IEEE 1588 PTP

从时钟，或光传输网络 (G.709)

使用低成本 TCXO/OCXO 实现符合标准的同步

和保持模式

•

用于以太网交换机和路由器的 400G 线卡、网络卡

无线基站 (BTS)、无线回程

•

两个具备业界领先性能的模拟锁相环 (APLL)：

–

312.5MHz 频率下 50fs RMS 抖动 (APLL1)

155.52MHz 频率下 125fs RMS 抖动 (APLL2)

测试与测量、医疗成像

56G/112G PAM-4 PHY、ASIC、FPGA、SoC 和

处理器的抖动消除、漂移衰减和基准时钟生成

两个基准时钟输入

–

基于优先级的输入选择

3 说明

在缺失参考时实现数字保持

具有可编程驱动器的八个时钟输出

LMK05318 是一款高性能网络同步器时钟器件，提供

抖动消除、时钟生成、先进的时钟监控和卓越的无中断

切换性能，可满足通信基础设施和工业应用的严格计

时要求。该器件具有超低抖动和高电源噪声抑制

(PSNR) 性能，可降低高速串行链路中的误码率

(BER)。

–

多达 6 个不同的输出频率

AC-LVDS、AC-CML、AC-LVPECL、HCSL 和

1.8V LVCMOS 输出格式

•

加电后自定义时钟的 EEPROM/ROM

灵活的配置选项

–

输入和输出为 1Hz (1PPS) 至 800MHz

XO/TCXO/OCXO 输入：10 至 100MHz

该器件可使用 TI 专有的体声波 (BAW) VCO 技术生成

具有 50fs RMS 抖动的输出时钟，而不受 XO 和基准

输入的抖动和频率的影响。

DCO 模式：< 0.001ppb/阶跃，可进行精确的时

钟控制（IEEE 1588 PTP 从运行）

器件信息⁽¹⁾

–

先进的时钟监控和状态

I²C 或 SPI 接口

器件型号

LMK05318

封装

VQFN (48)

封装尺寸（标称值）

•

PSNR：–83dBc（3.3V 电源下噪声为 50mVpp）

3.3V 电源，提供 1.8V、2.5V 或 3.3V 输出

7.00mm × 7.00mm

(1) 如需了解所有可用封装，请参阅数据表末尾的可订购产品附

录。

简化方框图

VDD

VDDO

3.3 V

1.8 / 2.5 / 3.3 V

Output

Muxes

LMK05318

Ultra-Low Jitter

Power Conditioning

OUT0

OUT1

Network Synchronizer Clock

÷OD

Differential

or HCSL

DPLL

DCO

APLL1

VCO1

PRIREF

Differential

OUT2

OUT3

÷R

or LVCMOS

SECREF

Hitless

Switching

XO/

TCXO/

OCXO

÷OD

OUT4

OUT5

×1, ×2

÷

Differential,

HCSL, or

1.8-V LVCMOS

APLL2

VCO2

EEPROM,

ROM

I²C/SPI

OUT6

OUT7

Registers

÷

LOGIC I/Os

STATUS

Device Control

and Status

1

本文档旨在为方便起见，提供有关 TI 产品中文版本的信息，以确认产品的概要。有关适用的官方英文版本的最新信息，请访问 www.ti.com，其内容始终优先。 TI 不保证翻译的准确

性和有效性。在实际设计之前，请务必参考最新版本的英文版本。

English Data Sheet: SNAS771

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3 Feature Description................................................. 26

9.4 Device Functional Modes........................................ 51

9.5 Programming .......................................................... 57

10 Application and Implementation........................ 64

10.1 Application Information.......................................... 64

10.2 Typical Application ................................................ 68

10.3 Do's and Don'ts..................................................... 73

11 Power Supply Recommendations ..................... 74

11.1 Power Supply Bypassing ...................................... 74

11.2 Device Current and Power Consumption.............. 75

12 Layout................................................................... 76

12.1 Layout Guidelines ................................................. 76

12.2 Layout Example .................................................... 76

12.3 Thermal Reliability................................................. 77

13 器件和文档支持 ..................................................... 78

13.1 器件支持................................................................ 78

13.2 接收文档更新通知 ................................................. 78

13.3 社区资源................................................................ 78

13.4 商标....................................................................... 78

13.5 静电放电警告......................................................... 78

13.6 术语表 ................................................................... 78

14 机械、封装和可订购信息....................................... 79

1

2

3

4

5

6

特性.......................................................................... 1

应用.......................................................................... 1

说明.......................................................................... 1

修订历史记录 ........................................................... 2

（说明（续））....................................................... 3

Pin Configuration and Functions......................... 4

6.1 Device Start-Up Modes............................................. 7

Specifications......................................................... 8

7.1 Absolute Maximum Ratings ...................................... 8

7.2 ESD Ratings.............................................................. 8

7.3 Recommended Operating Conditions....................... 8

7

7.4 Thermal Information: 4-Layer JEDEC Standard

PCB............................................................................ 9

7.5 Thermal Information: 10-Layer Custom PCB............ 9

7.6 Electrical Characteristics........................................... 9

7.7 Timing Diagrams..................................................... 15

7.8 Typical Characteristics............................................ 17

Parameter Measurement Information ................ 19

8.1 Output Clock Test Configurations........................... 19

Detailed Description ............................................ 21

9.1 Overview ................................................................. 21

9.2 Functional Block Diagram ....................................... 22

8

9

4 修订历史记录

注：之前版本的页码可能与当前版本有所不同。

Changes from Original (June 2018) to Revision A

Page

•

将器件状态从“预告信息”更改为“生产数据”.............................................................................................................................. 1

2

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

5 （说明（续））

DPLL 支持抖动和漂移衰减的可编程环路带宽，而两个 APLL 支持分频率转换，从而可以实现灵活的时钟生成。

DPLL 上支持的同步选项包括采用相位消除的无中断切换、数字保持和频率阶跃小于 0.001ppb（十亿分之一）的

DCO 模式，从而实现精确的时钟控制（IEEE 1588 PTP 从运行）。DPLL 可以锁相到 1 PPS（每秒脉冲数）基准

输入，并且在一个输出上支持可选零延迟模式，以实现具有可编程失调电压的确定性输入到输出相位校准。先进的

基准输入监控块可确保稳健的时钟故障检测并在发生基准缺失 (LOR) 时帮助将输出时钟干扰降至最低。

该器件可以使用通用的低频 TCXO 或 OCXO 来根据同步标准设置自由运行型或保持型输出频率稳定性。否则，在

自由运行型或保持型频率稳定性和漂移不重要时，该器件可以使用标准 XO。该器件可通过 I²C 或 SPI 接口实现完

全编程，在通电后支持通过内部 EEPROM 或 ROM 进行自定义频率配置。EEPROM 在出厂时进行了预编程，且

可根据需要进行系统内编程。

请参阅Typical Characteristics，以了解测试条件。

图 1. 312.5MHz 输出相位噪声 (APLL1)，< 50fs RMS 抖动

3

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

6 Pin Configuration and Functions

RGZ Package

48-Pin QFN

Top View

STATUS0

STATUS1/ FDEC

CAP_DIG

1

36

35

34

33

32

31

30

29

28

27

26

25

VDD_PLL2

CAP_PLL2

LF2

2

3

VDD_DIG

4

VDD_XO

XO_N

VDD_IN

5

PRIREF_P

6

XO_P

GND

PRIREF_N

7

GPIO2/SDO/ FINC

LF1

REFSEL

8

HW_SW_CTRL

SECREF_P

SECREF_N

GPIO0/SYNCN

9

CAP_PLL1

VDD_PLL1

SCL/SCK

10

11

12

SDA/SDI

Not to scale

Pin Functions

NAME

POWER

TYPE⁽¹⁾

DESCRIPTION

NO.

Ground / Thermal Pad.

GND

PAD

G

The exposed pad must be connected to PCB ground for proper electrical and thermal performance.

A 5×5 via pattern is recommended to connect the IC ground pad to the PCB ground layers.

Core Supply (3.3 V) for Primary and Secondary Reference Inputs.

Place a nearby 0.1-µF bypass capacitor on each pin.

VDD_IN

5

P

Core Supply (3.3 V) for XO Input.

Place a nearby 0.1-µF bypass capacitor on each pin.

VDD_XO

33

VDD_PLL1

VDD_PLL2

VDD_DIG

27

36

4

P

Core Supply (3.3 V) for PLL1, PLL2, and Digital Blocks.

Place a nearby 0.1-µF bypass capacitor on each pin.

(1) G = Ground, P = Power, I = Input, O = Output, I/O = Input or Output, A = Analog.

4

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Pin Functions (continued)

TYPE⁽¹⁾

DESCRIPTION

NAME

VDDO_01

VDDO_23

VDDO_4

VDDO_5

VDDO_6

VDDO_7

CORE BLOCKS

LF1

NO.

18

19

37

40

43

46

P

Output Supply (1.8, 2.5, or 3.3 V) for Clock Outputs 0 to 7.

Place a nearby 0.1-µF bypass capacitor on each pin.

29

34

A

External Loop Filter Capacitor for APLL1 and APLL2.

Place a nearby capacitor on each pin. For LF1, 0.47-µF is suggested for typical APLL1 loop

bandwidths around 2.5 kHz. For LF2, 0.1-µF is suggested for typical APLL2 loop bandwidth around

500 kHz.

LF2

CAP_PLL1

CAP_PLL2

CAP_DIG

28

35

3

A

External Bypass Capacitors for APLL1, APLL2, and Digital Blocks.

Place a nearby 10-µF bypass capacitor on each pin.

INPUT BLOCKS

PRIREF_P

PRIREF_N

SECREF_P

SECREF_N

XO_P

6

I

DPLL Primary and Secondary Reference Clock Inputs.

Each input pair can accept a differential or single-ended clock as a reference to the DPLL. Each

pair has a programmable input type with internal termination to support AC- or DC-coupled clocks.

A single-ended LVCMOS clock can be applied to the P input with the N input pulled down to

ground. An unused input pair can be left floating.

7

10

11

31

XO/TCXO/OCXO Input.

This input pair can accept a differential or single-ended clock signal from a low-jitter local oscillator

as a reference to the APLLs. This input has a programmable input type with internal termination to

support AC- or DC-coupled clocks. A single-ended LVCMOS clock (up to 2.5 V) can be applied to

the P input with the N input pulled down to ground. A low-frequency TCXO or OCXO can be used to

set the clock output frequency accuracy and stability during free-run and holdover modes.

In DPLL mode, the XO frequency must have a non-integer relationship to the VCO1 frequency so

APLL1 can operate in fractional mode (required for proper DPLL operation). In APLL-only mode, the

XO frequency can have either an integer or non-integer relationship to the VCO1 frequency.

XO_N

32

I

OUTPUT BLOCKS

OUT0_P

OUT0_N

OUT1_P

OUT1_N

OUT2_P

OUT2_N

OUT3_P

OUT3_N

OUT4_P

OUT4_N

OUT5_P

OUT5_N

OUT6_P

OUT6_N

OUT7_P

OUT7_N

14

15

17

16

20

21

23

22

39

38

42

41

45

44

48

47

O

Clock Outputs 0 to 3 Bank.

Each programmable output driver pair can support AC-LVDS, AC-CML, AC-LVPECL and HCSL.

Unused differential outputs should be terminated if active or left floating if disabled through registers.

The OUT[0:3] bank is preferred for PLL1 clocks to minimize output crosstalk.

Clock Outputs 4 to 7 Bank.

Each programmable output driver pair can support AC-LVDS, AC-CML, AC-LVPECL, HCSL, or 1.8-

V LVCMOS clocks (one or two per pair). Unused differential outputs should be terminated if active

or left floating if disabled through registers.

The OUT[4:7] bank is preferred for PLL2 clocks to minimize output crosstalk. When PLL2 is not

used, the OUT[4:7] bank can be used for PLL1 clocks without risk of cross-coupling from PLL2.

5

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Pin Functions (continued)

TYPE⁽¹⁾

DESCRIPTION

NAME

NO.

(2)(3)

LOGIC CONTROL / STATUS

Device Start-Up Mode Select (3-level, 1.8-V compatible).

This input selects the device start-up mode that determines the memory page used to initialize the

registers, serial interface, and logic pin functions. The input level is sampled only at device power-

on reset (POR).

HW_SW_CTRL

PDN

9

I

See Table 1 for start-up mode descriptions and logic pin functions.

Device Power-Down (active low).

When PDN is pulled low, the device is in hard-reset and all blocks including the serial interface are

powered down. When PDN is pulled high, the device is started according to device mode selected

by HW_SW_CTRL and begins normal operation with all internal circuits reset to their initial state.

13

I

I²C Serial Data I/O (SDA) or SPI Serial Data Input (SDI). See Table 1.

When HW_SW_CTRL is 0 or 1, the serial interface is I²C. SDA and SCL pins (open drain) require

external I²C pullup resistors. The default 7-bit I²C address is 11001xxb, where the MSB bits

(11001b) are initialized from on-chip EEPROM and the LSB bits (xxb) are determined by the logic

input pins. When HW_SW_CTRL is 0, the LSBs are determined by the GPIO1 input state (3-level)

during POR. When HW_SW_CTRL is 1, the LSBs are fixed to 00b.

SDA/SDI

25

I/O

When HW_SW_CTRL is Float, the serial interface is SPI (4-wire, Mode 0) using the SDI, SCK,

SCS, and SDO pins.

SCL/SCK

26

12

24

I

I²C Serial Clock Input (SCL) or SPI Serial Clock Input (SCK). See Table 1.

GPIO0/SYNCN

GPIO1/SCS

Multifunction Inputs or Outputs.

See Table 1.

GPIO2/SDO/

FINC

30

1

I/O

STATUS0

Status Outputs 0 and 1.

Each output has programmable status signal selection, driver type (3.3-V LVCMOS or open-drain),

and status polarity. Open-drain requires an external pullup resistor. Leave pin floating if unused.

In I²C mode, the STATUS1/FDEC pin can function as a DCO mode control input pin. See Table 1.

STATUS1/

FDEC

2

8

I/O

I

Manual DPLL Reference Clock Input Selection. (3-level, 1.8-V compatible).

REFSEL = 0 (PRIREF), 1 (SECREF), or Float or V_IM(Auto Select). This control pin must be

enabled by register default or programming. Leave pin floating if unused.

REFSEL

(2) Internal resistors: PDN pin has 200-kΩ pullup to VDD_IN. HW_SW_CTRL, GPIO, REFSEL, and STATUS pins each have a 150-kΩ bias

to V_IM(approximately 0.8 V) when PDN = 0 or 400-kΩ pulldown when PDN = 1.

(3) Unless otherwise noted: Logic inputs are 2-level, 1.8-V compatible inputs. Logic outputs are 3.3-V LVCMOS levels.

6

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

6.1 Device Start-Up Modes

The HW_SW_CTRL input pin selects the device start-up mode that determines the memory page (EEPROM or

ROM) used to initialize the registers, the serial interface, and the logic pin functions at power-on reset. The initial

register settings determine the device's frequency configuration on start-up. After start-up, the device registers

can be accessed through the serial interface for device monitoring and programming, and the logic pins will

function as defined by the selected mode.

Table 1. Device Start-Up Modes

HW_SW_CTRL

INPUT LEVEL⁽¹⁾

START-UP

MODE

MODE DESCRIPTION

Registers are initialized from EEPROM, and I²C interface is enabled.

Logic pins:

•

SDA/SDI, SCL/SCK: I²C Data, I²C Clock (open drain)

EEPROM + I²C

(Soft pin mode)

GPIO0/SYNCN: Output SYNC Input (active low). Pull up externally if not used.

GPIO1/SCS⁽¹⁾: I²C Address LSB Select (Low = 00b, Float = 01b, High = 10b)

GPIO2/SDO/FINC⁽²⁾: DPLL DCO Frequency Increment (active high)

STATUS1/FDEC⁽²⁾: DPLL DCO Frequency Decrement (active high), or Status output

0

Registers are initialized from EEPROM, and SPI interface is enabled.

Logic pins:

•

SDA/SDI, SCL/SCK: SPI Data In (SDI), SPI Clock (SCK)

GPIO0/SYNCN: Output SYNC Input (active low). Pull up externally if not used.

GPIO1/SCS: SPI Chip Select (SCS)

Float

EEPROM + SPI

(Soft pin mode)

(V_IM

)

GPIO2/SDO/FINC: SPI Data Out (SDO)

Registers are initialized from the ROM page selected by GPIO pins, and I²C interface is enabled.

Logic pins:

ROM + I²C

(Hard pin mode)

•

SDA/SDI, SCL/SCK: I²C Data, I²C Clock (open drain)

1

GPIO[2:0] ⁽¹⁾: ROM Page Select Inputs (000b to 111b) during POR.

After POR, GPIO2/SDO/FINC and STATUS1/FDEC pins can function the same as for

HW_SW_CTRL = 0.

(1) The input levels on these pins are sampled only during POR.

(2) FINC and FDEC pins are only available when DCO mode and GPIO pin control are enabled by registers.

NOTE

To ensure proper start-up into EEPROM + SPI Mode, the HW_SW_CTRL, STATUS0, and

STATUS1 pins must all be floating or biased to V_IM(0.8-V typical) before the PDN pin is

pulled high. These three pins momentarily operate as 3-level inputs and get sampled at

the low-to-high transition of PDN to determine the device start-up mode during POR. If

any of these pins are connected to a system host (MCU or FPGA), TI recommends using

external biasing resistors on each pin (10-kΩ pullup to 3.3 V with 3.3-kΩ pulldown to GND)

to set the inputs to V_IMduring POR. After power-up, the STATUS pins can operate as

LVCMOS outputs to overdrive the external resistor bias for normal status operation.

7

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

7 Specifications

7.1 Absolute Maximum Ratings

Over operating free-air temperature range (unless otherwise noted)⁽¹⁾

MIN

-0.3

MAX

UNIT

V

VDD⁽²⁾

VDDO⁽³⁾

V_IN

Core supply voltages

3.6

Output supply voltages

V

Input voltage range for clock and logic inputs

Output voltage range for logic outputs

Output voltage range for clock outputs

Junction temperature

VDD+0.3

VDDO+0.3

150

V

V_{OUT_LOGIC}

V_OUT

V

T_J

°C

Tstg

Storage temperture range

-65

150

(1) Stresses beyond those listed under Absolute Maximum Ratings may cause permanent damage to the device. These are stress ratings

only, which do not imply functional operation of the device at these or any other conditions beyond those indicated under Recommended

Operating Conditions. Exposure to absolute-maximum-rated conditions for extended periods may affect device reliability.

(2) VDD refers to all core supply pins or voltages. All VDD core supplies should be powered-on before the PDN is pulled high to trigger the

internal power-on reset (POR).

(3) VDDO refers to all output supply pins or voltages. VDDO_x refers to the output supply for a specific output channel, where x denotes

the channel index.

7.2 ESD Ratings

VALUE

UNIT

Human body model (HBM), per ANSI/ESDA/JEDEC JS-001, all pins⁽¹⁾

±2000

Electrostatic

discharge

V_(ESD)

V

Charged device model (CDM), per JEDEC specification JESD22-C101, all

pins⁽²⁾

±750

(1) JEDEC document JEP155 states that 500-V HBM allows safe manufacturing with a standard ESD control process.

(2) JEDEC document JEP157 states that 250-V CDM allows safe manufacturing with a standard ESD control process.

7.3 Recommended Operating Conditions

Over operating free-air temperature range (unless otherwise noted)

MIN

3.135

1.71

0

NOM

3.3

MAX

3.465

1.89

3.465

135

UNIT

V

VDD⁽¹⁾

VDDO_x⁽²⁾

V_IN

Core supply voltages

Output supply voltage for AC-LVDS/CML/LVPECL or HCSL driver

Output supply voltage for 1.8-V LVCMOS driver⁽³⁾

Input voltage range for clock and logic inputs

Junction temperature

1.8, 2.5, 3.3

1.8

V

T_J

°C

t_VDD

Power supply ramp time⁽⁴⁾

EEPROM program cycles⁽⁵⁾

0.01

100

ms

cycles

n_EEcyc

100

(1) VDD refers to all core supply pins or voltages. All VDD core supplies should be powered-on before internal power-on reset (POR).

(2) VDDO refers to all output supply pins or voltages. VDDO_x refers to the output supply for a specific output channel, where x denotes

the channel index.

(3) The LVCMOS driver supports full rail-to-rail swing when VDDO_x is 1.8 V ±5%. When VDDO_x is 2.5 V or 3.3 V, the LVCMOS driver

will not fully swing to the positive rail due to the dropout voltage of the output channel's internal LDO regulator.

(4) Time for VDD to ramp monotonically above 2.7 V for proper internal power-on reset. For slower or non-monotonic VDD ramp, hold PDN

low until after VDD voltages are valid.

(5) n_EEcycspecifies the maximum EEPROM program cycles allowed for customer programming. The initial count of factory-programmed

cycles is non-zero due to production tests, but factory-programmed cycles are excluded from the n_EEcyclimit. The total number of

EEPROM program cycles can be read from the 8-bit NVM count status register (NVMCNT), which automatically increments by 1 on

each successful programming cycle. TI does not ensure EEPROM endurance if the n_EEcyclimit is exceeded by the customer.

8

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

7.4 Thermal Information: 4-Layer JEDEC Standard PCB

LMK05318

THERMAL METRIC^{(1) (2) (3)}

RGZ (VQFN)

48 PINS

23.3

UNIT

R_θJA

Junction-to-ambient thermal resistance

Junction-to-case (top) thermal resistance

Junction-to-board thermal resistance

°C/W

R_θJC(top)

R_θJB

R_θJC(bot)

ψ_JT

13.2

7.4

Junction-to-case (bottom) thermal resistance

Junction-to-top characterization parameter

Junction-to-board characterization parameter

1.4

0.2

ψ_JB

7.3

(1) For more information about traditional and new thermal metrics, see the Semiconductor and IC Package Thermal Metrics application

report, SPRA953.

(2) The thermal information is based on a 4-layer JEDEC standard board with 25 thermal vias (5 x 5 pattern, 0.3 mm holes).

(3) Ψ_JBcan allow the system designer to measure the board temperature (T_PCB) with a fine-gauge thermocouple and back-calculate the

device junction temperature, T_J= T_PCB+ (Ψ_JBx Power). Measurement of Ψ_JBis defined by JESD51-6.

7.5 Thermal Information: 10-Layer Custom PCB

LMK05318

THERMAL METRIC^{(1) (2) (3)}

RGZ (VQFN)

UNIT

48 PINS

9.1

R_θJA

R_θJB

ψ_JT

Junction-to-ambient thermal resistance

Junction-to-board thermal resistance

°C/W

4.4

Junction-to-top characterization parameter

Junction-to-board characterization parameter

0.2

ψ_JB

4.4

(1) For more information about traditional and new thermal metrics, see the Semiconductor and IC Package Thermal Metrics application

report, SPRA953.

