AN-1026 [ADI]

High Speed Differential ADC Driver Design Considerations; 高速差分ADC驱动器设计考虑
AN-1026
型号: AN-1026
厂家: ADI    ADI
描述:

High Speed Differential ADC Driver Design Considerations
高速差分ADC驱动器设计考虑

驱动器
文件: 总16页 (文件大小:361K)
中文:  中文翻译
下载:  下载PDF数据表文档文件
AN-1026  
APPLICATION NOTE  
One Technology Way • P. O. Box 9106 • Norwood, MA 02062-9106, U.S.A. • Tel: 781.329.4700 • Fax: 781.461.3113 • www.analog.com  
High Speed Differential ADC Driver Design Considerations  
by John Ardizzoni and Jonathan Pearson  
For the discussions that follow, some definitions are in order.  
If the input signal is balanced, VIP and VIN are nominally equal  
in amplitude and opposite in phase with respect to a common  
reference voltage. When the input is single-ended, one input is  
at a fixed voltage, and the other varies with respect to it. In either  
case, the input signal is defined as VIP – VIN.  
INTRODUCTION  
Most modern high performance ADCs use differential inputs to  
reject common-mode noise and interference, increase dynamic  
range by a factor of 2, and improve overall performance due to  
balanced signaling. Though ADCs with differential inputs can  
accept single-ended input signals, optimum ADC performance  
is achieved when the input signal is differential. ADC drivers—  
circuits often specifically designed to provide such signals—perform  
many important functions, including amplitude scaling, single-  
ended-to-differential conversion, buffering, common-mode offset  
adjustment, and filtering. Since the introduction of the AD8138,  
differential ADC drivers have become essential signal conditioning  
elements in data acquisition systems.  
The differential-mode input voltage, VIN, dm, and the common-  
mode input voltage, VIN, cm, are defined in Equation 1 and  
Equation 2.  
V
IN, dm = VIP VIN  
(1)  
VIP +VIN  
VIN, cm  
=
(2)  
2
This common-mode definition is intuitive when applied to  
balanced inputs, but it is also valid for single-ended inputs.  
R
F1  
R
G1  
V
V
A
+
V
IP  
V
ON  
The output also has a differential mode and a common mode,  
defined in Equation 3 and Equation 4.  
V
V
IN, dm  
OCM  
OUT, dm  
V
V
+
IN  
OP  
V
R
A
G2  
R
F2  
VOUT, dm = VOP VON  
VOP +VON  
(3)  
Figure 1. Differential Amplifier  
VOUT, cm  
=
(4)  
A basic fully differential voltage-feedback ADC driver is shown  
in Figure 1. Two differences from a traditional op-amp feedback  
circuit can be seen. The differential ADC driver has an additional  
output terminal (VON) and an additional input terminal (VOCM).  
These terminals provide great flexibility when interfacing  
signals to ADCs that have differential inputs.  
2
Note the difference between the actual output common-mode  
voltage, VOUT, cm, and the VOCM input terminal, which establishes  
the output common-mode level.  
The analysis of differential ADC drivers is considerably more  
complex than that of traditional op amps. To simplify the algebra, it  
is expedient to define two feedback factors, β1 and β2, as given  
in Equation 5 and Equation 6.  
Instead of a single-ended output, the differential ADC driver  
produces a balanced differential output, with respect to VOCM  
,
between VOP and VON. (P indicates positive and N indicates  
negative.) The VOCM input controls the output common-mode  
voltage. As long as the inputs and outputs stay within their  
specified limits, the output common-mode voltage must equal  
the voltage applied to the VOCM input. Negative feedback and  
high open-loop gain cause the voltages at the amplifier input  
terminals, VA+ and VA–, to be essentially equal.  
RG1  
F1 + RG1  
β1 =  
β2 =  
(5)  
(6)  
R
RG2  
F2 + RG2  
R
Rev. 0  
Information furnished by Analog Devices is believed to be accurate and reliable. However, no  
responsibility is assumed by Analog Devices for its use, nor for any infringements of patents or other  
rights of third parties that may result from its use. Specifications subject to change without notice. No  
license is granted by implication or otherwise under any patent or patent rights of Analog Devices.  
Trademarks and registeredtrademarks arethe property of their respective owners.  
One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A.  
Tel: 781.329.4700  
Fax: 781.461.3113  
www.analog.com  
©2009 Analog Devices, Inc. All rights reserved.  
 