(2) The thermal information is based on a 10-layer 200 mm x 250 mm x 1.6 mm board with 25 thermal vias (5 x 5 pattern, 0.3 mm holes).

(3) Ψ_JBcan allow the system designer to measure the board temperature (T_PCB) with a fine-gauge thermocouple and back-calculate the

device junction temperature, T_J= T_PCB+ (Ψ_JBx Power). Measurement of Ψ_JBis defined by JESD51-6.

7.6 Electrical Characteristics

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

TEST CONDITIONS

MIN

TYP

MAX

UNIT

POWER SUPPLY CHARACTERISTICS

Core Current Consumption

(VDD_DIG)

IDD_DIG

18

38

mA

Core Current Consumption

(VDD_IN)

IDD_IN

Core Current Consumption

(VDD_PLL1)

IDD_PLL1

DPLL and APLL1 enabled

110

20

Core Current Consumption

(VDD_XO)

IDD_XO

APLL2 disabled

APLL2 enabled

20

mA

Core Current Consumption

(VDD_PLL2)

IDD_PLL2

120

9

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Electrical Characteristics (continued)

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

TEST CONDITIONS

MIN

TYP

MAX

UNIT

Output mux and divider enabled,

excludes driver(s)

50

mA

Divider value = 2 to 6

Output mux and divider enabled,

excludes driver(s)

Divider value > 6

70

mA

Output Current Consumption, per

channel⁽¹⁾

(VDDO_x)

IDDO_x

AC-LVDS

11

14

16

25

6

mA

AC-CML

AC-LVPECL

HCSL, 50-Ω load to GND

1.8-V LVCMOS (x2), 100 MHz

Total Current Consumption (all

VDD and VDDO pins, 3.3 V)

Device powered-down (PDN pin held

low)

IDD_PDN

56

mA

XO INPUT CHARACTERISTICS (XO)

f_IN

Input frequency range

10

1

100

2.6

2

MHz

Vpp

|V|

V_IN-SE

V_IN-DIFF

V_ID

Single-ended input voltage swing LVCMOS input, DC-coupled to XO_P

Differential input voltage swing⁽²⁾Differential input

Input slew rate⁽³⁾

0.4

0.2

40

1

dV/dt

IDC

0.5

V/ns

%

Input duty cycle

60

50-Ω and 100-Ω internal terminations

I_IN

Input leakage

disabled

-350

350

µA

REFERENCE INPUT CHARACTERISTICS (PRIREF, SECREF)

f_IN

Input frequency range

Differential input⁽⁴⁾

5

800

250

MHz

LVCMOS input

1E-6

LVCMOS input, DC-coupled to

xREF_P

V_IN-SE

Single-ended input voltage swing

1

Vpp

V_IN-DIFF

V_ID

Differential input voltage swing⁽²⁾Differential input

Input slew rate⁽³⁾

0.4

0.2

2

1

Vpp

V

dV/dt

0.5

V/ns

50-Ω and 100-Ω internal terminations

disabled

I_IN

Input leakage

-350

350

µA

VCO CHARACTERISTICS

f_VCO1VCO1 Frequency Range

f_VCO2VCO2 Frequency Range

(1) IDDO_x current for an operating output is the sum of mux, divider and an output format.

(2) Minimum limit applies for the minimum setting of the differential input amplitude monitor (xREF_LVL_SEL = 0).

2499.875

5500

2500 2500.125

6250

MHz

(3) In order to meet the jitter performance listed in the subsequent sections of this data sheet, the minimum recommended slew rate for all

input clocks is 0.5 V/ns. This is especially true for single-ended clocks. Phase noise performance will begin to degrade as the clock

input slew rate is reduced. However, the device will function at slew rates down to the minimum listed. When compared to single-

ended clocks, differential clocks (LVDS, LVPECL) will be less susceptible to degradation in phase noise performance at lower slew rates

due to their common mode noise rejection. However, it is also recommended to use the highest possible slew rate for differential clocks

to achieve optimal phase noise performance at the device outputs.

(4) For a differential input clock below 5 MHz, TI recommends to disable the differential input amplitude monitor and enable at least one

other monitor (frequency, window detectors) to validate the input clock. Otherwise, consider using an LVCMOS clock for an input below

5 MHz.

10

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Electrical Characteristics (continued)

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

APLL CHARACTERISTICS

TEST CONDITIONS

MIN

TYP

MAX

UNIT

APLL1 Phase Detector

Frequency

f_PD1

f_PD2

1

50

MHz

APLL2 Phase Detector

Frequency

10

150

AC-LVDS OUTPUT CHARACTERISTICS (OUTx)

f_OUT

V_OD

Output frequency⁽⁵⁾

800

450

MHz

mV

Output voltage swing (V_OH- V_OL

)

f

_OUT≥ 25 MHz; TYP at 156.25 MHz

250

100

350

Differential output voltage swing,

peak-to-peak

V_OUT-DIFF

V_OS

2×V_OD

Vpp

mV

ps

Output common mode

430

100

350

250

Same post divider, output divide

values, and output type

t_SK

Output-to-output skew

20% to 80%, < 300 MHz

225

85

ps

t_R/t_F

Output rise/fall time⁽⁶⁾

± 100 mV around center point, ≥ 300

MHz

ps

PN_FLOOR

ODC

Output phase noise floor

Output duty cycle⁽⁷⁾

f_OUT= 156.25 MHz; f_OFFSET> 10 MHz

-160

dBc/Hz

%

45

55

AC-CML OUTPUT CHARACTERISTICS (OUTx)

f_OUT

V_OD

Output frequency⁽⁵⁾

800

MHz

mV

Output voltage swing (V_OH- V_OL

)

TYP at f_OUT= 156.25 MHz

400

150

600

Differential output voltage swing,

peak-to-peak

V_OUT-DIFF

V_OS

2×V_OD

Vpp

mV

ps

Output common mode

550

100

300

150

Same post divider, output divide

values, and output type

t_SK

Output-to-output skew

20% to 80%, < 300 MHz

225

50

ps

t_R/t_F

Output rise/fall time⁽⁶⁾

± 100 mV around center point, ≥ 300

MHz

ps

PN_FLOOR

ODC

Output phase noise floor

Output duty cycle⁽⁷⁾

f_OUT= 156.25 MHz; f_OFFSET> 10 MHz

-160

dBc/Hz

%

45

55

AC-LVPECL OUTPUT CHARACTERISTICS (OUTx)

f_OUT

V_OD

Output frequency⁽⁵⁾

800

MHz

mV

Output voltage swing (V_OH- V_OL

)

TYP at f_OUT= 156.25 MHz

500

300

800

1000

Differential output voltage swing,

peak-to-peak

V_OUT-DIFF

V_OS

2×V_OD

Vpp

mV

ps

Output common mode

700

100

300

100

Same post divider, output divide

values, and output type

t_SK

Output-to-output skew

20% to 80%, < 300 MHz

200

35

ps

t_R/t_F

Output rise/fall time⁽⁶⁾

± 100 mV around center point, ≥ 300

MHz

ps

PN_FLOOR

ODC

Output phase noise floor

Output duty cycle⁽⁷⁾

f_OUT= 156.25 MHz; f_OFFSET> 10 MHz

-162

dBc/Hz

%

45

55

(5) An output frequency over f_OUTmax spec is possible, but output swing may be less than V_ODmin specification.

(6) Measured on the differential output waveform (OUTx_P - OUTx_N).

(7) Parameter is specified for PLL outputs divided from either VCO domain.

11

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Electrical Characteristics (continued)

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

TEST CONDITIONS

MIN

TYP

MAX

UNIT

HCSL OUTPUT CHARACTERISTICS (OUTx)

f_OUT

V_OH

V_OL

Output frequency⁽⁵⁾

Output high voltage

Output low voltage

400

880

150

MHz

mV

600

-150

mV

Same post divider, output divide

values, and output type

t_SK

Output-to-output skew

Output slew rate⁽⁶⁾

100

4

ps

V/ns

dBc/Hz

%

dV/dt

PN_FLOOR

ODC

± 150 mV around center point

1

Output phase noise floor (f_OFFSET

> 10 MHz)

Output duty cycle⁽⁷⁾

100 MHz

-160

45

55

200

0.4

1.8-V LVCMOS OUTPUT CHARACTERISTICS (OUT[4:7])

f_OUT

V_OH

V_OL

I_OH

Output frequency

Output high voltage

Output low voltage

Output high current

Output low current

Output rise/fall time

1E-6

1.2

MHz

V

I_OH= 1 mA

I_OL= 1 mA

V

-23

20

mA

ps

I_OL

t_R/t_F

20% to 80%

250

Same post divider, output divide

values, and output type

t_SK

Output-to-output skew

100

1.5

ps

ns

Same post divider, output divide

values, LVCMOS-to-DIFF

PN_FLOOR

ODC

Output phase noise floor

Output duty cycle⁽⁷⁾

Output impedance

f_OUT= 66.66 MHz; f_OFFSET> 10 MHz

-160

50

dBc/Hz

45

55

%

R_OUT

Ω

3-LEVEL LOGIC INPUT CHARACTERISTICS (HW_SW_CTRL, GPIO1, REFSEL, STATUS[1:0])

V_IH

Input high voltage

1.4

0.7

V

Input floating with internal bias and

PDN pulled low

V_IM

Input mid voltage

0.9

V_IL

I_IH

I_IL

Input low voltage

Input high current

Input low current

0.4

40

V

V_IH= VDD

V_IL= GND

-40

µA

2-LEVEL LOGIC INPUT CHARACTERISTICS (PDN, GPIO[2:0], SDI, SCK, SCS)

V_IH

V_IL

I_IH

Input high voltage

Input low voltage

Input high current

Input low current

Input Slew Rate

1.2

V

0.6

40

V_IH= VDD

V_IL= GND

-40

0.5

µA

V/ns

I_IL

SR

12

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Electrical Characteristics (continued)

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

TEST CONDITIONS

MIN

TYP

MAX

UNIT

LOGIC OUTPUT CHARACTERISTICS (STATUS[1:0], SDO)

V_OH

V_OL

Output high voltage

Output low voltage

I_OH= 1 mA

I_OL= 1 mA

1.2

V

0.6

20% to 80%, LVCMOS mode, 1 kΩ to

GND

t_R/t_F

Output rise/fall time

500

ps

SPI TIMING REQUIREMENTS (SDI, SCK, SCS, SDO)

SPI clock rate

f_SCK

20

5

MHz

ns

SPI clock rate; NVM write

t₁

t₂

t₃

t₄

t₅

t₆

t₇

t₈

SCS to SCK setup time

SDI to SCK setup time

SDI to SCK hold time

SCK high time

10

25

ns

SCK low time

ns

SCK to SDO valid read-back data

SCS pulse width

20

ns

20

10

ns

SDI to SCK hold time

ns

I²C-COMPATIBLE INTERFACE CHARACTERISTICS (SDA, SCL)

V_IH

V_IL

I_IH

Input high voltage

Input low voltage

Input leakage

1.2

-15

V

0.5

15

µA

V

V_OL

Output low voltage

I_OL= 3 mA

0.3

100

400

Standard

f_SCL

I²C clock rate

kHz

Fast mode

t_SU(START)

t_H(START)

t_W(SCLH)

t_W(SCLL)

t_SU(SDA)

t_H(SDA)

START condition setup time

START condition hold time

SCL pulse width high

SCL pulse width low

SDA setup time

SCL high before SDA low

SCL low after SDA low

0.6

1.3

100

0

µs

ns

µs

ns

µs

SDA hold time

SDA valid after SCL low

0.9

300

t_R(IN)

SDA/SCL input rise time

SDA/SCL input fall time

SDA output fall time

t_F(IN)

t_F(OUT)

C_BUS≤ 400 pF

t_SU(STOP)

STOP condition setup time

0.6

1.3

Bus free time between STOP and

START

t_BUS

µs

13

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Electrical Characteristics (continued)

Over Recommended Operating Conditions (unless otherwise noted)

PARAMETER

TEST CONDITIONS

MIN

TYP

MAX

UNIT

POWER SUPPLY NOISE REJECTION (PSNR) / CROSSTALK SPURS

V_DD= 3.3 V, V_{DDO_x}= 3.3 V, 156.25

MHz, AC-DIFF output

–83

–78

–63

–58

–45

–75

dBc

Spur induced by power supply

noise (V_N= 50 mVpp)

PSNR

(8) (9)

V_DD= 3.3 V, V_{DDO_x}= 3.3 V, 156.25

MHz, HCSL output

V_DD= 3.3 V, V_{DDO_x}= 1.8 V, 156.25

MHz, AC-DIFF output

Spur induced by power supply

noise (V_N= 25 mVpp)^{(8) (9)}

V_DD= 3.3 V, V_{DDO_x}= 1.8 V, 156.25

MHz, HCSL output

V_DD= 3.3 V, V_{DDO_x}= 1.8 V, 156.25

MHz, LVCMOS output

Spur level due to output-to-output f_OUTx= 156.25 MHz, f_OUTy= 155.52

SPUR_XTALK

SPUR

crosstalk (adjacent channels)⁽⁹⁾

MHz, AC-LVPECL

Highest spur level within 12 kHz

to 40 MHz band (excludes output

crosstalk and integer-boundary

spurs)⁽⁹⁾

f_VCO1= 2500 MHz, f_VCO2= 6065.28

MHz, f_OUTx= 156.25 MHz, f_OUTy

155.52 MHz, AC-LVPECL

=

–80

dBc

PLL CLOCK OUTPUT PERFORMANCE CHARACTERISTICS

312.5 MHz AC-DIFF output from

APLL1, f_XO= 48.0048 MHz, f_PD1

f_XO/2, f_VCO1= 2.5 GHz

RMS Phase Jitter (12 kHz to 20

MHz), including spurs⁽¹⁰⁾

RJ

=

50

60

100 fs RMS

156.25 MHz AC-DIFF output from

RMS Phase Jitter (12 kHz to 20

MHz), including spurs⁽¹⁰⁾

APLL1, f_XO= 48.0048 MHz, f_PD1

f_XO/2, f_VCO1= 2.5 GHz

=

153.6 MHz AC-DIFF output from

APLL2, f_XO= 48.0048 MHz, f_PD1

f_XO/2, f_VCO1= 2.5 GHz, f_PD2= f_VCO1/18,

f_VCO2= 5.5296 GHz

RMS Phase Jitter (12 kHz to 20

MHz), including spurs⁽¹⁰⁾

=

RJ

125

250 fs RMS

155.52 MHz AC-DIFF output from

APLL2, f_XO= 48.0048 MHz, f_PD1=

f_XO/2, f_VCO1= 2.5 GHz, f_PD2= f_VCO1/18,

f_VCO2= 5.59872 GHz

RMS Phase Jitter (12 kHz to 20

MHz), including spurs⁽¹⁰⁾

BW

J_PK

DPLL bandwidth range⁽¹¹⁾

Programmed bandwidth setting

0.01

4000

Hz

dB

DPLL closed-loop jitter

peaking⁽¹²⁾

f_REF= 25 MHz, f_OUT= 10 MHz, DPLL

BW = 0.1 Hz or 10 Hz

0.1

Jitter modulation = 10 Hz, 25.78125

Gbps

J_TOL

Jitter tolerance

6455

UI p-p

ps

Valid for a single switchover event

between two clock inputs at the same

frequency

Phase hit between two reference

inputs with 0 ppm error

t_HITLESS

± 50

± 10

Valid for a single switchover event

between two clock inputs at the same

frequency

Frequency transient during hitless

switch

f_HITLESS

ppb

(8) PSNR is the single-sideband spur level (in dBc) measured when sinusoidal noise with ampitude V_Nand frequency f_N(between 100 kHz

and 1 MHz) is injected onto VDD and VDDO_x pins.

(9) DJ_SPUR(ps pk-pk) = [2 × 10^(dBc/20)/ (π × f_OUT) × 1E6], where dBc is the PSNR or SPUR level (in dBc) and f_OUTis the output frequency

(in MHz).

(10) Excluding output coupling spurs

(11) Actual loop bandwidth may be lower. The valid loop bandwidth range may be constrained by the DPLL TDC frequency used in a given

configuration.

(12) DPLL closed-loop jitter peaking of 0.1 dB or less is based on the DPLL bandwidth setting configured by the TICS Pro software tool.

14

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

7.7 Timing Diagrams

t₁

t₄

t₅

SCK

t₂

–

SDI Write/Read

SDO Read

W/R

A14

D0/A0

DON‘T CARE

A13...D1/A1

t₆

DON‘T CARE

D7

D1

D0

t₇

SCS

t₈

图 2. SPI Timing Parameters

ACK

STOP

START

t_W(SCLL)

t_f(SM)

t_W(SCLH)

t_r(SM)

V_IH(SM)

V_IL(SM)

SCL

t_h(START)

t_SU(SDATA)

t_r(SM)

t_h(SDATA)

t_SU(START)

t_BUS

t_SU(STOP)

t_f(SM)

V_IH(SM)

V_IL(SM)

SDA

图 3. I²C Timing Diagram

OUTx_N

V_OH

V_OD= V_OH- V_OL

V_OL

OUTx_P

80%

0 V

20%

V_OUT-DIFF= 2 × V_OD

t_R

t_F

图 4. Differential Output Voltage and Rise/Fall Time

15

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Timing Diagrams (接下页)

80%

V_OUT,SE

OUT_REFx/2

20%

t_R

t_F

图 5. Single-Ended Output Voltage and Rise/Fall Time

INx_P

Single Ended

Differential

INx_N

t_PHO,DIFF

OUTx_P

Differential, PLL

OUTx_N

OUTx_P

OUTx_N

t_SK,DIFF,INT

Differential, PLL

t_{SK,SE-DIFF,INT}

Single Ended, PLL

OUTx_P/N

t_{PHO, SE}

t_SK,SE,INT

OUTx_P/N

Single Ended, PLL

图 6. Differential and Single-Ended Output Skew and Phase Offset

16

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

7.8 Typical Characteristics

Unless otherwise noted: VDD = 3.3 V, VDDO = 1.8 V, T_A= 25 °C, AC-LVPECL output measured.

DPLL: f_REF= 25 MHz, f_TDC= 25 MHz, BW_DPLL= 10 Hz, DPLL locked to reference.

APLL1: f_XO= 48.0048 MHz, f_PD1= 24.0024 MHz (f_XO÷2), f_VCO1= 2500 MHz, BW_APLL1= 2.5 kHz, DPLL mode.

APLL2: f_PD2= 138.8 MHz (f_VCO1÷18), BW_APLL2= 500 kHz, Cascaded APLL2 mode for 图 11 and 图 12.

Jitter = 40 fs RMS (12 kHz to 20 MHz)

DPLL Mode (APLL2 Disabled)

Jitter = 56 fs RMS (12 kHz to 20 MHz)

DPLL Mode (APLL2 Disabled)

图 7. 625-MHz Output Phase Noise (APLL1)

图 8. 156.25-MHz Output Phase Noise (APLL1)

Jitter = 63 fs RMS (12 kHz to 20 MHz)

DPLL Mode (APLL2 Disabled)

Jitter = 74 fs RMS (12 kHz to 20 MHz)

DPLL Mode (APLL2 Disabled)

图 9. 125-MHz Output Phase Noise (APLL1)

图 10. 100-MHz Output Phase Noise (APLL1)

17

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Typical Characteristics (接下页)

APLL2: f_PD2= 138.8 MHz (f_VCO1÷18), BW_APLL2= 500 kHz, Cascaded APLL2 mode for 图 11 and 图 12.

Jitter = 117 fs RMS (12 kHz to 20 MHz)

DPLL Mode With Cascaded APLL2

f_VCO2= 5598.72 MHz

Jitter = 120 fs RMS (12 kHz to 20 MHz)

DPLL Mode With Cascaded APLL2

f_VCO2= 5737.5 MHz

图 11. 155.52-MHz Output Phase Noise (APLL2)

图 12. 212.5-MHz Output Phase Noise (APLL2)

0

-10

-20

-30

-40

-50

-60

-70

-80

-90

-100

0

-10

-20

-30

-40

-50

-60

-70

-80

-90

-100

CML

HCSL

LVCMOS

LVDS

LVPECL

HCSL

LVDS

LVPECL

0

200 400 600 800 1000 1200 1400 1600 1800 2000

0

200 400 600 800 1000 1200 1400 1600 1800 2000

Noise Frequency (Hz)

Noise Frequency (kHz)

D002

D001

50-mVpp noise injected onto supplies (VDD = 3.3 V, VDDO = 3.3

V)

25-mVpp noise injected onto supplies (VDD = 3.3 V, VDDO = 1.8

V)

图 13. PSNR vs. Noise Frequency (50 mVpp)

图 14. PSNR vs. Noise Frequency (25 mVpp)

(1)

For 156.25-MHz Output

(1) DJ_SPUR(ps pk-pk) = 2 × 10^(dBc/20)/ (π × f_OUT) × 1E6, where dBc is the PSNR spur level (in dBc) and f_OUTis the output frequency (in

MHz).

18

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

8 Parameter Measurement Information

8.1 Output Clock Test Configurations

High-impedance

probe

LVCMOS

DUT

Oscilloscope

2 pF

图 15. LVCMOS Output Test Configuration

Phase Noise/

Spectrum

Analyzer

LVCMOS

DUT

图 16. LVCMOS Output Phase Noise Test Configuration

Oscilloscope

(50-ꢀ inputs)

AC-LVPECL, AC-LVDS, AC-CML

DUT

图 17. AC-LVPECL, AC-LVDS, AC-CML Output AC Test Configuration

Phase Noise/

Spectrum Analyzer

DUT

Balun

AC-LVPECL, AC-LVDS, AC-CML

图 18. AC-LVPECL, AC-LVDS, AC-CML Output Phase Noise Test Configuration

0 ꢀ

Oscilloscope

(50-ꢀ inputs)

DUT

HCSL

0 ꢀ

图 19. HCSL Output Test Configuration

19

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Output Clock Test Configurations (接下页)

Opt œ 33 ꢀ

HCSL

Phase Noise/

Spectrum

Analyzer

DUT

Balun

Opt œ 33 ꢀ

HCSL

50 ꢀ

图 20. HCSL Output Phase Noise Test Configuration

Sine wave

Modulator

Power Supply

Phase Noise/

Spectrum

Analyzer

Signal Generator

DUT

Device Output

Balun

Reference

Input

Single-sideband spur level measured in dBc with a known noise amplitude and frequency injected onto the device

power supply.

图 21. Power Supply Noise Rejection (PSNR) Test Configuration

20

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9 Detailed Description

9.1 Overview

The LMK05318 has two reference inputs, one digital PLL (DPLL), two analog PLLs (APLLs) with integrated

VCOs, and eight output clocks. with RMS phase jitter of 50-fs typical from APLL1 and 125-fs typical from APLL2.

APLL1 uses an ultra-high performance BAW VCO (VCO1) with a very high quality factor, and thus has no

dependency on the phase noise or frequency of the external oscillator (XO) input clock. This minimizes the

overall solution cost and allows the use of an off-the-shelf XO, TCXO, or OCXO selected to meet the free-run

and holdover frequency stability requirements of the application. APLL1 is cascaded with the DPLL, allowing the

APLL1 domain to be locked to the DPLL reference input for synchronous clock generation. APLL2 can be used

to generate unrelated clock frequencies either locked to the APLL1 domain or the free-running XO input.

The DPLL reference input mux supports automatic input selection or manual input selection through software or

pin control. The device provides hitless switching with proprietary phase cancellation for superior phase transient

performance (±50 ps typical). The reference clock input monitoring block monitors the clock inputs and will

perform a hitless switchover or holdover when a loss of reference (LOR) is detected. A LOR condition can be

detected upon any violation of the threshold limits set for the input monitors, which include amplitude, frequency,

missing pulse, runt pulse, and 1-PPS (pulse-per-second) detectors. The threshold limits for each input detector

can be set and enabled per clock input. The tuning word history monitor feature allows the initial output

frequency accuracy upon entry into holdover to be determined by the historical average frequency when locked,

minimizing the frequency and phase disturbance during a LOR condition.

The device has eight outputs with programmable drivers, allowing up to eight differential clocks, or a combination

of differential clocks and up to four 1.8-V LVCMOS pairs (two outputs per pair). The output clocks can be

selected from either APLL/VCO domain through the output muxes. The output dividers have a SYNC feature to

allow multiple outputs to be phase-aligned. A 1-PPS output can be supported on Output 7 (OUT7). If needed, the

user can enable the zero-delay mode (ZDM) synchronization to achieve deterministic phase alignment between

an APLL1 clock on OUT7 and the selected reference input.

To support IEEE 1588 PTP slave clock or other clock steering applications, the DPLL also supports DCO mode

with less than 0.001-ppb (part per billion) frequency resolution for precise frequency and phase adjustment

through external software or pin control.

The device is fully programmable through I²C or SPI and supports custom start-up frequency configuration with

the internal EEPROM, which is factory pre-programmed and in-system programmable if needed. Internal LDO

regulators provide excellent PSNR to reduce the cost and complexity of the power delivery network. The clock

input and PLL monitoring status can be observed through the status pins and interrupt registers for full diagnostic

capability.

9.1.1 ITU-T G.8262 (SyncE) Standards Compliance

The LMK05318 meets the applicable requirements of the ITU-T G.8262 (SyncE) standard. See the Application

Report, ITU-T G.8262 Compliance Test Result for the LMK05318 (SNAA316).

21

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.2 Functional Block Diagram

VDDO (x6)

1.8 / 2.5 / 3.3 V

VDD (x5)

3.3 V

Power Conditioning

(all blocks)

Outputs

SYNC

Reference Inputs

APLL1

OUT0

VCO1

÷R

5-b

0

1

2

3

×1, ×2

XO

0

1

÷OD

8-b

PFD

OUT1

OUT2

÷N

40-b Frac-N

PRIREF

0

1

2

3

DPLL

÷OD

8-b

÷R

16-b

TDC

OUT3

OUT4

SECREF

REFSEL

÷FB

40-b Frac-N

Input

Monitors

0

1

2

3

÷OD

8-b

FINC/FDEC

DCO

0

1

2

3

÷RP

/3 to /6

÷RS

/1 to /32

÷OD

8-b

OUT5

OUT6

OUT7

Post

Dividers

APLL2

VCO2

2

3

PFD

/2 to /7

Digital

0

1

2

3

÷OD

8-b

SDA/SDI

SCL/SCK

GPIO2/SDO/FINC

GPIO1/SCS

EEPROM

ROM

I²C/

SPI

÷N

24-b Frac-N

Registers

/2 to /7

0

1

2

3

GPIO0/SYNCN

PLL

÷OD

8-b × 24-b

Monitors

Device Control and Status

HW_SW_CTRL

PDN

STATUS1/FDEC

STATUS0

LF1

LF2

CAP

(x3)

图 22. Top-Level Device Block Diagram

22

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Functional Block Diagram (接下页)

9.2.1 PLL Architecture Overview

图 23 shows the PLL architecture implemented in the LMK05318. The primary "PLL1" channel consists of a

digital PLL (DPLL) and analog PLL (APLL1) with integrated BAW VCO (VCO1) capable of generating clocks with

RMS phase jitter of 50-fs typical. A secondary APLL (APLL2) with integrated LC VCO (VCO2) can be used as an

additional clock generation domain with RMS phase jitter of 125-fs typical.

The DPLL is comprised of a time-to-digital converter (TDC), digital loop filter (DLF), and 40-bit fractional

feedback (FB) divider with sigma-delta-modulator (SDM). The APLLs are comprised of a reference (R) divider,

phase-frequency detector (PFD), loop filter (LF), fractional feedback (N) divider with SDM, and VCO. APLL2 has

a reference selection mux that allows APLL2 to be either locked to APLL1's VCO domain (Cascaded APLL2) or

locked to the XO input (Non-Cascaded APLL2). Otherwise, APLL2 can be disabled (powered-down) if this clock

domain is not needed. APLL1's VCO feeds the output clock distribution blocks directly, whereas APLL2's VCO

drives the clock distribution blocks through its VCO post-dividers.

Post

Dividers

APLL2

VCO

f_PD2

f_PD

VCO2

0

1

f_VCO2/P1

f_VCO2

PFD

LF

÷P1

f_VCO

R Dividers

÷RP

/2 to /7

÷P2

÷N

f_VCO2/P2

/3 to /6

24-bit Frac-N SDM

To

÷RS

Output

Muxes

/1 to /32

×1, ×2

÷R

XO

DPLL

APLL1

PRIREF

0

1

f_PD1

f_TDC

VCO1

÷R

5-bit

f_VCO1

SECREF

TDC

DLF

LF

PFD

16-bit

(x2)

÷FB

40-bit Frac-N SDM

÷N

f_VCO1

÷PR

÷2

40-bit Frac-N SDM

To

Output

Muxes

38-bit

DCO option

DCO

FINC/FDEC

DPLL feedback clock

FDEV

(1) DCO frequency adjustments can be software or pin controlled.

图 23. PLL Architecture

The following sections describe the basic principle of operation for DPLL mode and APLL-only mode. See PLL

Operating Modes for more details on the PLL modes of operation including holdover.

9.2.2 DPLL Mode

In DPLL mode, the external XO input source determines the free-run and holdover frequency stability and

accuracy of the output clocks. The BAW VCO1 determines the APLL1 output clock phase noise and jitter

performance over the 12-kHz to 20-MHz integration band, regardless of the frequency and jitter of the XO input.

This allows the use a cost-effective, low-frequency TCXO or OCXO as the external XO input to support

standards-compliant frequency stability and low loop bandwidth (≤10 Hz) required in synchronization applications

like SyncE and IEEE 1588.

23

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Functional Block Diagram (接下页)

The principle of operation for DPLL mode after power-on reset and initialization is as follows. If APLL2 is in

cascaded mode as shown in 图 24, VCO1 is held at its nominal center frequency of 2.5 GHz while APLL2 locks.

Then APLL1 locks the VCO1 frequency to the external XO input and operates in free-run mode. Once a valid

DPLL reference input is detected, the DPLL begins lock acquisition. The DPLL TDC compares the phase of the

selected reference input clock and the FB divider clock (from VCO1) and generates a digital correction word

corresponding to the phase error. The correction word is filtered by the DLF, and the DLF output controls the

APLL1 N divider SDM to pull the VCO1 frequency into lock with the reference input. VCO2 will track the VCO1

domain during DPLL lock acquisition and locked modes, allowing APLL2's clock domain to be synchronized to

the DPLL reference input. Cascading APLL2 provides a high-frequency, ultra-low-jitter reference clock from

VCO1 to minimize the APLL2 in-band phase noise/jitter impact that would otherwise occur if the APLL2's

reference is from a XO/TCXO/OCXO with low frequency and/or high phase noise floor.

If APLL2 is not cascaded as shown in 图 25, VCO2 will lock to the XO input after initialization and operate

independently of the DPLL/APLL1 domain.

When all reference inputs to the DPLL are lost, the PLLs will enter holdover mode and track the stability and

accuracy of the external XO source.

If DCO mode is enabled on the DPLL, a frequency deviation step value (FDEV) can be programmed and used to

adjust (increment or decrement) the DPLL's FB divider SDM, where the frequency adjustment effectively

propagates through the APLL1 domain (and APLL2 domain if cascaded) to the output clocks.

The programmed DPLL loop bandwidth (BW_DPLL) should be lower than all of the following:

1. 1/100th of the DPLL TDC rate

2. the APLL1 loop bandwidth (1 to 10 kHz typical)

3. the maximum DPLL bandwidth setting of 4 kHz.

Post

Dividers

APLL2

VCO

f_PD2

f_PD

VCO2

0

1

f_VCO2/P1

f_VCO2

PFD

LF

÷P1

f_VCO

R Dividers

÷RP

/2 to /7

÷P2

÷N

f_VCO2/P2

/3 to /6

24-bit Frac-N SDM

To

÷RS

Output

Muxes

/1 to /32

×1, ×2

÷R

XO

DPLL

APLL1

PRIREF

0

1

f_PD1

f_TDC

VCO1

÷R

5-bit

f_VCO1

SECREF

TDC

DLF

LF

PFD

16-bit

(x2)

÷FB

40-bit Frac-N SDM

÷N

f_VCO1

÷PR

÷2

40-bit Frac-N SDM

To

Output

Muxes

38-bit

DCO option

DCO

FINC/FDEC

DPLL feedback clock

FDEV

图 24. DPLL Mode With Cascaded APLL2

24

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Functional Block Diagram (接下页)

Post

Dividers

APLL2

VCO

f_PD2

f_PD

VCO2

0

1

f_VCO2/P1

f_VCO2

PFD

LF

÷P1

f_VCO

R Dividers

÷RP

/2 to /7

÷P2

÷N

f_VCO2/P2

/3 to /6

24-bit Frac-N SDM

To

÷RS

Output

Muxes

/1 to /32

×1, ×2

÷R

XO

DPLL

APLL1

PRIREF

0

1

f_PD1

f_TDC

VCO1

÷R

5-bit

f_VCO1

SECREF

TDC

DLF

LF

PFD

16-bit

(x2)

÷FB

40-bit Frac-N SDM

÷N

f_VCO1

÷PR

÷2

40-bit Frac-N SDM

To

Output

Muxes

38-bit

DCO option

DCO

FINC/FDEC

DPLL feedback clock

FDEV

图 25. DPLL Mode With Non-Cascaded APLL2

25

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Functional Block Diagram (接下页)

9.2.3 APLL-Only Mode

In APLL-only mode, the external XO input source determines the free-run frequency stability and accuracy of the

output clocks. The BAW VCO1 determines the APLL1 output clock phase noise and jitter performance over the

12-kHz to 20-MHz integration band, regardless of the frequency and jitter of the XO input.

The principle of operation for APLL-only mode after power-on reset and initialization is as follows. If APLL2 is in

cascaded mode as shown in 图 26, VCO1 is held at its nominal center frequency of 2.5 GHz while APLL2 locks.

Then APLL1 locks the VCO1 frequency to the external XO input and operates in free-run mode. The DPLL

blocks are not used and do not affect the APLLs. VCO2 will track the VCO1 domain. Cascading APLL2 provides

a high-frequency, ultra-low-jitter reference clock from VCO1 to minimize the APLL2 in-band phase noise/jitter

impact that would occur otherwise if the APLL2's reference is from a XO/TCXO/OCXO with low frequency, high

phase noise floor, or both.

If APLL2 is not cascaded as shown in 图 25, VCO2 will lock to the XO input after initialization and operate

independent of the DPLL/APLL1 domain.

Post

Dividers

APLL2

VCO

f_PD2

f_PD

VCO2

0

1

f_VCO2/P1

f_VCO2

PFD

LF

÷P1

f_VCO

R Dividers

÷RP

/3 to /7

÷P2

÷N

f_VCO2/P2

/3 to /6

24-bit Frac-N SDM

To

÷RS

Output

Muxes

/1 to /32

×1, ×2

÷R

XO

DPLL

APLL1

PRIREF

0

1

f_PD1

f_TDC

VCO1

÷R

5-bit

f_VCO1

SECREF

TDC

DLF

LF

PFD

16-bit

(x2)

÷FB

40-bit Frac-N SDM

÷N

f_VCO1

÷PR

÷2

40-bit Frac-N SDM

To

Output

Muxes

38-bit

DCO option

DCO

FINC/FDEC

DPLL feedback clock

FDEV

图 26. APLL-Only Mode With Cascaded APLL2

9.3 Feature Description

The following sections describe the features and functional blocks of the LMK05318.

9.3.1 Oscillator Input (XO_P/N)

The XO input is the reference clock for the fractional-N APLLs. The XO input determines the output frequency

accuracy and stability in free-run or holdover modes.

For DPLL mode, the XO frequency must have a non-integer relationship with the VCO1 frequency so APLL1

can operate in fractional mode. For APLL-only mode, the XO frequency can have an integer or fractional

relationship with the VCO1 and/or VCO2 frequencies.

26

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Feature Description (接下页)

In DPLL mode applications, such as SyncE and IEEE 1588, the XO input can be driven by a low-frequency

TCXO, OCXO, or external traceable clock that conforms to the frequency accuracy and holdover stability

required by the applicable synchronization standard. TCXO and OCXO frequencies of 12.8, 19.2, 19.44, 24,

24.576, and 30.72 MHz are commonly available and cost-effective options that allow the APLL1 to operate in

fractional mode for a VCO1 frequency of 2.5 GHz.

An XO/TCXO/OCXO source with low frequency or high phase jitter/noise floor will have no impact on the output

jitter performance because the BAW VCO determines the jitter and phase noise over the 12-kHz to 20-MHz

integration bandwidth.

The XO input buffer has programmable input on-chip termination and AC-coupled input biasing configurations as

shown in 图 27. The buffered XO path also drives the input monitoring blocks.

28 pF

XO_P

100 kꢀ

S1

S2

V_AC-DIFF

(weak bias)

Differential or

Single-Ended*

S3

50 ꢀ

100 ꢀ

XO path

S2

100 kꢀ

28 pF

XO_ N

*Supports 2.5-V

single-ended swing

S1

50 ꢀ

图 27. XO Input Buffer

表 2 lists the typical XO input buffer configurations for common clock interface types.

表 2. XO Input Buffer Modes

INTERNAL SWITCH SETTINGS

XO_TYPE

INPUT TYPES

INTERNAL TERM. (S1, S2)⁽¹⁾

INTERNAL BIAS (S3)⁽²⁾

LVDS, CML, LVPECL

(DC-coupled)

0h

1h

3h

4h

8h

Ch

OFF

LVDS, CML, LVPECL

(AC-coupled)

ON (1.3 V)

OFF

LVDS, CML, LVPECL

(AC-coupled, internal 100-Ω)

100 Ω

50 Ω

OFF

HCSL

(DC-coupled, internal 50-Ω)

LVCMOS

(DC-coupled)

OFF

Single-ended

(DC-coupled, internal 50-Ω)

50 Ω

OFF

(1) S1, S2: OFF = External termination is assumed.

(2) S3: OFF = External input bias or DC coupling is assumed.

27

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3.2 Reference Inputs (PRIREF_P/N and SECREF_P/N)

The reference inputs (PRIREF and SECREF) can accept differential or single-ended clocks. Each input has

programmable input type, termination, and AC-coupled input biasing configurations as shown in 图 28. Each

input buffer drives the reference input mux of the DPLL block. The DPLL input mux can select from any of the

reference inputs. The DPLL can switch between inputs with different frequencies provided they can be divided-

down to a common frequency by DPLL R dividers. The reference input paths also drive the various detector

blocks for reference input monitoring and validation.

To LVCMOS input

slew rate detector

3.6 kꢀ

S4

28 pF

PRIREF_P/

SECREF_P

100 kꢀ

S1

S2

Differential or

Single-Ended*

V_AC-DIFF

(weak bias)

S3

50 ꢀ

100 ꢀ

REF path

S2

100 kꢀ

28 pF

PRIREF_ N/

SECREF_N

*Supports 3.3-V

S-E input swing

S1

50 ꢀ

图 28. Reference Input Buffer

表 3 lists the reference input buffer configurations for common clock interface types.

表 3. Reference Input Buffer Modes

INTERNAL SWITCH SETTINGS

LVCMOS SLEW

RATE DETECT

(S4)⁽³⁾

REFx_TYPE

INPUT TYPES

INTERNAL TERM.

(S1, S2)⁽¹⁾

INTERNAL BIAS

(S3)⁽²⁾

LVDS, CML, LVPECL

(DC-coupled)

0h

1h

3h

4h

8h

Ch

OFF

ON (1.3 V)

OFF

ON

LVDS, CML, LVPECL

(AC-coupled)

LVDS, CML, LVPECL

(AC-coupled, internal 100-Ω)

100 Ω

50 Ω

OFF

HCSL

(DC-coupled, internal 50-Ω)

LVCMOS

(DC-coupled)

OFF

Single-ended

(DC-coupled, internal 50-Ω)

50 Ω

OFF

ON

(1) S1, S2: OFF = External termination is assumed.

(2) S3: OFF = External input bias or DC coupling is assumed.

(3) S4: OFF = Differential input amplitude detector is used for all input types except LVCMOS or Single-ended.

28

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.3 Clock Input Interfacing and Termination

图 29 through 图 35 show the recommended input interfacing and termination circuits. Unused clock inputs can

be left floating or pulled down.

VDD

Rs

R1

R2

LVCMOS

Driver

XO_P

XO_N

50 W

LMK05318

(R_OUT

)

Rs = 50 œ R_OUT

VDD

3.3 V

2.5 V

1.8 V

R1 (ꢀ) R2 (ꢀ)

125

0

375

open

0

图 29. Single-Ended LVCMOS to XO Input (XO_P)

Rs

LVCMOS

3.3V LVCMOS

Driver

LMK05318

图 30. Single-Ended LVCMOS (1.8, 2.5, 3.3 V) to Reference (PRIREF_P/SECREF_P)

Vcco

LVPECL Driver

LVPECL

LMK05318

50 ꢀ

Vcco œ 2 V

图 31. DC-Coupled LVPECL to Reference (PRIREF_P/SECREF_P) or XO Inputs

LMK05318

LVDS Driver

100 ꢀ

LVDS

图 32. DC-Coupled LVDS to Reference (PRIREF/SECREF) or XO Inputs

29

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

CML

Driver

LMK05318

CML

图 33. DC-Coupled CML (Source Terminated) to Reference (PRIREF/SECREF) or XO Inputs

50 ꢀ

HCSL

Driver

LMK05318

HCSL

50 ꢀ

图 34. HCSL (Load Terminated) to Reference (PRIREF/SECREF) or XO Inputs

Driver

LVDS

R_B(ꢀ)

open

150

LMK05318

100 ꢀ

Differential

Driver

CML*

3.3-V LVPECL

2.5-V LVPECL

HCSL

82

Internal input biasing

R_B

50

*CML driver has 50-ꢀ pull-up

图 35. AC-Coupled Differential to Reference (PRIREF/SECREF) or XO Inputs

30

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.4 Reference Input Mux Selection

For the DPLL block, the reference input mux selection can be done automatically using an internal state machine

with a configurable input priority scheme, or manually through software register control or hardware pin control.

The input mux can select from PRIREF or SECREF. The priority for all inputs can be assigned through registers.

The priority ranges from 0 to 2, where 0 = ignore (never select), 1 = first priority, and 2 = second priority. When

both inputs are configured with the same priority setting, PRIREF will be given first priority. The selected input

can be monitored through the status pins or register.

9.3.4.1 Automatic Input Selection

There are two automatic input selection modes that can be set by register: Auto Revertive and Auto Non-

Revertive.

•

Auto Revertive: In this mode, the DPLL automatically selects the valid input with the highest configured

priority. If a clock with higher priority becomes valid, the DPLL will automatically switch over to that clock

immediately.

•

Auto Non-Revertive: In this mode, the DPLL automatically selects the highest priority input that is valid. If a

higher priority input because valid, the DPLL will not switch-over until the currently selected input becomes

invalid.

9.3.4.2 Manual Input Selection

There are two manual input selection modes that can be set by a register: Manual with Auto-Fallback and

Manual with Auto-Holdover. In either manual mode, the input selection can be done through register control (see

表 4) or hardware pin control (see 表 5).

•

Manual with Auto-Fallback: In this mode, the manually selected reference is the active reference until it

becomes invalid. If the reference becomes invalid, the DPLL will automatically fallback to the highest priority

input that is valid or qualified. If no prioritized inputs are valid, the DPLL will enter holdover mode (if tuning

word history is valid) or free-run mode. The DPLL will exit holdover mode when the selected input becomes

valid.

•

Manual with Auto-Holdover: In this mode, the manually selected reference is the active reference until it

becomes invalid. If the reference becomes invalid, the DPLL will automatically enter holdover mode (if tuning

word history is valid) or free-run mode. The DPLL will exit holdover mode when the selected input becomes

valid.

表 4. Manual Input Selection by Register Bits

DPLL_REF_MAN_REG_SEL BIT

DPLLx_REF_MAN_SEL BIT

SELECTED INPUT

PRIREF

0

1

0

SECREF

表 5. Manual Input Selection by Hardware Pins

REFSEL PIN

DPLL_REF_MAN_SEL BIT

SELECTED INPUT

PRIREF

0

1

Float (V_IM

)

Auto Select

SECREF

1

31

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

The reference input selection flowchart is shown in 图 36.

See Device POR and

PLL Initialization and

DPLL Modes Flowcharts

DPLL

Locked

Yes: With

Auto-Holdover

Yes: Auto

Revertive

No

Input Select Mode

= Manual?

Input Select Mode

= Auto?

Yes: With

Auto-Fallback

LOR on

Selected Input, or

Higher Priority Input

Valid?

Yes: Auto

Non-Revertive

No

Loss of Ref (LOR)

on Selected Input?

Yes

No

Loss of Ref (LOR) on

Selected Input?

Yes

Holdover Mode

No

Higher Priority

Input Valid?

Manually Selected

Input Valid?

Yes: Auto-Switch

according to Priority

settings

Yes: Switch to

Selected Input

Lock Acquisition

(Fastlock, Hitless Switch)

图 36. Reference Input Selection Flowchart

9.3.5 Hitless Switching

The DPLL supports hitless switching through TI's proprietary phase cancellation scheme. When hitless switching

is enabled, it will prevent a phase transient (phase hit) from propagating to the outputs when the two switched

inputs have a fixed phase offset and are frequency-locked. The inputs are frequency-locked when they have

same exact frequency (0-ppm offset), or have frequencies that are integer-related and can each be divided to a

common frequency by integers. When hitless switching is disabled, a phase hit equal to the phase offset

between the two inputs will be propagated to the output at a rate determined by the DPLL fastlock bandwidth.

The hitless switching specifications (t_HITLESSand f_HITLESS) are valid for reference inputs with no wander. In the

case where two inputs are switched but are not frequency-locked, the output smoothly transitions to the new

frequency with reduced transient.

9.3.5.1 Hitless Switching With 1-PPS Inputs

Hitless switching between 1-PPS inputs is supported when zero-delay mode (ZDM) synchronization is disabled,

but the switchover event should only occur after the DPLL has acquired lock. If a switchover occurs before the

DPLL has locked initially, the switchover will not be hitless and the DPLL will take an indeterminate amount of

time to lock. In this case, a soft-reset should be issued for the DPLL to lock to the selected input. In an

application, the system host can monitor the DPLL lock status through a STATUS pin or bit to determine when

the DPLL has locked before allowing a switchover between 1-PPS inputs. The DPLL lock time is governed by the

DPLL bandwidth (typically 10 mHz for a 1-PPS input).

Hitless switching between 1-PPS inputs is not supported when ZDM synchronization is enabled.

32

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.6 Gapped Clock Support on Reference Inputs

The DPLL supports locking to an input clock that has missing periods and is referred to as a gapped clock.

Gapping severely increases the jitter of a clock, so the DPLL provides the high input jitter tolerance and low loop

bandwidth necessary to generate a low-jitter periodic output clock. The resulting output will be a periodic non-

gapped clock with an average frequency of the input with its missing cycles. The gapped clock width cannot be

longer than the reference clock period after the R divider (R_PRI/SECREF/ f_PRI/SECREF). The reference input monitors

should be configured to avoid any flags due to the worst-case clock gapping scenario to achieve and maintain

lock. Reference switchover between two gapped clock inputs may violate the hitless switching specification if the

switch occurs during a gap in either input clock.

9.3.7 Input Clock and PLL Monitoring, Status, and Interrupts

The following section describes the input clock and PLL monitoring, status, and interrupt features.

XO

Status Bits

EN

LOS_XO

Frequency

XO Input Monitor

Ref Inputs

PRIREF

REF

Mux

÷R

PLLs

Clock Status

SECREF

Ref Input Monitors (x2)

EN

DIFF: Min. Swing

LVCMOS: Slew rate

Amplitude

Frequency

EN

Valid / Invalid ppm

Late detect window

Early detect window

Jitter threshold

LOR

Validation Timer

Starts when LOR‰0

PRI/SECREF

Valid

Missing pulse

Runt pulse

LOR_AMP

DPLL

Selected

Input

LOR_FREQ

LOR_MISSCLK

REFSWITCH

Valid time

Phase valid*

5

Detector Status (1 = fault)

PRI/SECREF

Status

*Enable for 1-PPS input

图 37. Clock Monitors for Reference and XO Inputs

9.3.7.1 XO Input Monitoring

The XO input has a coarse frequency monitor to help qualify the input before it is used to lock the APLLs.

The XO frequency detector clears its LOS_XO flag when the input frequency is detected within the supported

range of 10 MHz to 100 MHz. The XO frequency monitor uses a RC-based detector and cannot precisely detect

if the XO input clock has sufficient frequency stability to ensure successful VCO calibration during the PLL start-

up when the external XO clock has a slow or delayed start-up behavior. See Slow or Delayed XO Start-Up for

more information.

The XO frequency detector can be bypassed by setting the XO_FDET_BYP bit (shown as EN in 图 37) so that

the XO input is always considered valid by the PLL control state machine. The user can observe the LOS_XO

status flag through the status pins and status bit.

33

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3.7.2 Reference Input Monitoring

Each DPLL reference clock input is independently monitored for input validation before it is qualified and

available for selection by the DPLL. The reference monitoring blocks include amplitude, frequency, missing

pulse, and runt pulse monitors. For a 1-PPS input, the phase valid monitor and LVCMOS input amplitude monitor

are supported, while the differential input amplitude, frequency, missing pulse, and runt pulse monitors are not

supported and must be disabled. A validation timer sets the minimum time for all enabled reference monitors to

be clear of flags before an input is qualified.

The enablement and valid threshold for all reference monitors and validation timers are programmable per input.

The reference monitors and validation timers are optional to enable, but are critical to achieve reliable DPLL lock

and optimal transient performance during holdover or switchover events, and are also used to avoid selection of

an unreliable or intermittent clock input. If a given detector is not enabled, it will not set a flag and will be ignored.

The status flag of any enabled detector can be observed through the status pins for any reference input (selected

or not selected). The status flags of the enabled detectors can also be read through the status bits for the

selected input of the DPLL.

9.3.7.2.1 Reference Validation Timer

The validation timer sets the amount of time for each reference to be clear of flags from all enabled input

monitors before the timer is qualified and valid for selection. The validation timer and enable settings are

programmable.

9.3.7.2.2 Amplitude Monitor

The reference amplitude detector determines if the input meets the amplitude-related threshold depending on the

input buffer configuration. For differential input mode, the amplitude detector clears its LOR_AMP flag when the

differential input voltage swing (peak-to-peak) is greater than the minimum threshold selected by the registers

(400, 500, or 600 mVpp nominal). For LVCMOS input mode, the input slew rate detector clears its LOR_AMP

flag when its slew rate is faster than 0.2 V/ns on the clock edge selected by the registers (rising edge, falling

edge, or both edges). If either the differential or LVCMOS input clock does not meet the specified thresholds, the

amplitude detector will set the LOR_AMP flag and disqualify the input.

If the input frequency is below 5 MHz, the differential input detector may signal a false flag. In this case, the

amplitude detector should be disabled and at least one other input monitor (frequency, window, or 1-PPS phase

valid detector) should be enabled to validate the input clock. The LVCMOS input detector can be used for low-

frequency clocks down to 1 Hz or 1 PPS.

9.3.7.2.3 Frequency Monitoring

The precision frequency detector measures the frequency offset or error (in ppm) of all input clocks relative to the

XO input's frequency accuracy, which is considered as the "0-ppm reference clock" for frequency comparison.

The valid and invalid ppm frequency thresholds are configurable through the registers. The monitor will clear the

LOR_FREQ flag when the relative input frequency error is less than the valid ppm threshold. Otherwise, the

monitor will set the LOR_FREQ flag when the relative input frequency error is greater than the invalid ppm

threshold. The ppm delta between the valid and invalid thresholds provides hysteresis to prevent the LOR_FREQ

flag from toggling when the input frequency offset is crossing these thresholds.

A measurement accuracy (ppm) and averaging factor are used in computing the frequency detector register

settings. A higher measurement accuracy (smaller ppm) or higher averaging factor will increase the

measurement delay to set or clear the flag, which allow more time for the input frequency to settle, and can also

provide better measurement resolution for an input with high drift or wander. Note that higher averaging reduces

the maximum frequency ppm thresholds that can be configured.

9.3.7.2.4 Missing Pulse Monitor (Late Detect)

The missing pulse monitor uses a window detector to validate input clock pulses that arrive within the nominal

clock period plus a programmable late window threshold (T_LATE). When an input pulse arrives before T_LATE, the

pulse is considered valid and the missing pulse flag will be cleared. When an input pulse does not arrive before

T_LATE(due to a missing or late pulse), the flag will be set immediately to disqualify the input.

34

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Typically, T_LATEshould be set higher than the input's longest clock period (including cycle-to-cycle jitter), or

higher than the gap width for a gapped clock. The missing pulse monitor can act as a coarse frequency detector

with faster detection than the ppm frequency detector. The missing pulse monitor is supported for input

frequencies between 2 kHz and f_VCO1/12 and should be disabled when outside this range.

The missing pulse and runt pulse monitors operate from the same window detector block for each reference

input. The status flags for both these monitors are combined by logic-OR gate and can be observed through

status pin. The window detector flag for the selected DPLL input can also be observed through the corresponding

MISSCLK status bit.

9.3.7.2.5 Runt Pulse Monitor (Early Detect)

The runt pulse monitor uses a window detector to validate input clock pulses that arrive within the nominal clock

period minus a programmable early window threshold (T_EARLY). When an input pulse arrives after T_EARLY, the

pulse is considered valid and the runt pulse flag will be cleared. When an early or runt input pulse arrives before

T_EARLY, the monitor will set the flag immediately to disqualify the input.

Typically, T_EARLYshould be set lower than the input's shortest clock period (including cycle-to-cycle jitter). The

early pulse monitor can act as a coarse frequency detector with faster detection than the ppm frequency

detector. The early pulse monitor is supported for input frequencies between 2 kHz and f_VCO1/12 and should be

disabled when outside of this range.

Ideal Reference Period

Ideal Edge

Ideal Reference Input

(rising-edge triggered)

Early Pulse (Input disqualified at this input rising edge)

Example A: Input with

Early (Runt) Pulse

Late Pulse (Input disqualified after T_LATE

)

Example B: Input with

Missing (Late) Pulse

Gapped Clock (To avoid disqualifying input at the

missing clock cycle, set T_LATEwindow > Gap width)

Example C: Input with

Missing (Gapped) Clock

Gap width

Valid

Invalid

Valid Windows

Valid Window size can be relaxed by increasing the Window size.

Window Step Size = 2 / f_VCO1

Early Window

(T_EARLY

)

Late Window

(T_LATE

)

Minimum Valid Window

is ±3 × (2 / f_VCO1

)

图 38. Early and Late Window Detector Examples

9.3.7.2.6 Phase Valid Monitor for 1-PPS Inputs

The phase valid monitor is designed specifically for 1-PPS input validation because the frequency and window

detectors do not support this mode. The phase valid monitor uses a window detector to validate 1-PPS input

pulses that arrive within the nominal clock period (T_IN) plus a programmable jitter threshold (T_JIT). When the input

pulse arrives within the counter window (T_V), the pulse is considered valid and the phase valid flag will be

cleared. When the input pulse does not arrive before T_V(due to a missing or late pulse), the flag will be set

immediately to disqualify the input. T_JITshould be set higher than the worst-case input cycle-to-cycle jitter.

35

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Ideal Edge

(T_IN< T_V

Counter resets at

valid edge (T_IN‘ < T_V)

Counter time-out (T_IN‘‘ > T_V).

Input is disqualified here

)

Late Pulse

(Large peak jitter)

Ideal Input Period

T_IN

T_IN‘

T_IN‘‘

T_IN‘ > T_IN

T_IN‘‘ >> T_IN

Example:

1-PPS Input

T_JIT

Valid Counter (T_V)

T_V= T_IN+ T_JIT

T_V

图 39. 1-PPS Input Window Detector Example

9.3.7.3 PLL Lock Detectors

The loss-of-lock (LOL) status is available for each APLL and the DPLL. The APLLs are monitored for loss-of-

frequency lock only. The DPLL is monitored for both loss-of-frequency lock (LOFL) and loss-of-phase lock

(LOPL). The DPLL lock threshold and loss-of-lock threshold are programmable for both LOPF and LOFL

detectors.

The DPLL frequency lock detector will clear its LOFL flag when the DPLL's frequency error relative the selected

reference input is less than the lock ppm threshold. Otherwise, it will set the LOFL flag when the DPLL's

frequency error is greater than the unlock ppm threshold. The ppm delta between the lock and unlock thresholds

provides hysteresis to prevent the LOFL flag from toggling when the DPLL frequency error is crossing these

thresholds.

A measurement accuracy (ppm) and averaging factor are used in computing the frequency lock detector register

settings. A higher measurement accuracy (smaller ppm) or higher averaging factor will increase the

measurement delay to set or clear the LOFL flag. Higher averaging may be useful when locking to an input with

high wander or when the DPLL is configured with a narrow loop bandwidth. Note that higher averaging reduces

the maximum frequency ppm thresholds that can be configured.

The DPLL phase lock detector will clear its LOPL flag when the phase error of the DPLL is less than the phase

lock threshold. Otherwise, the lock detector will set the LOPL flag when the phase error is greater than the phase

unlock threshold.

Users can observe the APLL and DPLL lock detector flags through the status pins and the status bits.

36

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

PLLs

Status Bits

2

DPLL Frequency Lock

Detector

APLL Lock

Detectors

LOL_PLL[1:2]

Lock

Unlock

LOFL

LOL_PLL1

APLL

LOL_PLL2

Thresh

(ppm)

Thresh

(ppm)

XO

APLL1

APLL2

f_VCO1

f_TDC

DPLL

PLLs Status

Free-run

Tuning Word

LOFL_DPLL

DPLL Phase Lock

Detector

LOPL_DPLL

LOPL

DPLL

HIST

Lock

Unlock

Tuning Word History

History

HLDOVR

Update

Count

Delay

Holdover

Active

Thresh

(ns)

Thresh

(ns)

EN Average Ignore

time time

图 40. PLL Lock Detectors and History Monitor

37

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3.7.4 Tuning Word History

The DPLL domain has a tuning word history monitor block that determines the initial output frequency accuracy

upon entry into holdover. The tuning word can be updated from one of three sources depending on the DPLL

operating mode:

a. Locked Mode: From the output of the digital loop filter when locked

b. Holdover Mode: From the final output of the history monitor

c. Free Run Mode: From the free-run tuning word register (user defined)

When the history monitor is enabled and the DPLL is locked, it effectively averages the reference input frequency

by accumulating history from the digital loop filter output during a programmable averaging time (T_AVG). Once the

input becomes invalid, the final tuning word value is stored to determine the initial holdover frequency accuracy.

Generally, a longer T_AVGtime will produce a more accurate initial holdover frequency. The stability of the 0-ppm

reference clock (XO input) determines the long-term stability and accuracy of the holdover output frequency.

There is also a separate programmable delay timer (T_IGN) that can be set to ignore the history data that is

corrupted just prior to entry into holdover. The history data could be corrupted if a tuning word update occurs

while the input clock is failing and before it is detected by the input monitors. Both T_AVGand T_IGNtimes are

programmable through the HISTCNT and HISTDLY register bits, respectively, and are related to the TDC rate.

The tuning word history is initially cleared after a device hard reset or soft reset. After the DPLL locks to a new

reference, the history monitor waits for the first T_AVGtimer to expire before storing the first tuning word value and

begins to accumulate history. The history monitor will not clear the previous history value during reference

switchover or holdover exit. The history can be manually cleared or reset by toggling the history enable bit

(HIST_EN = 1 → 0 → 1), if needed.

Initial start of history

Ref Lost

Ref Valid

when LOFL‰ 0 only

LOR‰ 1

LOR‰ 0

History

Reset

No History

History Data Accumulating

History Valid

History Data Accumulating

T_AVG(0)

T_AVG(1)

T_IGN

T_AVG(2..n)

History Delay⁽¹⁾

Delay to ignore

history updates

prior to LOR.

History Count⁽¹⁾

Timer to average history data to

Initial holdover

frequency determined

by averaged history.

Previous history is persistent

(not cleared or reset after

exiting holdover).

compute initial holdover frequency accuracy.

Time

Free Run

Lock Acq.

Locked

Holdover

Lock Acq.

Locked

LOFL = 1, LOPL = 1

LOFL‰ 0, then LOPL‰ 0

LOFL = 0, LOPL‰ 1

LOFL = 0, LOPL = 1 LOFL = 0, LOPL‰ 0

(1) History count and delay windows are programmable.

图 41. Tuning Word History Windows

If the T_AVGperiod is set very long (minutes or hours) to obtain a more precise historical average frequency, it is

possible for a switchover or holdover event to occur before the first tuning word is stored and available for use.

To overcome this, there is an intermediate history update option (HIST_INTMD). If the history is reset, then the

intermediate average can be updated at intervals of T_AVG/2^K, where K = HIST_INTMD to 0, during the first T_AVG

period only. If HIST_INTMD = 0, there is no intermediate update and the first average is stored after the first

T_AVGperiod. However, if HIST_INTMD = 4, then four intermediate averages are taken at T_AVG/16, T_AVG/8, T_AVG/4,

and T_AVG/2, as well as at T_AVG. After the first T_AVGperiod, all subsequent history updates occur at the T_AVG

period.

When no tuning word history exists, the free-run tuning word value (TUNING_FREE_RUN) determines the initial

holdover output frequency accuracy.

38

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.7.5 Status Outputs

The STATUS0 and STATUS1 pins can be configured to output various status signals and interrupt flag for device

diagnostic and debug purposes. The status signal, output driver type, and output polarity settings are

programmable. The status signals available at these pins are listed in 表 6. When the status signal is asserted,

the status output will be driven high (active high) assuming the output polarity is not inverted (or active low).

表 6. Status Pin Signals Available per Device Block

DEVICE BLOCK MONITORED

STATUS SIGNAL (ACTIVE HIGH)

XO Loss of Signal (LOS)

XO

APLLx Lock Detected (LOL)

APLLx VCO Calibration Active

APLLx N Divider, div-by-2

APLL1 and APLL2

APLLx Digital Lock Detect (DLD)

APLL2 R Divider, div-by-2

EEPROM

EEPROM Active

All Inputs and PLLs

Interrupt (INTR)

PRIREF/SECREF Monitor Divider Output, div-by-2

PRIREF/SECREF Amplitude Monitor Fault

PRIREF/SECREF Frequency Monitor Fault

PRIREF/SECREF Missing or Early Pulse Monitor Fault

PRIREF/SECREF Validation Timer Active

PRIREF/SECREF Phase Validation Monitor Fault

DPLL R Divider, div-by-2

PRIREF and SECREF

DPLL FB Divider, div-by-2

DPLL Phase Lock Detected (LOPL)

DPLL PRIREF/SECREF Selected

DPLL Holdover Active

DPLL

DPLL Reference Switchover Event

DPLL Tuning History Update

DPLL FastLock Active

DPLL Loss of Lock (LOFL)

39

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3.7.6 Interrupt

Any of the two status pins can be configured as a device interrupt output pin. The interrupt logic configuration is

set through registers. When the interrupt logic is enabled, the interrupt output can be triggered from any

combination of interrupt status indicators, including LOS for the XO, LOR for the selected DPLL input, LOL for

each APLL and the DPLL, and holdover and switchover events for the DPLL. When the interrupt polarity is set

high, a rising edge on the live status bit will assert its interrupt flag (sticky bit). Otherwise, when the polarity is set

low, a falling edge on the live status bit will assert its interrupt flag. Any individual interrupt flag can be masked so

it does not trigger the interrupt output. The unmasked interrupt flags are combined by the AND/OR gate to

generate the interrupt output, which can be selected on either status pin.

When a system host detects an interrupt from the LMK05318, the host can read the interrupt flag or "sticky"

registers to identify which bits were asserted to resolve the fault conditions in the system. After the system faults

have been resolved, the host can clear the interrupt output by writing zeros to the sticky bits that were asserted.

INTR

Flag*

INTR

Enable

INTR

Mask

INTR

Polarity

INT_AND_OR

Status Bits

F

LOS_FDET_XO

LOS_XO

2

Status Pins (x2)

F

LOL_PLL[1:2]

LOFL_DPLL

STATx_SEL

LOPL_DPLL

HIST

INTR

AND/OR

Gate

Polarity

Type

0xA

HLDOVR

Status

Select

STATUS0

STATUS1

F

Other

status

signals

REFSWITCH

LOR_MISSCLK

LOR_FREQ

LOR_AMP

Sticky Status Registers

0x013 to 0x014

Live Status Registers

0x00D to 0x00E

*Write 0 to clear INTR flag bits

图 42. Status and Interrupt

40

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.8 PLL Relationships

图 43 shows the PLL architecture implemented in the LMK05318. The PLLs can be configured in the different

PLL modes described in PLL Architecture Overview.

Post

Dividers

APLL2

VCO

f_PD2

f_PD

VCO2

0

1

f_VCO2/P1

f_VCO2

PFD

LF

÷P1

f_VCO

R Dividers

÷RP

/2 to /7

÷P2

÷N

f_VCO2/P2

/3 to /6

24-bit Frac-N SDM

To

÷RS

Output

Muxes

/1 to /32

×1, ×2

÷R

XO

DPLL

APLL1

PRIREF

0

1

f_PD1

f_TDC

VCO1

÷R

5-bit

f_VCO1

SECREF

TDC

DLF

LF

PFD

16-bit

(x2)

÷FB

40-bit Frac-N SDM

÷N

f_VCO1

÷PR

÷2

40-bit Frac-N SDM

To

Output

Muxes

38-bit

DCO option

DCO

FINC/FDEC

DPLL feedback clock

FDEV

图 43. PLL Architecture

9.3.8.1 PLL Frequency Relationships

The following equations provide the PLL frequency relationships required to achieve closed-loop operation

according to the selected PLL mode. The TICS Pro programming software can be used to generate valid divider

settings based on the desired frequency plan configuration and PLL mode.

•

To operate APLL1 in Free-run mode (locked to the XO input), the conditions in 公式 1 and 公式 2 must be

met.

•

To operate APLL1 in DPLL mode, the conditions in 公式 1, 公式 2, 公式 3, and 公式 4 must be met.

To operate APLL2 in Cascaded mode, the conditions in 公式 1, 公式 2, 公式 5, and 公式 7 must be met.

To operate APLL2 in Non-cascaded mode, the conditions in 公式 6 and 公式 7 must be met.

Note that any divider in the following equations refer to the actual divide value (or range) and not its

programmable register value.

公式 1 and 公式 2 relate to APLL1:

f_PD1= f_XO× D_XO/ R_XO

where

•

f_PD1= APLL1 phase detector frequency

f_XO: XO input frequency

D_XO: XO input doubler (1 = disabled, 2 = enabled)

R_XO: APLL1 XO Input R divider value (1 to 32)

(1)

41

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

f_VCO1= f_PD1× (INT_APLL1+ NUM_APLL1/ DEN_APLL1

)

where

•

f_VCO1: VCO1 frequency

INT_APLL1: APLL1 N divider integer value (12 bits, 1 to 2¹²– 1)

NUM_APLL1: APLL1 N divider numerator value (40 bits, 0 to 2⁴⁰– 1)

DEN_APLL1: APLL1 N divider denominator value (fixed, 2⁴⁰)

–

0.125 < NUM_APLL1/ DEN_APLL1< 0.875 (In DPLL Mode)

(2)

公式 3 and 公式 4 relate to the DPLL:

f_TDC= f_PRIREF/ R_PRIREF= f_SECREF/ R_SECREF

where

•

f_TDC: DPLL TDC input frequency (see 公式 3)

f_PRIREFor f_SECREF: PRIREF or SECREF input frequency

R_PRIREFor R_SECREF: PRIREF or SECREF R divider value (16 bits, 1 to 2¹⁶– 1)

(3)

(4)

f_VCO1= f_TDC× 2 × PR_DPLL× (INT_DPLL+ NUM_DPLL/ DEN_DPLL

)

where

•

PR_DPLL: DPLL prescaler divider value (2 to 17)

INT_DPLL: DPLL FB divider integer value (30 bits, 1 to 2³⁰– 1)

NUM_DPLL: DPLL FB divider numerator value (40 bits, 0 to 2⁴⁰– 1)

DEN_DPLL: DPLL FB divider denominator value (40 bits, 1 to 2⁴⁰)

公式 5, 公式 6, and 公式 7 relate to APLL2:

Cascaded APLL2: f_PD2= f_VCO1/ (R_{APLL2_PRE}× R_{APLL2_SEC}

)

where

•

f_PD2: APLL2 phase detector frequency

R_{APLL2_PRE}: Cascaded APLL2 Pre R divider value (3 to 6)

R_{APLL2_SEC}: Cascaded APLL2 Secondary R divider value (1 to 32)

(5)

(6)

Non-Cascaded APLL2: f_PD2= f_XO× D_XO

f_VCO2= f_PD2× (INT_APLL2+ NUM_APLL2/ DEN_APLL2

)

where

•

f_VCO2: VCO2 frequency

INT_APLL2: APLL2 N divider integer value (9 bits, 1 to 2⁹– 1)

NUM_APLL2: APLL2 N divider numerator value (24 bits, 0 to 2²⁴– 1)

DEN_APLL2: APLL2 N divider denominator value (fixed, 2²⁴)

(7)

公式 8, 公式 9, 公式 10, and 公式 11 relate to the output frequency, which depends on the selected APLL clock

source and output divider value:

APLL1 selected: f_CHxMUX= f_VCO1

(8)

(9)

APLL2 selected: f_CHxMUX= f_VCO2/ Pn_APLL2

OUT[0:6]: f_OUTx= f_CHxMUX/ OD_OUTx

OUT7: f_OUT7= f_CH7MUX/ (OD_OUT7× OD2)

(10)

where

•

f_CHxMUX: Output mux source frequency (APLL1 or APLL2 post-divider clock)

Pn_APLL2: APLL2 primary "P1" or secondary "P2" post-divide value (2 to 7)

f_OUTx: Output clock frequency (x = 0 to 7)

OD_OUTx: OUTx output divider value (8 bits, 1 to 2⁸)

OD2: OUT7 secondary output divider value (24 bits, 1 to 2²⁴)

–

If OD2 > 1, then OD_OUT7≥ 6

(11)

42

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.8.2 Analog PLLs (APLL1, APLL2)

APLL1 has a 40-bit fractional-N divider and APLL2 has a 24-bit fractional-N divider to support high-resolution

frequency synthesis and very low phase noise and jitter. APLL1 has the ability to tune its VCO1 frequency

through sigma-delta modulator (SDM) control in DPLL mode. APLL2 has the ability to lock its VCO2 frequency to

the VCO1 frequency.

In free-run mode, APLL1 uses the XO input as an initial reference clock to VCO1. APLL1's PFD compares the

fractional-N divided clock with its reference clock and generates a control signal. The control signal is filtered by

the APLL1 loop filter to generate VCO1's control voltage to set its output frequency. The SDM modulates the N

divider ratio to get the desired fractional ratio between the PFD input and the VCO output. APLL2 operates

similar to APLL1, but the user can select APLL2's reference from either the VCO1 clock or XO clock.

In DPLL mode, the APLL1 fractional SDM is controlled by the DPLL loop to pull the VCO1 frequency into lock

with the DPLL reference input. If APLL2 derives its reference from VCO1, then VCO2 will be effectively locked to

the DPLL reference input, assuming there is no synthesis error introduced by the fractional N divide ratio of

APLL2.

9.3.8.3 APLL Reference Paths

9.3.8.3.1 APLL XO Doubler

The APLL XO doubler can be enabled to double the PFD frequency up to 50 MHz for APLL1 and up to 150 MHz

for APLL2 in Non-Cascaded mode. Enabling the XO doubler adds minimal noise and can be useful to increase

the PFD frequency to optimize phase noise, jitter, and fractional spurs. The flat portion of the APLL phase noise

can improve when the PFD frequency is increased.

9.3.8.3.2 APLL1 XO Reference (R) Divider

APLL1 has a 5-b XO R divider that can be used to meet the maximum APLL1 PFD frequency specification. It

can also be used to ensure the APLL1 fractional-N divide ratio (NUM/DEN) is between 0.125 to 0.875, which is

recommended to support the DPLL frequency tuning range. Otherwise, the XO R divider can be bypassed

(divide by 1).

9.3.8.3.3 APLL2 Reference (R) Dividers

APLL2 has a cascaded primary R divider (÷3 to ÷6) and secondary R divider (÷1 to ÷32) to divide-down the

VCO1 clock to meet the maximum APLL2 PFD frequency specification in Cascaded APLL2 mode. The dividers

can also be used to operate APLL2 in integer mode or avoid near-integer spurs in fractional mode.

9.3.8.4 APLL Phase Frequency Detector (PFD) and Charge Pump

The APLL1 PFD frequency can operate up to 50 MHz and can be computed by 公式 1. APLL1 has

programmable charge pump settings from 0 to 1500 µA in 100-µA steps. Best performance from APLL1 is

achieved with a charge pump currents of 800 µA or higher.

The APLL2 PFD frequency can operate up to 150 MHz and can be computed by 公式 5 in Cascaded mode or 公

式 6 in Non-cascaded mode. APLL2 has programmable charge pump settings of 1.6, 3.2, 4.8, or 6.4 mA.

9.3.8.5 APLL Feedback Divider Paths

The VCO output of each APLL is fed back to its PFD block through the fractional feedback (N) divider. The

VCO1 output is also fed back to the DPLL feedback path in DPLL mode.

9.3.8.5.1 APLL1 N Divider With SDM

The APLL1 fractional N divider includes a 12-b integer portion (INT), a 40-b numerator portion (NUM), a fixed 40-

b denominator portion (DEN), and a sigma-delta modulator. The INT and NUM are programmable, while the

denominator is fixed to 2⁴⁰for very high frequency resolution on the VCO1 clock. The total APLL1 N divider value

is: N = INT + NUM / 2⁴⁰.

In APLL free-run mode, the PFD frequency and total N divider for APLL1 determine the VCO1 frequency, which

can be computed by 公式 2.

43

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

9.3.8.5.2 APLL2 N Divider With SDM

The APLL2 fractional N divider includes a 9-b integer portion (INT), a 24-b numerator portion (NUM), a fixed 24-b

denominator portion (DEN), and a sigma-delta modulator. The INT and NUM are programmable, while the

denominator is fixed to 2²⁴for high frequency resolution on the VCO2 clock. The total APLL2 N divider value is:

N = INT + NUM / 2²⁴.

The PFD frequency and total N divider for APLL2 determine the VCO2 frequency, which can be computed by 公

式 7.

9.3.8.6 APLL Loop Filters (LF1, LF2)

APLL1 supports a programmable loop bandwidth from 100 Hz to 10 kHz (typical range), and APLL2 supports a

programmable loop bandwidth from 100 kHz to 1 MHz (typical range). The loop filter components can be

programmed to optimize the APLL bandwidth depending on the reference input frequency and phase noise. The

LF1 and LF2 pins each require an external "C2" capacitor to ground. See the suggested values for the LF1 and

LF2 capacitors in the Pin Configuration and Functions section.

图 44 shows the APLL loop filter structure between the PFD/charge pump output and VCO control input.

VCO

Programmable

Loop Filter

R3

R4

PFD /

Charge Pump

C1

R2

C3

C4

LF1, LF2

C2

图 44. Loop Filter Structure of Each APLL

9.3.8.7 APLL Voltage Controlled Oscillators (VCO1, VCO2)

Each APLL contains a fully-integrated VCO, which takes the voltage from its loop filter and converts this into a

frequency. VCO1 uses proprietary BAW resonator technology with a very high quality factor to deliver the lowest

phase jitter and has a tuning range of 2.5 GHz ± 50 ppm. VCO2 uses a high-performance LC VCO with a wider

tuning range of 5.5 to 6.25 GHz to cover a additional unrelated clock frequencies, if needed.

9.3.8.7.1 VCO Calibration

Each APLL VCO must be calibrated to ensure that the PLL can achieve lock and deliver optimal phase noise

performance. VCO calibration establishes an optimal operating point within the VCO tuning range. VCO

calibration is executed automatically during initial PLL start-up after device power-on, hard-reset, or soft-reset

once the XO input is detected by its input monitor. To ensure successful calibration and APLL lock, it is critical

for the XO clock to be stable in amplitude and frequency before the start of calibration; otherwise, the calibration

can fail and prevent PLL lock and output clock start-up. Before VCO calibration and APLL lock, the output drivers

are typically held in the mute state (configurable per output) to prevent spurious output clocks.

A VCO calibration can be triggered manually for a single APLL by toggling a PLL power-down cycle (PLLx_PDN

bit = 1 → 0) through host programming. This may be needed after the APLL N divider value (VCO frequency) is

changed dynamically through programming.

44

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.8.8 APLL VCO Clock Distribution Paths (P1, P2)

APLL1 has no VCO post-dividers. The primary VCO1 clock (P1) and a secondary VCO1 inverted clock (P2) are

distributed to all output channel muxes. The inverted clock is optional, but it can help to reduce spurious in some

cases.

APLL2 has two VCO2 post-dividers to provide more flexible clock frequency planning. The primary VCO2 post-

divider clock (P1) and secondary post-divider clock (P2) are distributed to all output channel muxes. Both VCO2

post-dividers support independently programmable dividers (÷2 to ÷7). Note that output SYNC is not supported

between output channels selecting a VCO2 post-divider of 2.

A PLL2 or device soft-reset is recommended after changing the APLL2 post-divider value to initialize it for

deterministic divider operation.

9.3.8.9 DPLL Reference (R) Divider Paths

Each reference input clock (PRIREF and SECREF) has its own 16-b reference divider to the DPLL TDC block.

The R divider output of the selected reference sets the TDC input frequency. To support hitless switching

between inputs with different frequencies, the R dividers can be used to divide the clocks to a single common

frequency to the DPLL TDC input.

9.3.8.10 DPLL Time-to-Digital Converter (TDC)

The TDC input compares the phase of the R divider clock of the selected reference input and the DPLL feedback

divider clock from VCO1. The TDC output generates a digital correction word corresponding to the phase error

which is processed by the DPLL loop filter.

The DPLL TDC input frequency (f_TDC) can operate up to 26 MHz and can be computed by 公式 3.

9.3.8.11 DPLL Loop Filter (DLF)

The DPLL supports a programmable loop bandwidth from 10 mHz to 4 kHz and can achieve jitter peaking below

0.1 dB (typical). The low-pass jitter transfer characteristic of the DPLL attenuates its reference input noise with up

to 60-dB/decade roll-off above the loop bandwidth.

The DPLL loop filter output controls the fractional SDM of APLL1 to steer the VCO1 frequency into lock with the

selected DPLL reference input.

9.3.8.12 DPLL Feedback (FB) Divider Path

The DPLL feedback path has a fixed prescaler (÷2), programmable prescaler (÷2 to ÷17), and a fractional

feedback (FB) divider. The programmable DPLL FB divider includes a 30-b integer portion (INT), 40-b numerator

portion (NUM), and 40-b denominator portion (DEN). The total DPLL FB divider value is: FB_DPLL= INT + NUM /

DEN.

In DPLL mode, the TDC frequency and total DPLL feedback divider and prescalers determine the VCO1

frequency, which can be computed by 公式 4.

9.3.9 Output Clock Distribution

The output clock distribution blocks shown in 图 45 include six output muxes, six output dividers, and eight

programmable output drivers. The output dividers support output synchronization (SYNC) to allow phase

synchronization between two or more output channels. Also, the OUT7 channel has an optional zero-delay mode

(ZDM) synchronization feature to support deterministic input-to-output phase alignment (typically for 1-PPS

clocks) with programmable offset.

45

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Clock Bus

OUT0

OUT1

0

1

2

3

÷OD

8-b

OUT[0:3] bank

preferred for

PLL1 clocks

PLL1

f_VCO1

0

1

OUT2

OUT3

0

1

2

3

÷OD

8-b

VCO1

Output Channel Configuration

Power-

down

Output Type

Mux

0

1

2

3

÷OD

8-b

OUT4

2

PLL2

SYNC EN

(1)

Auto Mute,

Mute Level

f_VCO2

2

3

0

1

2

3

÷P1

/2 to /7

÷P2

÷OD

8-b

OUT5

OUT6

OUT7

SYNC

VCO2

OUT[4:7] bank

preferred for

PLL2 clocks

0

1

2

3

÷OD

8-b

0

1

2

3

÷OD

8-b

÷OD_2

24-b

SYNC_SW

SYNC

GPIO0/SYNCN

(active-low pin)

图 45. Output Clock Distribution

9.3.10 Output Channel Muxes

Each of the six output channels has as output mux. Each output mux for the OUT0 to OUT7 channels can

individually select between the PLL1 VCO clocks (normal or inverted) and PLL2 VCO post-divider clocks.

9.3.11 Output Dividers (OD)

Each of the six output channels has an output divider after the output mux. The OUT[0:1] channel has a single

output divider that is similar to the OUT[2:3] channel output divider. Each OUT[4:7] channel has an individual

output divider. The output divider is used to generate the final clock output frequency from the source selected by

the output mux.

Each OUT[0:6] channel has an 8-bit divider (OD) that can support output frequencies from 10 to 800 MHz (or up

to the maximum frequency supported by the configured output driver type). It is possible to configure the PLL

post-divider and output divider to achieve higher clock frequencies, but the output swing of the driver may fall out

of specification.

46

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

The OUT7 channel has cascaded 8-bit (OD) and 24-bit (OD2) output dividers to support output frequencies from

1 Hz (1 PPS) to 800 MHz. The total OUT7 divide value is the product of the cascaded divider values (OD ×

OD2).

Each output divider is powered from the same VDDO_x supply used for the clock output drivers. The output

divider can be powered down if not used to save power. For either OUT[0:1] or OUT[2:3] channel, the output

divider is automatically powered down when both output drivers are disabled. For any OUT[4:7] channel, the

output divider is automatically powered down when its output driver is disabled.

9.3.12 Clock Outputs (OUTx_P/N)

Each clock output can be individually configured as a differential driver (AC-LVDS/CML/LVPECL), HCSL driver,

or 1.8-V LVCMOS drivers (two per pair). Otherwise, it can be disabled if not used to save power.

Each output channel has its own internal LDO regulator to provide excellent PSNR and minimize jitter and spurs

induced by supply noise. The OUT[0:1] channel (mux, divider, and drivers) are powered through a single output

supply pin (VDDO_01), and similarly for the OUT[2:3] channel (VDDO_23). Each OUT[4:7] channel have their

own output supply pin (VDDO[4:7]). Each output supply can be separately powered by 1.8 V, 2.5 V, or 3.3 V for

a differential or HCSL output, or 1.8 V for an LVCMOS output.

For differential and HCSL driver modes, the output clock specifications (such as output swing, phase noise, and

jitter) are not sensitive to the VDDO_x voltage because of the channel's internal LDO regulator. When an output

channel is left unpowered, the channel's output(s) will not generate any clocks.

表 7. Output Driver Modes

OUTx_FMT

00h

OUTPUT FORMAT⁽¹⁾

Disabled (powered-down)

AC-LVDS

10h

14h

AC-CML

18h

AC-LVPECL

2Ch

2Dh

30h

HCSL (External 50-Ω to GND)

HCSL (Internal 50-Ω to GND)

LVCMOS (HiZ / HiZ)

LVCMOS (HiZ / –)

LVCMOS (HiZ / +)

LVCMOS (Low / Low)

LVCMOS (– / HiZ)

LVCMOS (– / –)

32h

33h

35h

38h

3Ah

3Bh

3Ch

3Eh

3Fh

LVCMOS (– / +)

LVCMOS (+ / HiZ)

LVCMOS (+ / –)

LVCMOS (+ / +)

(1) LVCMOS modes are only available on OUT[4:7].

9.3.12.1 AC-Differential Output (AC-DIFF)

The programmable differential output driver uses a switched-current mode type shown in 图 46. A tail current of

4, 6, or 8 mA (nominal) can be programmed to achieve V_ODswing compatible with AC-coupled LVDS, CML, or

LVPECL receivers, respectively, across a 100-Ω differential termination. The differential output driver is ground-

referenced (similar to an HCSL driver), meaning the differential output has a low common-mode voltage (V_OS).

The differential driver has internal biasing, so external pullup or pulldown resistors should not be applied. The

differential output should be interfaced through external AC-coupling to a differential receiver with proper input

termination and biasing.

47

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

VDDO_x

LDO

I₁= 4 mA

From

output

channel

Output tail current (I₁+ I₂) can be programmed

P

N

to 4, 6, or 8 mA for LVDS-, CML-, and LVPECL-

compatible swing across AC-coupled

50-W single-ended or 100-W differential load.

clk_p

N

clk_n

OUTx_P

I₂= 0, 2, or 4 mA

P

N

P

N

OUTx_N

图 46. AC-LVDS/CML/LVPECL Output Driver Structure

9.3.12.2 HCSL Output

The HCSL output is an open-drain differential driver that can be DC-coupled to an HCSL receiver. The HCSL

output has programmable internal 50-Ω termination to ground which can be enabled if the receiver side does not

provide termination. If the internal termination is disabled, external 50-Ω to ground (on P and N) is required at

either the driver side (source terminated) or the receiver side (load terminated).

9.3.12.3 1.8-V LVCMOS Output

The LVCMOS driver has two outputs per pair. Each output on P and N can be configured for normal polarity,

inverted polarity, or disabled as HiZ or static low level. The LVCMOS output high level (V_OH) is determined by the

VDDO_x voltage of 1.8 V for rail-to-rail LVCMOS output voltage swing. If a VDDO_x voltage of 2.5 V or 3.3 V is

applied to the LVCMOS driver, the output V_OHlevel not will not swing to the VDDO_x rail due to the channel's

internal LDO regulator.

Because an LVCMOS output clock is an unbalanced signal with large voltage swing, it can be a strong aggressor

and couple noise onto other jitter-sensitive differential output clocks. If an LVCMOS clock is required from an

output pair, configure the pair with both outputs enabled but with opposite polarity (+/– or –/+) and leave the

unused output floating with no trace connected.

9.3.12.4 Output Auto-Mute During LOL

Each output driver can automatically mute or squelch its clock when the selected output mux clock source is

invalid, as configured by its CHx_MUTE bit. The source can be invalid based on the LOL status of each PLL by

configuring the APLL and DPLL mute control bits (MUTE_APLLx_LOCK, MUTE_DPLL_LOCK,

MUTE_DPLL_PHLOCK). The mute level can be configured per output channel by its CHx_MUTE_LVL bits,

where the mute level depends on the configured output driver type (Differential/HCSL or LVCMOS). The mute

level for a differential or HCSL driver can be set to output common mode, differential high, or differential low

levels. The mute level for an LVCMOS driver pair can be set to output low level for each of its outputs (P and N)

independently. When auto-mute is disabled or bypassed (CHx_MUTE = 0 and CHx_MUTE_LVL = 0), the output

clock can have incorrect frequency or be unstable before and during the VCO calibration. For this reason, the

mute bypass mode should only be used for diagnostic or debug purposes.

48

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.3.13 Glitchless Output Clock Start-Up

When APLL auto-mute is enabled, the outputs will start up in synchronous fashion without clock glitches once

APLL lock is achieved after any the following events: device power-on, exiting hard-reset, exiting soft-reset, or

deasserting output SYNC (when SYNC_MUTE = 1).

9.3.14 Clock Output Interfacing and Termination

图 47 to 图 51 show the recommended output interfacing and termination circuits. Unused clock outputs can be

left floating and powered down by programming.

LVCMOS

1.8 V LVCMOS

LMK05318

Receiver

图 47. 1.8-V LVCMOS Output to 1.8-V LVCMOS Receiver

LVDS

Receiver

LMK05318

AC-LVDS

100 ꢀ

图 48. AC-LVDS Output to LVDS Receiver With Internal Termination/Biasing

50 ꢀ

CML

Receiver

LMK05318

AC-CML

50 ꢀ

图 49. AC-CML Output to CML Receiver With Internal Termination/Biasing

AC-LVPECL

LVPECL Receiver

LMK05318

50 ꢀ

VDD_IN œ 1.3 V

图 50. AC-LVPECL Output to LVPECL Receiver With External Termination/Biasing

49

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

33 ꢀ (optional)

LMK05318

HCSL

HCSL Receiver

33 ꢀ (optional)

50 ꢀ

If HCSL Internal Termination (50-Ω to GND) is enabled, short 33-Ω and remove 50-Ω external resistors.

图 51. HCSL Output to HCSL Receiver With External Source Termination

9.3.15 Output Synchronization (SYNC)

Output SYNC can be used to phase-align two or more output clocks with a common rising edge by allowing the

output dividers to exit reset on the same PLL output clock cycle. Any output dividers selecting the same PLL

output can be synchronized together as a SYNC group by triggering a SYNC event through the hardware pin or

software bit.

The following requirements must be met to establish a SYNC group for two or more output channels:

•

Output dividers have their respective sync enable bit set (CHx_SYNC_EN = 1).

Output dividers have their output mux selecting the same PLL output.

The PLL (post-divider) output has its sync enable bit set (for example, PLL1_P1_SYNC_EN = 1).

A SYNC event can be asserted by the hardware GPIO0/SYNCN pin (active low) or the SYNC_SW register bit

(active high). When SYNC is asserted, the SYNC-enabled dividers held are reset and clock outputs are muted.

When SYNC is deasserted, the outputs will start with their initial clock phases synchronized or aligned. SYNC

can also be used to mute any SYNC-enabled outputs to prevent output clocks from being distributed to

downstream devices until they are configured and ready to accept the incoming clock.

Output channels with their sync disabled (CHx_SYNC_EN bit = 0) will not be affected by a SYNC event and will

continue normal output operation as configured. Also, VCO and PLL post-divider clocks do not stop running

during the SYNC so they can continue to source output channels that do not require synchronization. Output

dividers with divide-by-1 (divider bypass mode) are not gated during the SYNC event.

表 8. Output Synchronization

GPIO0/SYNCN PIN

SYNC_SW BIT

OUTPUT DIVIDER AND DRIVER STATE

Output driver(s) muted and output divider(s) reset

0

0→1

1

1→0

0

Outputs in a SYNC group are unmuted with their initial clock phases aligned

Normal output driver/divider operation as configured

注

Output SYNC is not supported (output-to-output skew specifications is not ensured)

between output channels selecting a PLL2 output (P1 or P2) with VCO2 post-divider of 2.

9.3.16 Zero-Delay Mode (ZDM) Synchronization for 1-PPS Input and Output

Zero-delay mode synchronization can be enabled to achieve zero phase delay between the selected DPLL

reference input clock and the OUT7 clock as shown in 图 52. This is primarily used to achieve deterministic

phase relationship between a 1-PPS input and 1-PPS output. This feature can be configured through registers by

enabling ZDM (DPLL_ZDM_SYNC_EN bit = 1) and enabling OUT7 divider synchronization (CH7_SYNC_EN bit

= 1). The OUT7 clock must be derived from the DPLL and APLL1 VCO domain (f_VCO1).

50

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

When the DPLL is not locked and the DPLL reference input is invalid, the OUT7 clock is held in mute state (no

clock). Once the reference input is validated and selected, the OUT7 channel divider is reset or SYNCed using

the DPLL reference input clock edge to achieve a deterministic phase relationship between the reference input

and OUT7 clock. OUT7 is not affected by normal output SYNC events, and OUT[0:6] are not be affected by a

ZDM SYNC event. The input-to-output phase offset can be adjusted through the DPLL phase offset register

control (DPLL_REF_SYNC_PH_OFFSET bits). If the DPLL phase offset is programmed on-the-fly with 1-PPS

input, it can take a long time to adjust due to the narrow DPLL bandwidth (10 mHz typical).

Hitless switching between 1-PPS inputs is not supported when ZDM is enabled. If a switchover event between 1-

PPS inputs occurs when ZDM is enabled, a soft-reset should be issued for the DPLL to relock and realign the 1-

PPS output to the selected input.

DPLL + APLL1

OUT7 Channel

÷OD

SYNC

÷R

f_TDC

f_VCO1

REF

OUT7

Phase Offset

DPLL_REF_SYNC

_PH_OFFSET

DPLL_ZDM_SYNC_EN

图 52. DPLL ZDM Synchronization Between Reference Input and OUT7

9.4 Device Functional Modes

9.4.1 Device Start-Up Modes

The LMK05318 can start up in one of three device modes depending on the 3-level input level sampled on the

HW_SW_CTRL pin during power-on reset (POR):

•

HW_SW_CTRL = 0: EEPROM + I²C Mode (Soft pin mode)

HW_SW_CTRL = Float (V_IM): EEPROM + SPI Mode (Soft pin mode)

HW_SW_CTRL = 1: ROM + I²C Mode (Hard pin mode)

The device start-up mode determines:

•

The memory bank (EEPROM or ROM) used to initialize the register settings that sets the frequency

configuration.

•

The serial interface (I²C or SPI) used for register access.

The logic pin functionality for device control and status.

After start-up, the I²C or SPI interface is enabled for register access to monitor the device status and control (or

reconfigure) the device if needed. The register map configurations are the same for I²C and SPI.

Table 1 summarizes the device start-up mode and corresponding logic pin functionality.

图 53 shows the device power-on reset configuration sequence.

51

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Device Functional Modes (接下页)

Power-On Reset

(POR)

Device POR

Configuration Sequence

PDN = 0

Hard Reset?

Outputs muted

PDN = 1

HW_SW_CTRL = 1

HW_SW_CTRL = 0

GPIO[2:0] = 000b to 111b

(ROM Page Select)

GPIO1 = 0, Float, 1 (I²C Addr. Select)

Start-up Mode?

(sample pin states)

HW_SW_CTRL = Float

STATUS[1:0] = Float

EEPROM + I²C

EEPROM + SPI

ROM + I²C

(Soft Pin) Mode

(Hard Pin) Mode

Registers initialize from EEPROM/ROM and I²C/SPI,

Control & Status pins activate (after hard reset only).

All blocks reset to initial states.

RESET_SW = 0

Device Block

Configuration

Soft Reset?

RESET_SW = 1

Register and EEPROM programming available.

Normal Operation

See PLL Initialization

Flowchart

图 53. Device POR Configuration Sequence

9.4.1.1 EEPROM Mode

In EEPROM mode, the frequency configuration of the device is loaded to the registers from the non-volatile

EEPROM. The factory default start-up configuration for EEPROM mode is summarized in EEPROM Start-Up

Mode Default Configuration. If a different custom start-up configuration is needed, a different EEPROM image

can be programmed in-system through the serial interface. The EEPROM supports up to 100 programming

cycles to facilitate clock reconfiguration for system-level prototyping, debug, and optimization.

The EEPROM image can store a single frequency configuration (one register page). Upon request, a factory pre-

programmed device with a custom EEPROM image could be assigned by TI with a unique orderable part number

(OPN).

TI suggests to use the EEPROM mode when either of the following is true:

•

A single custom start-up frequency configuration is required from a single OPN.

A host device is available to program the registers (and EEPROM if needed) with a new configuration after

power-up through I²C or SPI. SPI is not supported in ROM mode.

9.4.1.2 ROM Mode

In ROM mode, the frequency configuration of the device is loaded to the registers from one of eight register

pages in ROM selected by the GPIO[2:0] control pins. All register pages in the ROM image can be factory-set in

hardware (mask ROM) and are not software programmable. Only the I²C interface is available for register access

after start-up in ROM mode.

The factory ROM image have default register pages intended for TI internal use, but ROM pages may be

allocated for future custom frequency configurations upon request.

52

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Device Functional Modes (接下页)

A benefit of ROM mode over EEPROM mode is that a custom ROM image can support up to eight different pin-

selectable frequency configurations from a single OPN. Upon request, a factory preset device with a custom

ROM image could be assigned by TI with a unique OPN.

9.4.2 PLL Operating Modes

The following sections describe the PLL modes of operation shown in 图 54.

See Device POR

and PLL Initialization

Flowchart

No valid input

available

Free-run Mode⁽¹⁾

Initial frequency accuracy

determined by free-run

tuning word register.

(1) Free-run/Holdover Mode frequency

stability determined by TCXO/OCXO/

XO.

No

Valid Input

Available for

Selection? ⁽²⁾

(2) See Input Selection Flowchart.

Yes

Lock Acquisition

(Fastlock, Hitless Switch)

Phase-locked to

selected input

Yes

Valid Input

Available for

Selection? ⁽²⁾

No

DPLL Locked

DCO Mode available.

No

Loss of Ref (LOR) on

Selected Input? ⁽²⁾

Holdover Mode⁽¹⁾

Initial holdover frequency

accuracy determined by

averaged history data.

Yes

No

Yes

Is Tuning Word

History Valid?

(1) Assumes DPLL_HLDOVR_MODE bit is 0 to enter free-run mode if history is not valid.

图 54. PLL Operating Mode

53

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Device Functional Modes (接下页)

9.4.2.1 Free-Run Mode

After device POR configuration and initialization, APLL1 will automatically lock to the XO clock once it is detected

by its input monitor. Then, APLL2 will acquire lock to either VCO1 or XO frequency as selected. The output clock

frequency accuracy and stability in free-run mode are equal to that of the XO input. The reference inputs remain

invalid (unqualified) during free-run mode.

9.4.2.2 Lock Acquisition

The DPLL constantly monitors its reference inputs for a valid input clock. When at least one valid input clock is

detected, the PLL1 channel will exit free-run mode or holdover mode and initiate lock acquisition through the

DPLL. The device supports the Fastlock feature where the DPLL temporarily engages a wider loop bandwidth to

reduce the lock time. Once the lock acquisition is done, the loop bandwidth is set to its normal configured loop

bandwidth setting (BW_DPLL).

9.4.2.3 Locked Mode

Once locked, the APLL1 output clocks are frequency and phase locked to the selected DPLL input clock. While

the DPLL is locked, the APLL1 output clocks will not be affected by frequency drift on the XO input. The DPLL

has a programmable frequency lock detector and phase lock detectors to indicate loss-of-frequency lock (LOFL)

and loss-of-phase lock (LOPL) status flags, which can be observed through the status pins or status bits. Once

frequency lock is detected (LOFL → 0), the tuning word history monitor (if enabled) will begin to accumulate

historical averaging data used to determine the initial output frequency accuracy upon entry into holdover mode.

9.4.2.4 Holdover Mode

When a loss-of-reference (LOR) condition is detected and no valid input is available, the PLL1 channel enters

holdover mode. If the tuning word history is valid, the initial output frequency accuracy upon entry into holdover

will be pulled to the computed average frequency accuracy just prior to the loss of reference. If history is not valid

(no history exists) and the DPLL_HLDOVR_MODE bit is 0, the holdover frequency accuracy will be determined

by the free-run tuning word register (user programmable). Otherwise, if history is not valid and

DPLL_HLDOVR_MODE is 1, the DPLL will hold its last digital loop filter output control value (which is not tuning

word history).

If history is valid, the initial holdover frequency accuracy depends on the DPLL loop bandwidth and the elapsed

time used for historical averaging. See Tuning Word History. In general, the longer the historical average time,

the more accurate the initial holdover frequency assuming the 0-ppm reference clock (XO input) is drift-free. The

stability of the XO reference clock determines the long-term stability and accuracy of the holdover output

frequency. Upon entry into holdover, the LOPL flag will be asserted (LOPL → 1). The LOFL flag will not be

asserted, however, as long as the holdover frequency accuracy does not drift beyond of the programmed loss-of-

frequency-lock threshold. When a valid input becomes available for selection, the PLL1 channel will exit holdover

mode and automatically phase lock with the new input clock without any output glitches.

9.4.3 PLL Start-Up Sequence

图 55 shows the general sequence for PLL start-up after device configuration. This sequence is also applicable

after a device soft-reset or individual PLL soft-reset. To ensure proper VCO calibration, it is critical for the

external XO clock to be stable in amplitude and frequency prior to the start of VCO calibration. Otherwise, the

VCO calibration can fail and prevent start-up of the PLL and its output clocks.

54

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Device Functional Modes (接下页)

Device Configured

See Device POR

Configuration Flowchart

XO Detected

VCO Calibration

PLL Initialization Sequence

APLL2 locks before APLL1 in Cascaded mode.

Outputs lock to XO frequency.

APLL(s) Locked

Outputs un-mute if DPLL auto-mute disabled.

Outputs auto-SYNC if enabled.

(Free-run from XO)

Input Monitoring (fastest to slowest detector):

1. Missing and/or Early clock detector

2. Amplitude or Slew rate detector

3. Frequency (ppm) detector

4. 1-PPS phase valid detector (skip #1 and 3 for 1-PPS)

5. After enabled detectors are valid, validation timer starts

and must finish before input is qualified.

Ref. Input

Validation

Valid Input Selected

DPLL

Lock Acquisition

Fastlock DPLL bandwidth is temporarily asserted

during lock acquisition.

Outputs lock to selected input clock frequency.

Outputs are un-muted if DPLL auto-mute enabled.

Configured DPLL bandwidth is asserted.

DCO Mode control

available (ZDM

must be disabled)

DPLL

Locked

DPLL frequency- and phase-lock detectors are monitored.

Output will have deterministic input-to-output phase

relationship if Zero-Delay Mode (ZDM) SYNC is enabled.

See DPLL Modes

and

Input Selection

Flowcharts

图 55. PLL Start-Up Sequence

9.4.4 Digitally-Controlled Oscillator (DCO) Mode

To support the IEEE 1588 slave clock and other clock steering applications, the DPLL supports DCO mode to

allow precise output clock frequency adjustment of less than 0.001 ppb/step. DCO mode can be enabled

(DPLL_FDEV_EN = 1) when the DPLL is locked.

The DCO frequency step size can be programmed through a 38-bit frequency deviation word register

(DPLL_FDEV bits). The DPLL_FDEV value is an offset added to or subtracted from the current numerator value

of the DPLL fractional feedback divider and determines the DCO frequency offset at the VCO output.

The DCO frequency increment (FINC) or frequency decrement (FDEC) updates can be controlled through

software control or pin control in I²C mode. DCO updates through software control are always available through

I²C or SPI by writing to the DPLL_FDEV_REG_UPDATE register bit. Writing a 0 will increment the DCO

frequency by the programmed step size, and writing a 1 will decrement it by the step size. SPI can achieve faster

DCO update rates than to I²C because the SPI has faster register transfer.

When pin control mode is enabled (GPIO_FDEV_EN = 1) in I²C mode, the GPIO2/SDO/FINC pin will function as

the FINC input and the STATUS1/FDEC pin will function as the FDEC input (STATUS1 output will be disabled).

A positive pulse on the FINC pin or FDEC pin will apply a corresponding DCO update to the DPLL. The minimum

positive pulse width applied to the FINC or FDEC pins should be greater than 100 ns to be captured by the

internal sampling clock. The DCO update rate should be limited to less than 1 MHz when using pin control.

55

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Device Functional Modes (接下页)

When DCO mode is disabled (DPLL_FDEV_EN = 0), the DCO frequency offset will be cleared and the VCO

output frequency will be determined by the original numerator value of the DPLL fractional feedback divider.

APLL1

APLL2

DPLL

f_VCO2

f_VCO1

f_TDC

FINC/FDEC Pin Control

GPIO_FDEV_EN

DPLL_FDEV_EN

GPIO2/FINC

FINC

FDEC

DCO

Step

STATUS1/FDEC

38-bit

DPLL_FDEV

FINC/FDEC Register Control

DPLL_FDEV_REG_UPDATE

I²C/SPI

The DPLL Numerator is incremented or decremented by the

DCO FDEV step word on the rising-edge of FINC or FDEC.

0x160[0]

Write:

0 = FINC

1 = FDEC

图 56. DCO Mode Control Options

9.4.4.1 DCO Frequency Step Size

公式 12 shows the formula to compute the DPLL_FDEV register value required to meet the specified DCO

frequency step size in ppb (part-per-billion) when DCO mode is enabled for the DPLL.

DPLL_FDEV = (Reqd_ppb / 10⁹) × DEN_DPLL× f_VCO1/ (2 × PR_DPLL) / (f_REF/ R_REF

)

where

•

DPLL_FDEV: Frequency deviation value (0 to 2³⁸- 1)

Reqd_ppb: Required DCO frequency step size (in ppb)

DEN_DPLL: DPLL FB divider denominator value (1 to 2⁴⁰)

f_VCO1: VCO1 frequency

PR: DPLL feedback prescaler divide value (2 to 17)

f_REF: PRIREF or SECREF input frequency

Rx: PRIREF or SECREF input divide value (1 to 2¹⁶- 1)

(12)

9.4.4.2 DCO Direct-Write Mode

An alternate method to update the DCO frequency is to take the current numerator value (DPLL_REF_NUM) of

the DPLL fractional feedback divider, compute the adjusted numerator value by adding or subtracting the

DPLL_FDEV step value computed above, and write the adjusted numerator value through I²C or SPI.

9.4.5 Zero-Delay Mode Synchronization

The DPLL supports a zero-delay mode (ZDM) synchronization option to achieve a known and deterministic

phase relationship between the selected DPLL reference input and the OUT7 clock. This is primarily intended to

achieve phase alignment between a 1-PPS input and 1-PPS output. See Zero-Delay Mode (ZDM)

Synchronization for 1-PPS Input and Output.

56

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.5 Programming

9.5.1 Interface and Control

A system host device (MCU or FPGA) can use either I²C or SPI to access the register, SRAM, and EEPROM

maps. The register and EEPROM map configurations are the same for I²C and SPI. The device can be

initialized, controlled, and monitored through register access during normal operation (when PDN is deasserted).

Some device features can also be controlled and monitored through the external logic control and status pins.

In the absence of a host, the LMK05318 can self-start from its on-chip EEPROM or ROM page depending on the

state of HW_SW_CTRL pin. The EEPROM or ROM page is used to initialize the registers upon device POR. A

custom EEPROM configuration can be programmed in-system through the register interface by either I²C or SPI.

The ROM configurations are fixed in hardware and cannot be modified.

图 57 shows the device control pin, register, and memory interfaces. The arrows refer to the control interface

directions between the different blocks.

The register map has 435 data bytes. Some registers,such as status registers and internal test/diagnostic

registers (above R352), do not need to be written during device initialization.

The SRAM/EEPROM map has one register page with 256 data bytes. The SRAM/EEPROM map has fewer

bytes because not all bit fields are mapped from the register space. To program the EEPROM, it is necessary to

write the register contents to SRAM (internal register commit or direct write), then Program EEPROM with the

register contents from SRAM.

The ROM map has eight register pages with 249 data bytes per page. The ROM contents are fixed in hardware

and cannot be modified.

Mask ROM

Select ROM Mode

(8 Pages)

(Page 0 to 7)

Addr: 0x000 to 0x7C7

Data: 1992 bytes

(249 bytes / Page)

- Initialize Registers from ROM Page

Memory

Interface

STATUS0

STATUS1/FDEC

HW_SW_CTRL

PDN

Control/

Status Pins

Serial

Interface

Device

Control

and

Device Blocks

(Inputs, PLLs,

Outputs,

Registers

Block Interface

GPIO0/SYNCN

GPIO2/SDO/FINC

GPIO1/SCS

SCL/SCK

Addr: 0x000 to 0x1B2

Data: 435 bytes

Status

Monitors, etc.)

I²C/SPI

Pins

SDA/SDI

Memory

Interface

Memory

Interface

- Write SRAM (Commit Registers)

- Read SRAM

- Initialize Registers from EEPROM

- Read EEPROM

Program

EEPROM

SRAM

NVM EEPROM

Addr: 0x000 to 0x100

Data: 256 bytes

Addr: 0x000 to 0x100

Data: 256 bytes

Select EEPROM Mode

图 57. Device Control, Register, and Memory Interfaces

57

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Programming (接下页)

9.5.2 I²C Serial Interface

When started in I²C mode (HW_SW_CTRL = 0 or 1), the LMK05318 operates as an I²C slave and supports bus

rates of 100 kHz (standard mode) and 400 kHz (fast mode). Slower bus rates can work as long as the other I²C

specifications are met.

In EEPROM mode, the LMK05318 can support up to four different I²C addresses depending on the GPIO1 pins.

The 7-bit I²C address is 11001xxb, where the two LSBs are determined by the GPIO1 input levels sampled at

device POR and the five MSBs (11001b) are initialized from the EEPROM. In ROM mode, the two LSBs are

fixed to 00b, while the five MSB (11001b) are initialized from the EEPROM.

Write Transfer

1

7

1

S

Slave Address

Wr

A

8

1

Register Address High

A

Register Address Low

A

8

1

Data Byte

A

P

Read Transfer

1

7

1

S

Slave Address

Wr

A

8

1

Register Address High

A

Register Address Low

A

7

Slave Address

8

1

A

1

Sr

Rd

1

Data Byte

A

P

Legend

S

Sr

Start condition sent by master device

Write bit = 0 sent by master device

Acknowledge sent by master device

Stop condition sent by master device

|

Repeated start condition sent by master device

Read bit = 1 sent by master device

Wr Rd

A

P

N

A

Acknowledge sent by slave device

Not-acknowledge sent by master device

Data Data sent by master

|

Not-acknowledge sent by slave device

Data sent by slave

N

Data

图 58. I²C Byte Write and Read Transfers

58

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Programming (接下页)

9.5.2.1 I²C Block Register Transfers

The device supports I²C block write and block read register transfers as shown in 图 59.

Block Write Transfer

1

7

1

S

Slave Address

Wr

A

8

1

Register Address High

A

Register Address Low

A

8

1

8

1

Data Byte

A

Data Byte

A

P

Block Read Transfer

1

7

1

S

Slave Address

Wr

A

8

1

Register Address High

A

Register Address Low

A

7

Slave Address

8

1

A

1

Sr

Rd

8

1

Data Byte

A

Data Byte

A

P

图 59. I²C Block Register Transfers

9.5.3 SPI Serial Interface

When started in SPI mode (HW_SW_CTRL = Float or V_IM), the device uses a 4-wire SPI interface with SDI,

SCK, SDO, and SCS signals. The host device must present data to the device MSB first. A message includes a

transfer direction bit (W/R), a 15-bit address field (A14 to A0), and a 8-bit data field (D7 to D0) as shown in 图 60.

The W/R bit is 0 for a SPI write and 1 for a SPI read.

MSB

LSB

0

23 22 21 20 19 18 17 16 15 14 13 12 11 10

9

8

7

6

5

4

3

2

1

MSB Transmitted First

Bit Definition

A

10

A

9

A

8

A

7

A

6

A

5

A

4

A

3

A

2

A

1

A

0

D

7

D

6

D

5

D

4

D

3

D

2

D

1

D

0

A

First Out

14 13 12 11

Message Field Definition

Register Address (15 bits)

Data Payload (8 bits)

图 60. SPI Message Format

A message frame is initiated by asserting SCS low. The frame ends when SCS is deasserted high. The first bit

transferred is the W/R bit. The next 15 bits are the register address, and the remaining eight bits are data. On

write transfers, data is committed in bytes as the final data bit (D0) is clocked in on the rising edge of SCK. If the

write access is not an even multiple of eight clocks, the trailing data bits are not committed. On read transfers,

data bits are clocked out from the SDO pin on the falling edges of SCK.

59

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Programming (接下页)

9.5.3.1 SPI Block Register Transfer

The device supports a SPI block write and block read transfers. A SPI block transfer is exactly (2 + N) bytes

long, where N is the number of data bytes to write or read. The host device (SPI master) is only required to

specify the lowest address of the sequence of addresses to be accessed. The device will automatically increment

the internal register address pointer if the SCS pin remains low after the host finishes the initial 24-bit

transmission sequence. Each transfer of eight bits (a data payload width) results in the device automatically

incrementing the address pointer (provided the SCS pin remains active low for all sequences).

9.5.4 Register Map and EEPROM Map Generation

The TICS Pro software tool for EVM programming has a step-by-step design flow to enter the user-selected

clock design parameters, calculate the frequency plan, and generate the device register settings for the desired

configuration. The register map data (hex format) or SRAM/EEPROM map data can be exported to enable host

programming of the LMK05318 on start-up.

If desired, customers may send their TICS Pro setup file (.tcs) to TI to review and optimize the configuration

settings or to support factory pre-programmed samples.

9.5.5 General Register Programming Sequence

For applications that use a system host to program the initial LMK05318 configuration after start-up, this general

procedure can be followed from the register map data generated and exported from TICS Pro:

1. Apply power to the device to start in I²C or SPI mode. The PDN pin must be pulled high or driven high.

2. Write the register settings from lower to higher addresses (R0 to R352) while applying the following register

mask (do not modify mask bits = 1):

–

Mask R12 = A7h (Device reset/control register)

Mask R157 = FFh (NVM control bits register)

Mask R164 = FFh (NVM unlock bits register)

Mask R353 to R435 = FFh (Internal test/diagostic registers should not be written)

3. Write 1b to R12[7] to assert device soft-reset. This does not reset the register values.

4. Write 0b to R12[7] to exit soft-reset and begin the PLL start-up sequence.

5. See EEPROM Programming Using Method #1 (Register Commit) to store the active configuration to the

EEPROM to enable auto-startup on the next power cycle.

9.5.6 EEPROM Programming Flow

Before the EEPROM can be programmed, it is necessary to program the desired configuration to the SRAM

through the memory control registers. The register data can be written to the SRAM by transferring the active

register configuration internally using Method #1, or by direct writes to the SRAM using Method #2.

•

Method #1 (Register Commit) requires the active registers be first programmed to the desired configuration,

but does not require knowledge of the SRAM/EEPROM map.

•

Method #2 (Direct Writes) bypasses any writes to the active registers, allowing the device to continue normal

operation without disruption while the SRAM/EEPROM are programmed.

The programming flow for the two methods are different and described as follows.

9.5.6.1 EEPROM Programming Using Method #1 (Register Commit)

This sequence can be followed to program the SRAM and EEPROM using the active register configuration.

1. Program the desired configuration to the active registers (see General Register Programming Sequence).

This requires the register settings in the register map format.

2. Write SRAM Using Register Commit.

3. Program EEPROM.

60

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Programming (接下页)

9.5.6.1.1 Write SRAM Using Register Commit

The SRAM array is volatile shadow memory mapped to a subset of the active configuration registers and is used

to program the EEPROM.

Once the active registers have been programmed, the data can be internally committed to the SRAM with a

single register transaction:

1. Write 40h to R157 (REGCOMMIT bit, self-clearing). This commits the current register data to the SRAM

internally.

2. (optional) Program any of the user-programmable fields to SRAM. See User-Programmable Fields In

EEPROM. This step should not precede the prior step.

9.5.6.1.2 Program EEPROM

The EEPROM array is non-volatile memory mapped directly from the SRAM array.

After the register settings have been written to the SRAM (by either Method #1 or #2), the EEPROM can be

programmed through the following sequence:

1. Write EAh to R164 (NVMUNLK). This unlocks the EEPROM to allow programming.

2. Write 03h to R157 (NVM_ERASE_PROG bits). This programs the EEPROM from the entire contents of the

SRAM. The total erase/program cycle takes about 230 ms.

–

NOTE: Steps 1 and 2 must be atomic writes without any other register transactions in-between.

3. (optional) Read or poll R157[2] (NVMBUSY bit). When this bit cleared, the EEPROM programming is done.

4. (optional) Write 00h to R164. This locks the EEPROM to protect against inadvertent programming.

On the next power-up or hard-reset, the device can self-start in EEPROM mode from the newly programmed

configuration. Also, the NVMCNT register value will be incremented by 1 after power-up or hard-reset to reflect

total number of EEPROM programming cycles completed successfully.

9.5.6.2 EEPROM Programming Using Method #2 (Direct Writes)

This sequence can be followed to program the EEPROM by writing the SRAM directly to avoid disruption to the

current device operation. This requires the register settings in the SRAM/EEPROM map format.

1. Write SRAM Using Direct Writes.

2. Program EEPROM.

9.5.6.2.1 Write SRAM Using Direct Writes

This SRAM direct write method can be used if it is required to store a different device configuration to EEPROM

without disrupting the current operational state of the device. This method requires that the SRAM/EEPROM map

data is already generated, which can be exported by TICS Pro.

The SRAM can be directly written without modifying the active configuration registers through the following

sequence:

1. Write the most significant five bits of the SRAM address to R159 (MEMADR byte 1) and write the least

significant eight bits of the SRAM address to R160 (MEMADR byte 0).

2. Write the SRAM data byte to R162 (RAMDAT byte) for the address specified in the previous step in the

same register transaction.

–

Any additional write (or read) transfers in same transaction will cause the SRAM address pointer to be

auto-incremented and a subsequent write (or read) will take place at the next SRAM address.

–

Byte or Block write transfers to R162 can be used to write the entire SRAM map sequentially from Byte 0

to 252.

–

Bytes 253 to 255 must not be modified or overwritten and shall be reserved for TI internal use only.

–

Alternatively, it is valid to write R159 and R160 to set the memory address pointer explicitly before each

write to R162.

–

Access to the SRAM will terminate at the end of current write transaction.

Note that reading the RAMDAT register will also cause the memory address pointer to be auto-

incremented.

61

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Programming (接下页)

9.5.6.2.2 User-Programmable Fields In EEPROM

表 9 summarizes the address of several user-programmable bytes in EEPROM. These bytes can be written

using the SRAM direct write method prior to programming the EEPROM. It is optional to modify these bytes from

their factory default settings.

表 9. User-Programmable Fields

ADDRESS BYTE #

FIELD NAME

DESCRIPTION

(DECIMAL)

I²C Slave Address MSB Bits [7:3].

Bits [7:3] can be written to set the five MSBs of the 7-bit slave address. Bits [2:0]

should be written with zeros. The two LSBs of the 7-bit address are determined by the

control pins on device start-up. Default SLAVEADR[7:0] value = C8h (corresponds to

7-bit address of 64h).

5

SLAVEADR[7:0]

After the EEPROM is programmed and a subsequent POR cycle, the SLAVEADR

value stored in EEPROM can be read from R10.

EEPROM Image Revision.

This byte can be written to set the EEPROM image revision number or any customer-

specific data for part traceability.

11

EEREV[7:0]

After the EEPROM is programmed and a subsequent POR cycle, the EEREV value

stored in EEPROM can be read from R11.

249

250

251

252

NVM_SPARE_BY0[7:0]

NVM_SPARE_BY1[7:0]

NVM_SPARE_BY2[7:0]

NVM_SPARE_BY3[7:0]

NVM Spare Bytes.

These four bytes can be written with any customer-specific data for part traceability.

After the EEPROM is programmed, these bytes can be read directly from EEPROM

(see Read EEPROM ).

9.5.7 Read SRAM

The contents of the SRAM can be read back, one word at a time, starting with that of the requested address

through the following sequence. This sequence can be used to verify the contents of the SRAM before it is

transferred to the EEPROM during an EEPROM program cycle.

1. Write the most significant five bits of the SRAM address to R159 (MEMADR byte 1) and write the least

significant eight bits of the SRAM address to R160 (MEMADR byte 0).

2. Read R162 (RAMDAT byte) to fetch the SRAM data byte from the address specified in the previous step in

the same register transaction.

–

Any additional read transfers that are part of the same transaction will cause the SRAM address to auto-

increment and a subsequent read will take place at the next SRAM address.

Byte or Block read transfers from R162 can be used to read the entire SRAM map sequentially from Byte

0 to 252.

Access to SRAM will terminate at the end of current register transaction.

9.5.8 Read EEPROM

The contents of the EEPROM can be read back, one word at a time, starting with that of the requested address

through the following sequence. This sequence can be used to verify the EEPROM contents after the last

successful program cycle.

1. Write the most significant five bit of the EEPROM address in R159 (MEMADR byte 1) and write the least

significant eight bits of the EEPROM address in R160 (MEMADR byte 0).

2. Read R161 (NVMDAT byte) to fetch the EEPROM data byte from the address specified in the previous step

in the same register transaction.

–

Any additional read transfer that is part of the same transaction will cause the EEPROM address pointer

to be auto-incremented and a subsequent read will take place of the next address.

Byte or Block read transfers from R161 can be used to read the entire EEPROM map sequentially from

Byte 0 to 252.

Access to EEPROM will terminate at the end of current register transaction.

62

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

9.5.9 EEPROM Start-Up Mode Default Configuration

The generic LMK05318 device is factory pre-programmed with the EEPROM default configuration in 表 10. A

different start-up configuration can be stored to the EEPROM through in-system programming.

表 10. LMK05318 EEPROM Start-Up Default Configuration

SYSTEM CLOCK

XO

FREQUENCY (MHz)

INPUT TYPE

XO DOUBLER

48.0048

AC-DIFF(ext. term)

Disabled

CLOCK INPUTS

PRIREF

FREQUENCY (MHz)

INPUT TYPE

AUTO PRIORITY

25

AC-DIFF(ext. term)

1st

SECREF

2nd

MANUAL REGISTER

SELECTION

INPUT SELECTION

DPLL

INPUT SELECT MODE

MANUAL SELECTION MODE

Manual with Auto-Fallback

Pin Select

PRIREF

CLOCK OUTPUTS

OUT0

FREQUENCY (MHz)

OUTPUT MUX

PLL 1

OUTPUT TYPE

Disabled

156.25

25

OUT1

PLL 1

Disabled

OUT2

PLL 1

AC-LVPECL

Disabled

OUT3

PLL 1

OUT4

PLL 1

OUT5

PLL 1

Disabled

OUT6

PLL 1

AC-LVPECL

HCSL (ext. 50-Ω term.)

OUT7

100

PLL 1

PLL CONFIGURATION

PLL MODE

DPLL Mode

Disabled

LOOP BW (Hz)

TDC or PFD RATE (MHz)

DPLL

APLL1

APLL2

100

1000

–

25

24.0048

–

REF INPUT MONITORS (1)

PRIREF

VALIDATION TIMER (s)

FREQ DET VALID (ppm)

FREQ DET INVALID (ppm)

0.1

-

SECREF

1-PPS JITTER THRESHOLD

REF INPUT MONITORS (2)

EARLY DETECT WINDOW (ns)

LATE DETECT WINDOW (ns)

(μs)

PRIREF

33.6

46.4

–

SECREF

FREQUENCY LOCK DETECT

DPLL

LOCK (ppm)

UNLOCK (ppm)

ACCURACY (ppm)

1

10

1

DCO MODE

DPLL

DCO CONTROL

STEP SIZE (ppb)

FINC/FDEC MODE

DCO Disabled

–

Register bit

ZERO-DELAY MODE

REF-to-OUT7

DPLL ZDM SYNC

PHASE OFFSET (ns)

ZDM Disabled

–

STATUS PINS

STATUS0

SIGNAL

TYPE

POLARITY

Active High

DPLL Loss of Frequency Lock

DPLL Holdover Active

3.3-V LVCMOS

STATUS1

63

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

10 Application and Implementation

注

Information in the following applications sections is not part of the TI component

specification, and TI does not warrant its accuracy or completeness. TI’s customers are

responsible for determining suitability of components for their purposes. Customers should

validate and test their design implementation to confirm system functionality.

10.1 Application Information

10.1.1 Device Start-Up Sequence

The device start-up sequence is shown in 图 61.

64

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

Application Information (接下页)

Power-On Reset

(POR)

Device POR

Configuration Sequence

PDN = 0

Hard Reset?

Outputs muted

PDN = 1

HW_SW_CTRL = 1

HW_SW_CTRL = 0

GPIO[2:0] = 000b to 111b

(ROM Page Select)

GPIO1 = 0, Float, 1 (I²C Addr. Select)

Start-up Mode?

(sample pin states)

HW_SW_CTRL = Float

STATUS[1:0] = Float

EEPROM + I²C

EEPROM + SPI

ROM + I²C

(Soft Pin) Mode

(Hard Pin) Mode

Registers initialize from EEPROM/ROM and I²C/SPI,

Control & Status pins activate (after hard reset only).

All blocks reset to initial states.

RESET_SW = 0

Soft Reset?

RESET_SW = 1

Device Block

Configuration

Register and EEPROM programming available.

XO Detected

VCO Calibration

PLL Initialization Sequence

APLL2 locks before APLL1 in Cascaded mode.

Outputs lock to XO frequency.

APLL(s) Locked

Outputs un-mute if DPLL auto-mute disabled.

Outputs auto-SYNC if enabled.

(Free-run from XO)

Input Monitoring (fastest to slowest detector):

1. Missing and/or Early clock detector

2. Amplitude or Slew rate detector

3. Frequency (ppm) detector

4. 1-PPS phase valid detector (skip #1 and 3 for 1-PPS)

5. After enabled detectors are valid, validation timer starts

and must finish before input is qualified.

Ref. Input

Validation

Valid Input Selected

DPLL

Lock Acquisition

Fastlock DPLL bandwidth is temporarily asserted

during lock acquisition.

Outputs lock to selected input clock frequency.

Outputs are un-muted if DPLL auto-mute enabled.

Normal DPLL bandwidth is asserted.

DCO Mode control

available (ZDM

must be disabled)

DPLL

Locked

DPLL frequency- and phase-lock detectors are monitored.

Output will have deterministic input-to-output phase

relationship if Zero-Delay Mode (ZDM) SYNC is enabled.

See DPLL Modes

and

Input Selection

Flowcharts

图 61. Device Start-Up Sequence

65

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

Application Information (接下页)

10.1.2 Power Down (PDN) Pin

The PDN pin (active low) can be used for device power down and used to initialize the POR sequence. When

PDN is pulled low, the entire device is powered down and the serial interface is disabled. When PDN is pulled

high, the device POR sequence is triggered to begin the device start-up sequence and normal operation as

depicted in 图 61. If the PDN pin is toggled to issue a momentary hard-reset, the negative pulse applied to the

PDN pin should be greater than 200 ns to be captured by the internal digital system clock.

表 11. PDN Control

PDN PIN STATE

DEVICE OPERATION

Device is disabled

Normal operation

0

1

10.1.3 Power Rail Sequencing, Power Supply Ramp Rate, and Mixing Supply Domains

10.1.3.1 Mixing Supplies

The LMK05318 incorporates flexible power supply architecture. While all VDD core supplies should be powered

by the same 3.3-V rail, the individual output supplies can be powered from separate 1.8-V, 2.5-V, or 3.3-V rails.

This can allow all VDDO output supplies to operate at 1.8 V to minimize power consumption.

10.1.3.2 Power-On Reset (POR) Circuit

The LMK05318 integrates a built-in power-on reset (POR) circuit that holds the device in reset until all of the

following conditions have been met:

•

All VDD core supplies have ramped above 2.72 V

PDN pin has ramped above 1.2 V (minimum V_IH)

10.1.3.3 Powering Up From a Single-Supply Rail

As long as all VDD core supplies are driven by the same 3.3-V supply rail that ramp in a monotonic manner from

0 V to 3.135 V, irrespective of the ramp time, then there is no requirement to add a capacitor on the PDN pin to

externally delay the device power-up sequence. As shown in 图 62, the PDN pin can be left floating or otherwise

driven by a system host to meet the clock sequencing requirements in the system.

VDD_PLLx, VDD_IN,

3.135 V

VDD_DIG,

VDD_XO,VDDOx,

PDN

Decision Point 2:

VDD_PLLx/VDD_IN/

VDD_DIG / VDD_XO /

VDDOx ≥ 2.72 V

VDD_IN

200 kꢀ

Decision Point 1:

PDN ≥ 1.2 V

PDN

0 V

图 62. Recommendation for Power Up From a Single-Supply Rail

66

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

10.1.3.4 Power Up From Split-Supply Rails

If some VDD core supplies are driven from different supply rails, TI recommends to start the PLL calibration after

all of the core supplies have ramped above 3.135 V. This can be realized by delaying the PDN low-to-high

transition. The PDN input incorporates a 200-kΩ resistor to VDD_IN and as shown in 图 63, a capacitor from the

PDN pin to GND can be used to form an R-C time constant with the internal pullup resistor. This R-C time

constant can be designed to delay the low-to-high transition of PDN until all the core supplies have ramped

above 3.135 V.

Alternatively, the PDN pin can be driven high by a system host or power management device to delay the device

power-up sequence until all VDD supplies have ramped.

VDD_PLLx,

VDD_IN,

3.135 V

Decision Point 3

VDD_PLLx/

VDD_DIG,

VDDXO

VDD_IN/ VDD_DIG/

VDDXO

VDD_IN

≥ 2.72 V

Decision Point 2:

VDDOx ≥ 1.7 V

VDDO_x,

PDN

200 kΩ

PDN

Decision Point 1:

PDN ≥ 1.2 V

C_PDN

Delay

0 V

图 63. Recommendation for Power Up From Split-Supply Rails

10.1.3.5 Non-Monotonic or Slow Power-Up Supply Ramp

In case the VDD core supplies ramp with a non-monotonic manner or with a slow ramp time from 0 V to 3.135 V

of over 100 ms, TI recommends to delay the VCO calibration until after all of the core supplies have ramped

above 3.135 V. This could be achieved by delaying the PDN low-to-high transition with one of the methods

described in Power Up From Split-Supply Rails.

If any core supply cannot ramp above 3.135 V before the PDN low-to-high transition, it is acceptable to issue a

device soft-reset after all core supplies have ramped to manually trigger the VCO calibration and PLL start-up

sequence.

10.1.4 Slow or Delayed XO Start-Up

Because the external XO clock input is used as the reference input for the VCO calibration, the XO input

amplitude and frequency must be stable before the start of VCO calibration to ensure successful PLL lock and

output start-up. If the XO clock is not stable prior to VCO calibration, the VCO calibration can fail and prevent

PLL lock and output clock start-up.

If the XO clock has a slow start-up time or has glitches on power-up (due to a slow or non-monotonic power

supply ramp, for example), TI recommends to delay the start of VCO calibration until after the XO is stable. This

could be achieved by delaying the PDN low-to-high transition until after the XO clock has stabilized using one of

the methods described in Power Up From Split-Supply Rails. It is also possible to issue a device soft-reset after

the XO clock has stabilized to manually trigger the VCO calibration and PLL start-up sequence.

67

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

10.2 Typical Application

图 64 shows a reference schematic to help implement the LMK05318 and its peripheral circuitry. Power filtering

examples are given for the core supply pins and independent output supply pins. Single-ended LVCMOS, AC-

coupled differential, and HCSL clock interfacing examples are shown for the clock input and output pins. An

external LVCMOS oscillator drives an AC-coupled voltage divider network as an example to interface the 3.3-V

LVCMOS output to meet the input voltage swing specified for the XO input. The required external capacitors are

placed close to the LMK05318 and are shown with the suggested values. External pullup and pulldown resistor

options at the logic I/O pins set the default input states. The I²C or SPI pins and other logic I/O pins can be

connected to a host device (not shown) to program and control the LMK05318 and monitor its status. This

example assumes the device will start up from EEPROM mode with an I²C interface (HW_SW_CTRL = 0).

68

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

10.2.1 Design Requirements

In a typical application, consider the following design requirements or parameters to implement the overall clock

solution:

1. Device initial configuration. The device should be configured as either host programmed (MCU or FPGA) or

factory pre-programmed.

2. Device start-up mode and serial interface. Typically, this will be EEPROM + I²C or SPI mode.

3. XO frequency, signal type, and frequency accuracy and stability. Consider a high-stability TCXO or OCXO

for the XO input if any of the following is required:

–

Standard-compliant frequency stability (such as SyncE, SONET/SDH, IEEE 1588)

Lowest possible close-in phase noise at offsets ≤ 100 Hz

Narrow DPLL bandwidth ≤ 10 Hz

4. For the DPLL/APLL1 domain, determine the following:

–

Input clocks: frequency, buffer mode, priority, and input selection mode

Output clocks: frequency, buffer mode

DPLL loop bandwidth and maximum TDC frequency

If the DCO Mode or Zero-Delay Mode is required

5. For the APLL2 domain, determine the following:

–

APLL2 reference: VCO1 for synchronous clocking with Cascaded APLL2, or XO for asynchronous

clocking with Non-cascaded APLL2

–

Output clocks: frequency, buffer mode

6. Input clock and PLL monitoring options

7. Status outputs and interrupt flag

8. Power supply rails

10.2.2 Detailed Design Procedure

In a typical application, TI recommends the following steps:

1. Use the LMK05318 GUI in the TICS Pro programming software for a step-by-step design flow to enter the

design parameters, calculate the frequency plan for each PLL domain, and generate the register settings for

the desired configuration. The register settings can be exported (in hex format) to enable host programming

or factory pre-programming.

–

If using a generic (non-custom) device, a host device can program the register settings through the serial

interface after power-up and issue a soft-reset (by RESET_SW bit) to start the device. The host can also

store the settings to the EEPROM to allow automatic start-up with these register settings on subsequent

power-on reset cycles.

–

Alternatively, a LMK05318 setup file for TICS Pro (.tcs) can be sent to TI to support factory pre-

programmed samples.

2. Tie the HW_SW_CTRL pin to ground to select EEPROM+I²C mode, or bias the pin to V_IMthrough the weak

internal resistors or external resistors to select EEPROM+SPI mode. Determine the logic I/O pin assignments

for control and status functions. See Device Start-Up Modes.

–

Connect I²C/SPI and logic I/O pins (1.8-V compatible levels) to the host device pins with the proper I/O

direction and voltage levels.

3. Select an XO frequency by following Oscillator Input (XO_P/N).

–

Choose an XO with target phase jitter performance that meets the frequency stability and accuracy

requirements required for the output clocks during free-run or holdover.

For a 3.3-V LVCMOS driver, follow the OSC clock interface example in 图 64. Power the OSC from a low-

noise LDO regulator or optimize its power filtering to avoid supply noise-induced jitter on the XO clock.

TICS Pro: Configure the XO input buffer mode to match the XO driver interface requirements. See 表 2.

70

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

4. Wire the clock I/O for each PLL domain in the schematic and use TICS Pro to configure the device settings

as follows:

–

Reference inputs: Follow the LVCMOS or differential clock input interface examples in 图 64 or Clock

Input Interfacing and Termination.

–

TICS Pro: For DPLL mode, configure the reference input buffer modes to match the reference clock

driver interface requirements. See 表 3.

–

LVCMOS clock input should be used for input frequencies below 5 MHz when amplitude monitoring is

enabled.

–

TICS Pro: For DPLL mode, configure the DPLL input selection modes and input priorities. See Reference

Input Mux Selection.

TICS Pro: If APLL2 is used, configure the APLL2 reference for VCO1 domain (Cascaded APLL2) or XO

clock (Non-cascaded APLL2).

TICS Pro: Configure each output with the required clock frequency and PLL domain. TICS Pro can

calculate the VCO frequencies and divider settings for the PLL and outputs. Consider the following output

clock assignment guidelines to minimize crosstalk and spurs:

–

OUT[0:3] bank is preferred for PLL1 clocks.

OUT[4:7] bank is preferred for PLL2 clocks.

Group identical output frequencies (or harmonic frequencies) on adjacent channels, and use the

output pairs with a single divider (OUT0/1 or OUT2/3) when possible to minimize power.

–

Separate clock outputs when the difference of the two frequencies, |f_OUTx– f_OUTy|, falls within the jitter

integration bandwidth (12 kHz to 20 MHz, for example). Any outputs that are potential aggressors

should be separated by at least four static pins (power pin, logic pin, or disabled output pins) to

minimize potential coupling. If possible, separate these clocks by the placing them on opposite output

banks, which are on opposite sides of the chip for best isolation.

–

Avoid or isolate any LVCMOS output (strong aggressor) from other jitter-sensitive differential output

clocks. If an LVCMOS output is required, use dual complementary LVCMOS mode (+/- or -/+) with the

unused LVCMOS output left floating with no trace.

If not all outputs pairs are used in the application, consider connecting an unused output to a pair of

RF coaxial test structures for testing purposes (such as SMA, SMP ports).

–

TICS Pro: Configure the output drivers.

–

Configure the output driver modes to match the receiver clock input interface requirements. See 表 7.

Configure any output SYNC groups that need their output phases synchronized. See Output

Synchronization (SYNC).

–

Configure the output auto-mute modes, output mute levels, and APLL and DPLL mute options. See

Output Auto-Mute During LOL.

Clock output Interfacing: Follow the single-ended or differential clock output interface examples in 图 64

or Clock Output Interfacing and Termination.

–

Differential outputs should be AC-coupled and terminated and biased at the receiver inputs.

HCSL outputs should have 50-Ω termination to GND (at source or load side) unless the internal

source termination is enabled by programming.

–

LVCMOS outputs have internal source termination to drive 50-Ω traces directly. LVCMOS V_OHlevel is

determined by VDDO voltage (1.8 V).

–

TICS Pro: Configure the DPLL loop bandwidth.

–

Below the loop bandwidth, the reference noise is added to the TDC noise floor and the

XO/TCXO/OCXO noise. Above the loop bandwidth, the reference noise will be attenuated with roll-off

up to 60 dB/decade. The optimal bandwidth depends on the relative phase noise between the

reference input and the XO. APLL1's loop bandwidth can be configured to provide additional

attenuation of the reference input, TDC, and XO phase noise above APLL1's bandwidth (typically

around 1 kHz).

TICS Pro: Configure the maximum TDC frequency to optimize the DPLL TDC noise contribution for the

desired use case.

–

Wired: The maximum TDC rate is preset to 400 kHz. This supports SyncE and other use cases using

a narrow loop bandwidth (≤10 Hz) with a TCXO/OCXO/XO to set the frequency stability and wander

performance.

71

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

–

Wireless: The maximum TDC rate is preset to 26 MHz for lowest in-band TDC noise contribution. This

supports wireless and other use cases where close-in phase noise is critical.

Custom: The maximum TDC rate can be specified for any value up to 26 MHz.

–

TICS Pro: If clock steering is needed (such as for IEEE 1588 PTP), enable DCO mode for the DPLL loop

and enter the frequency step size (in ppb). The FDEV step register will be computed according to DCO

Frequency Step Size. Enable the FINC/FDEC pin control on the GPIO pins if needed.

TICS Pro: If deterministic input-to-output clock phase is needed for 1-PPS input and 1-PPS output (on

OUT7), enable the ZDM and OUT7 divider synchronization features. See Zero-Delay Mode (ZDM)

Synchronization for 1-PPS Input and Output.

5. TICS Pro: Configure the reference input monitoring options for each reference input. Disable the monitor

when not required or when the input operates beyond the monitor's supported frequency range. See

Reference Input Monitoring.

–

Amplitude monitor: Set the LVCMOS detected slew rate edge or the differential input amplitude threshold

to monitor input signal quality. Disable the monitor for a differential input below 5 MHz or else use an

LVCMOS input clock.

–

Frequency monitor: Set the valid and invalid thresholds (in ppm).

Missing pulse monitor: Set the late window threshold (T_LATE) to allow for the longest expected input clock

period, including worst-case cycle-to-cycle jitter. For a gapped clock input, set T_LATEbased on the number

of allowable missing clock pulses.

–

Runt pulse monitor: Set the early window threshold (T_EARLY) to allow for the shortest expected input clock

period, including worst-case cycle-to-cycle jitter.

1-PPS Phase validation monitor: Set the phase validation jitter threshold, including worst-case input

cycle-to-cycle jitter.

Validation timer: Set the amount of time the reference input must be qualified by all enabled input

monitors before the input is valid for selection.

6. TICS Pro: Configure the DPLL lock detect and tuning word history monitoring options for each channel. See

PLL Lock Detectors and Tuning Word History.

–

DPLL tuning word history: Set the history count/averaging time (T_AVG), history delay/ignore time (T_IGN),

and intermediate averaging option.

–

DPLL frequency lock and phase lock detectors: Set the lock and unlock thresholds for each detector.

7. TICS Pro: Configure each status output pin and interrupt flag as needed. See Status Outputs and Interrupt.

–

Select the desired status signal selection, status polarity, and driver mode (3.3-V LVCMOS or open-

drain). Open-drain requires an external pullup resistor.

–

If the Interrupt is enabled and selected as a status output, configure the flag polarity and the mask bits for

any interrupt source, and the combinational AND/OR gate as needed.

8. Consider the following guidelines for designing the power supply:

–

Outputs with identical frequency or integer-related (harmonic) frequencies can share a common filtered

power supply.

–

Example: 156.25-MHz and 312.5-MHz outputs on OUT[0:1] and OUT[2:3] can share a filtered VDDO

supply (Group 1), while 100-MHz, 50-MHz, and 25-MHz outputs on OUT[4:7] can share a separate

VDDO supply (Group 2).

–

For lowest power, AC-DIFF or HCSL outputs can be powered from a 1.8-V supply with no degradation in

output swing or phase noise (compared to 2.5 V or 3.3 V).

–

1.8-V LVCMOS outputs should be powered from a 1.8-V supply.

See Power Rail Sequencing, Power Supply Ramp Rate, and Mixing Supply Domains.

72

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

10.2.3 Application Curves

Unless otherwise noted, test conditions are the same as in Typical Characteristics. 156.25-MHz output (APLL1) on OUT3 and

155.52 MHz output (APLL2) on OUT4 running simultaneously to demonstrate minimal coupling between the PLL domains

and minimal degradation in phase noise and jitter. Device operating in DPLL Mode with Cascaded APLL2. AC-LVPECL

outputs measured.

Jitter = 65 fs RMS (12 kHz to 20 MHz)

f_VCO1= 2500 MHz (VCO1 inverted)

Jitter = 136 fs RMS (12 kHz to 20 MHz)

f_VCO2= 6065.28 MHz (VCO2 post-divider = 3)

图 65. 156.25-MHz Output Phase Noise (OUT3)

With 155.52-MHz Output Enabled (OUT4)

图 66. 155.52-MHz Output Phase Noise (OUT4)

With 156.25-MHz Output Enabled (OUT3)

10.3 Do's and Don'ts

•

Power all the VDD pins with proper supply decoupling and bypassing connect as shown in 图 64.

Power down unused blocks through registers to minimize power consumption.

Use proper source or load terminations to match the impedance of input and output clock traces for any

active signals to/from the device.

•

Leave unused clock outputs floating and powered down through register control.

Leave unused clock inputs floating.

For EEPROM+SPI Mode: Leave HW_SW_CTRL and STATUS[1:0] pins floating during POR to ensure proper

start-up. These pins has internal biasing to V_IMinternally.

–

If HW_SW_CTRL or either STATUS pin is connected to a system host (MCU or FPGA), the host device

must be configured with high-impedance input (no pullup or pulldown resistors) to avoid conflict with the

internal bias to V_IM. If needed, external biasing resistors (10-kΩ pullup to 3.3 V and 3.3-kΩ pulldown) can

be connected on each STATUS pin to bias the inputs to V_IMduring POR.

•

Consider routing each STATUS pin to a test point or high-impedance input of a host device to monitor device

status outputs.

Consider using a LDO regulator to power the external XO/TCXO/OCXO source.

–

High jitter and spurious on the oscillator clock are often caused by high spectral noise and ripple on its

power supply.

•

Include dedicated header to access the I²C or SPI interface of the device, as well as a header pin for ground.

–

This can enabled off-board programming for device bring-up, prototyping, and diagnostics using the TI

USB2ANY interface and TICS Pro software tools.

73

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

11 Power Supply Recommendations

11.1 Power Supply Bypassing

图 67 shows two general placements of power supply bypass capacitors on either the back side or the

component side of the PCB. If the capacitors are mounted on the back side, 0402 components can be employed.

For component side mounting, use 0201 body size capacitors to facilitate signal routing. A combination of

component side and back side placement can be used. Keep the connections between the bypass capacitors

and the power supply on the device as short as possible. Ground the other side of the capacitor using a low-

impedance connection to the ground plane.

(Does not indicate actual location of the LMK05318 supply pins)

图 67. Generalized Placement of Power Supply Bypass Capacitors

74

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

11.2 Device Current and Power Consumption

The device power consumption is dependent on the actual configuration programmed to the device. The

individual supply pin current consumption values in Electrical Characteristics can be used to estimate device

power consumption and power supply dimensioning.

11.2.1 Current Consumption Calculations

Core supply currents:

IDD_CORE = IDD_DIG + IDD_IN + IDD_XO + IDD_PLL1 + IDD_PLL2

(13)

(14)

(15)

OUT[0:1] or OUT[2:3] channel supply current:

IDDO_XY = IDDO_XY_DIVIDER+ IDDO_X_DRIVER+ IDDO_Y_DRIVER

OUT[4:7] channel supply current:

IDDO_X = IDDO_X_DIVIDER+ IDDO_X_DRIVER

When an output channel's divider and drivers are disabled, its IDDO_x equals approximately 0 mA.

11.2.2 Power Consumption Calculations

Core power consumption:

P_CORE= IDD_CORE × VDD

(16)

(17)

(18)

Output power consumption:

P_OUT= (IDDO_01 × VDDO_01) + (IDDO_23 × VDDO_23) + ... + (IDDO_7 × VDDO_7)

Total device power consumption:

P_TOTAL= P_CORE+ P_OUT

11.2.3 Example

Estimate the current and power consumption for the following device configuration:

•

VDD = 3.3 V and VDDO_x = 1.8 V

DPLL/APLL1 mode with Cascaded APLL2

XO: 48 MHz, PRIREF and SECREF: 25 MHz

OUT[0:1]: 156.25 MHz AC-LVPECL (x2), PLL1

OUT[2:3]: 156.25 MHz AC-CML (x2), PLL1

OUT4: 133.33 MHz AC-LVDS, PLL2

OUT5: Disabled

OUT6: 100 MHz HCSL, PLL1

OUT7: 25 MHz LVCMOS (x2), PLL1

From 公式 13: IDD_CORE = 18 + 38 + 20 + 110 + 120 = 306 mA

From 公式 14 and 公式 15:

•

IDDO_01 = 70 + 16 + 16 = 102 mA

IDDO_23 = 70 + 14 + 14 = 98 mA

IDDO_4 = 70 + 10 = 80 mA

IDDO_5 = 0 mA

IDDO_6 = 70 + 25 = 95 mA

IDDO_7 = 70 + 6 = 76 mA

From 公式 16: P_CORE= 306 mA × 3.3 V = 1.01 W

From 公式 17: P_OUT= (102 + 98 + 80 + 95 + 76) mA × 1.8 V = 0.812 W

From 公式 18: P_TOTAL= 1.01 W + 0.812 W = 1.822 W

75

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

12 Layout

12.1 Layout Guidelines

•

Isolate input, XO/OCXO/TCXO and output clocks from adjacent clocks with different frequencies and other

nearby dynamic signals.

•

Consider the XO/OCXO/TCXO placement and layout in terms of the supply/ground noise and thermal

gradients from nearby circuitry (for example, power supplies, FPGA, ASIC) as well as system-level vibration

and shock. These factors can affect the frequency stability/accuracy and transient performance of the

oscillator.

•

Avoid impedance discontinuities on controlled-impedance 50-Ω single-ended (or 100-Ω differential) traces for

clock and dynamic logic signals.

Place bypass capacitors close to the VDD and VDDO pins on the same side as the IC, or directly below the

IC pins on the opposite side of the PCB. Larger decoupling capacitor values can be placed further away.

•

Place external capacitors close to the CAP_x and LFx pins.

Use multiple vias to connect wide supply traces to the respective power islands or planes if possible.

Use at least a 5×5 through-hole via pattern to connect the IC ground/thermal pad to the PCB ground planes.

See the Land Pattern Example, Solder Mask Details, and Solder Paste Example in the 机械、封装和可订购信

息.

12.2 Layout Example

图 68. General PCB Ground Layout for Thermal Reliability (8+ Layers Recommended)

76

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

12.3 Thermal Reliability

The LMK05318 is a high-performance device. To ensure good electrical and thermal performance, TI

recommends to design a thermally-enhanced interface between the IC ground/thermal pad and the PCB ground

using at least a 5×5 through-hole via pattern connected to multiple PCB ground layers as shown in 图 68.

12.3.1 Support for PCB Temperature up to 105°C

The device can maintain a safe junction temperature below the recommended maximum value of 125°C even

when operated on a PCB with a maximum board temperature (T_PCB) of 105 °C. This can shown by the following

example calculation, which assumes the total device power (P_TOTAL) computed with all blocks enabled using the

typical current consumption from the Electrical Characteristics (VDD = 3.3 V, VDDO = 1.8 V) and the thermal

data in Thermal Information: 10-Layer Custom PCB with no airflow.

T_J= T_PCB+ (Ψ_JB× P_TOTAL) = 113.8 °C

where

•

T_PCB= 105 °C

Ψ_JB= 4.4 °C/W

P_TOTAL= P_CORE+ P_OUTPUT= 2.0 W

–

P_CORE= (18 + 38 + 20 + 110 + 120) mA × 3.3 V = 1.01 W

DPLL, APLL1, APLL2, and all Inputs enabled

P_OUTPUT= (102 + 102 + 86 + 86 + 86 + 86) mA × 1.8 V = 0.986 W

All output channels enabled with output divider values > 6 and AC-LVPECL output types

–

(19)

77

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

13 器件和文档支持

13.1 器件支持

13.1.1 TICS Pro

TICS Pro 是用于 EVM 编程的离线软件工具，也可以用生成寄存器映射，为特定应用的器件配置编程。如需 TICS

Pro，请访问 www.ti.com.cn/tool/cn/TICSPRO-SW。

13.1.2 相关文档

请参阅以下文档：

•

《LMK05318 的 ITU-T G.8262 合规性测试结果》(SNAA316)

《通过 LMK05318 支持的同步模式》(SNAA324)

《使用 LMK05318 实现对高速 56G PAM-4 串行链路的计时》(SNAA325)

《LMK05318EVM 用户指南》(SNAU236)

13.2 接收文档更新通知

如需接收文档更新通知，请导航至 TI.com.cn 上的器件产品文件夹。单击右上角的通知我进行注册，即可每周接收

产品信息更改摘要。有关更改的详细信息，请查看任何已修订文档中包含的修订历史记录。

13.3 社区资源

下列链接提供到 TI 社区资源的连接。链接的内容由各个分销商“按照原样”提供。这些内容并不构成 TI 技术规范，

并且不一定反映 TI 的观点；请参阅 TI 的《使用条款》。

TI E2E™ 在线社区 TI 的工程师对工程师 (E2E) 社区。此社区的创建目的在于促进工程师之间的协作。在

e2e.ti.com 中，您可以咨询问题、分享知识、拓展思路并与同行工程师一道帮助解决问题。

设计支持

TI 参考设计支持可帮助您快速查找有帮助的 E2E 论坛、设计支持工具以及技术支持的联系信息。

13.4 商标

E2E is a trademark of Texas Instruments.

All other trademarks are the property of their respective owners.

13.5 静电放电警告

ESD 可能会损坏该集成电路。德州仪器 (TI) 建议通过适当的预防措施处理所有集成电路。如果不遵守正确的处理措施和安装程序 , 可

能会损坏集成电路。

ESD 的损坏小至导致微小的性能降级 , 大至整个器件故障。精密的集成电路可能更容易受到损坏 , 这是因为非常细微的参数更改都可

能会导致器件与其发布的规格不相符。

13.6 术语表

SLYZ022 — TI 术语表。

这份术语表列出并解释术语、缩写和定义。

78

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

14 机械、封装和可订购信息

以下页面包含机械、封装和可订购信息。这些信息是指定器件的最新可用数据。数据如有变更，恕不另行通知，且

不会对此文档进行修订。如需获取此数据表的浏览器版本，请查阅左侧的导航栏。

79

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

PACKAGE OUTLINE

VQFN - 1 mm max height

RGZ0048A

PLASTIC QUADFLAT PACK- NO LEAD

A

7.1

6.9

B

7.1

6.9

PIN 1 INDEX AREA

1 MAX

C

SEATING PLANE

0.05

0.00

0.08

C

2X 5.5

5.15 0.1

(0.2) TYP

13

24

44X 0.5

12

25

SYMM

2X

5.5

1

36

0.30

0.18

PIN1 ID

(OPTIONAL)

48X

48

37

SYMM

0.1

C A B

0.5

0.3

48X

0.05

C

4219044/A 05/2018

NOTES:

1. All linear dimensions are in millimeters. Any dimensions in parenthesis are for reference only. Dimensioning and tolerancing

per ASME Y14.5M.

2. This drawing is subject to change without notice.

3. The package thermal pad must be soldered to the printed circuit board for optimal thermal and mechanical performance.

www.ti.com

80

LMK05318

www.ti.com.cn

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

EXAMPLE BOARD LAYOUT

VQFN - 1 mm max height

RGZ0048A

PLASTIC QUADFLAT PACK- NO LEAD

2X (6.8)

5.15)

SYMM

(

48X (0.6)

48X (0.24)

44X (0.5)

35

48

1

34

SYMM

2X

(5.5)

2X

(6.8)

2X

(1.26)

2X

(1.065)

(R0.05)

TYP

23

12

21X (Ø0.2) VIA

TYP

22

13

2X (1.065)

2X (1.26)

2X (5.5)

LAND PATTERN EXAMPLE

SCALE: 15X

SOLDER MASK

OPENING

0.07 MIN

ALL AROUND

0.07 MAX

ALL AROUND

EXPOSED METAL

METAL

SOLDER MASK

OPENING

METAL UNDER

SOLDER MASK

NON SOLDER MASK

DEFINED

(PREFERRED)

SOLDER MASK

DEFINED

4219044/A 05/2018

SOLDER MASK DETAILS

NOTES: (continued)

4. This package is designed to be soldered to a thermal pad on the board. For more information, see Texas Instruments literature

number SLUA271 (www.ti.com/lit/slua271)

.

5. Vias are optional depending on application, refer to device data sheet. If any vias are implemented, refer to their locations shown

on this view. It is recommended that vias under paste be ﬁlled, plugged or tented.

www.ti.com

81

LMK05318

ZHCSID6A –DECEMBER 2018–REVISED DECEMBER 2018

www.ti.com.cn

EXAMPLE STENCIL DESIGN

VQFN - 1 mm max height

RGZ0048A

PLASTIC QUADFLAT PACK- NO LEAD

2X (6.8)

SYMM

(

1.06)

48X (0.6)

48X (0.24)

44X (0.5)

SYMM

2X

(5.5)

2X

(6.8)

2X

(0.63)

2X

(1.26)

(R0.05)

TYP

2X

(1.26)

2X (0.63)

2X (5.5)

SOLDER PASTE EXAMPLE

BASED ON 0.125 mm THICK STENCIL

EXPOSED PAD

67% PRINTED COVERAGE BY AREA

SCALE: 15X

4219044/A 05/2018

NOTES: (continued)

6. Laser cutting apertures with trapezoidal walls and rounded corners may oﬀer better paste release. IPC-7525 may have alternate

design recommendations.

www.ti.com

82

重要声明和免责声明

TI 均以“原样”提供技术性及可靠性数据（包括数据表）、设计资源（包括参考设计）、应用或其他设计建议、网络工具、安全信息和其他资

源，不保证其中不含任何瑕疵，且不做任何明示或暗示的担保，包括但不限于对适销性、适合某特定用途或不侵犯任何第三方知识产权的暗示

担保。

所述资源可供专业开发人员应用TI 产品进行设计使用。您将对以下行为独自承担全部责任：(1) 针对您的应用选择合适的TI 产品；(2) 设计、

验证并测试您的应用；(3) 确保您的应用满足相应标准以及任何其他安全、安保或其他要求。所述资源如有变更，恕不另行通知。TI 对您使用

所述资源的授权仅限于开发资源所涉及TI 产品的相关应用。除此之外不得复制或展示所述资源，也不提供其它TI或任何第三方的知识产权授权

许可。如因使用所述资源而产生任何索赔、赔偿、成本、损失及债务等，TI对此概不负责，并且您须赔偿由此对TI 及其代表造成的损害。

TI 所提供产品均受TI 的销售条款 (http://www.ti.com.cn/zh-cn/legal/termsofsale.html) 以及ti.com.cn上或随附TI产品提供的其他可适用条款的约

束。TI提供所述资源并不扩展或以其他方式更改TI 针对TI 产品所发布的可适用的担保范围或担保免责声明。IMPORTANT NOTICE

邮寄地址：上海市浦东新区世纪大道 1568 号中建大厦 32 楼，邮政编码：200122

PACKAGE MATERIALS INFORMATION

www.ti.com

5-Feb-2020

TAPE AND REEL INFORMATION

*All dimensions are nominal

Device

Package Package Pins

Type Drawing

SPQ

Reel

A0

B0

K0

P1

W

Pin1

Diameter Width (mm) (mm) (mm) (mm) (mm) Quadrant

(mm) W1 (mm)

LMK05318RGZR

LMK05318RGZT

VQFN

RGZ

48

2500

250

330.0

180.0

16.4

7.3

1.1

12.0

16.0

Q2

Pack Materials-Page 1

PACKAGE MATERIALS INFORMATION

www.ti.com

5-Feb-2020

*All dimensions are nominal

Device

Package Type Package Drawing Pins

SPQ

Length (mm) Width (mm) Height (mm)

LMK05318RGZR

LMK05318RGZT

VQFN

RGZ

48

2500

250

367.0

210.0

367.0

185.0

38.0

35.0

Pack Materials-Page 2

重要声明和免责声明

TI 均以“原样”提供技术性及可靠性数据（包括数据表）、设计资源（包括参考设计）、应用或其他设计建议、网络工具、安全信息和其他资

源，不保证其中不含任何瑕疵，且不做任何明示或暗示的担保，包括但不限于对适销性、适合某特定用途或不侵犯任何第三方知识产权的暗示

担保。

所述资源可供专业开发人员应用TI 产品进行设计使用。您将对以下行为独自承担全部责任：(1) 针对您的应用选择合适的TI 产品；(2) 设计、

验证并测试您的应用；(3) 确保您的应用满足相应标准以及任何其他安全、安保或其他要求。所述资源如有变更，恕不另行通知。TI 对您使用

所述资源的授权仅限于开发资源所涉及TI 产品的相关应用。除此之外不得复制或展示所述资源，也不提供其它TI或任何第三方的知识产权授权

许可。如因使用所述资源而产生任何索赔、赔偿、成本、损失及债务等，TI对此概不负责，并且您须赔偿由此对TI 及其代表造成的损害。

TI 所提供产品均受TI 的销售条款 (http://www.ti.com.cn/zh-cn/legal/termsofsale.html) 以及ti.com.cn上或随附TI产品提供的其他可适用条款的约

束。TI提供所述资源并不扩展或以其他方式更改TI 针对TI 产品所发布的可适用的担保范围或担保免责声明。IMPORTANT NOTICE

邮寄地址：上海市浦东新区世纪大道 1568 号中建大厦 32 楼，邮政编码：200122


型号：	LMK05318RGZR
厂家：	TEXAS INSTRUMENTS
描述：	采用 BAW 技术的超低抖动、单通道网络同步器时钟 \| RGZ \| 48 \| -40 to 85 时钟
文件：	总88页 (文件大小：2929K)
中文：	中文翻译
下载：	下载PDF数据表文档文件

LMK05318RGZR [TI]

相关型号：

SI9130DB

SI9135LG-T1

SI9135LG-T1-E3

SI9135_11

SI9136_11

SI9130CG-T1-E3

SI9130LG-T1-E3

SI9130_11

SI9137

SI9137DB

SI9137LG

SI9122E