 
AN-1026  
APPLICATION NOTE  
TABLE OF CONTENTS  
Introduction ...................................................................................... 1  
Noise ...............................................................................................7  
Supply Voltage ...............................................................................9  
Harmonic Distortion ................................................................. 10  
Bandwidth and Slew Rate.......................................................... 11  
Stability ........................................................................................ 11  
PC Board Layout ........................................................................ 12  
Revision History ............................................................................... 2  
Terminating the Input to an ADC Driver................................. 3  
Input Common-Mode Voltage Range (ICMVR)......................... 5  
Input and Output Coupling: AC or DC .................................... 6  
Output Swing ................................................................................ 7  
REVISION HISTORY  
11/09—Revision 0: Initial Version  
Rev. 0 | Page 2 of 16  
 
APPLICATION NOTE  
AN-1026  
In most ADC driving applications, β1 = β2, but the general closed-  
loop equation for VOUT, dm, in terms of VIP, VIN, VOCM, β1, and β2, is  
useful to gain insight into how beta mismatch affects performance.  
The equation for VOUT, dm, shown in Equation 7, includes the finite  
frequency-dependent, open-loop voltage gain of the amplifier, A(s).  
TERMINATING THE INPUT TO AN ADC DRIVER  
ADC drivers are frequently used in systems that process high  
speed signals. Devices separated by more than a small fraction  
of a signal wavelength must be connected by electrical transmission  
lines with controlled impedance to avoid losing signal integrity.  
Optimum performance is achieved when a transmission line is  
terminated at both ends in its characteristic impedance. The driver  
is generally placed close to the ADC, so controlled-impedance  
connections are not required between them; however, the incoming  
signal connection to the ADC driver input is often long enough  
to require a controlled-impedance connection, terminated in  
the proper resistance.  
(
β1 β2  
)
1+  
+VIP  
(
1β1  
)
VIN  
(
1β2  
)
2
VOCM  
VOUT,dm  
=
(7)  
2
β1 + β2  
A
(
s
)
(
β1 + β2  
)
When β1 ≠ β2, the differential output voltage depends on VOCM  
,
which is an undesirable outcome because it produces an offset and  
excess noise in the differential output. The gain bandwidth product  
of the voltage-feedback architecture is constant. Interestingly,  
the gain in the gain bandwidth product is the reciprocal of the  
averages of the two feedback factors.  
The input resistance of the ADC driver, whether differential or  
single-ended, must be greater than or equal to the desired  
termination resistance so that a termination resistor, RT, can be  
added in parallel with the amplifier input to achieve the required  
resistance. All ADC drivers in the examples considered here are  
designed to have balanced feedback ratios, as shown in Figure 2.  
When β1 = β2 β, Equation 7 reduces to Equation 8.  
VOUT, dm  
R
RF  
RG  
1
F
=
(8)  
1
R
G
VIN, dm  
1 +  
A(s)  
(β  
)
R
V
OCM  
IN, dm  
This is a more familiar-looking expression; the ideal closed-loop  
R
G
R
gain becomes simply RF/RG when A(s) → . The gain bandwidth  
product is also more familiar looking, with the noise gain equal  
to 1/β, just as with a traditional op amp.  
F
Figure 2. Differential Amplifier Input Impedance  
Because the voltage between the two amplifier inputs is driven  
to a null by negative feedback, they are virtually connected, and  
the differential input resistance, RIN, is simply 2 × RG. To match  
the transmission line resistance, RL, place resistor, RT, as calculated  
in Equation 11, across the differential input. Figure 3 shows  
typical resistances RF = RG = 200 Ω, desired RL, dm = 100 Ω, and  
RT = 133 Ω.  
The ideal closed-loop gain for a differential ADC driver with  
matched feedback factors is seen in Equation 9.  
VOUT, dm  
RF  
AV =  
=
(9)  
VIN, dm  
RG  
Output balance, an important performance metric for differential  
ADC drivers, has two components: amplitude balance and phase  
balance. Amplitude balance is a measure of how closely the two  
outputs are matched in amplitude; in an ideal amplifier, they are  
exactly matched. Output phase balance is a measure of how close  
the phase difference between the two outputs is to 180°. Any  
imbalance in output amplitude or phase produces an undesirable  
common-mode component in the output. The output balance  
error (Equation 10) is the log ratio of the output common-mode  
voltage produced by a differential input signal to the output  
differential-mode voltage produced by the same input signal,  
expressed in decibels.  
1
RT =  
(11)  
1
1
RL RIN  
200  
200Ω  
200Ω  
R
= 100Ω  
V
R
L, dm  
OCM  
T
200Ω  
1
R
=
= 133Ω  
T
1
1
100Ω  
400Ω  
ΔVOUT, cm  
Figure 3. Matching a 100 Ω Line  
Output Balance Error = 20log10  
(10)  
ΔV  
OUT, dm  
An internal common-mode feedback loop forces VOUT, cm to  
equal the voltage applied to the VOCM input, producing excellent  
output balance.  
Rev. 0 | Page 3 of 16  
 
 
 
AN-1026  
APPLICATION NOTE  
Terminating a single-ended input requires significantly more  
effort. Figure 4 illustrates how an ADC driver operates with a  
single-ended input and a differential output.  
3.5V  
In Figure 5, a single-ended-to-differential gain of 1, a 50 Ω  
input termination, and feedback and gain resistors with values  
in the neighborhood of 200 Ω are required to keep noise low.  
267  
200Ω  
200Ω  
V
ON  
2.5V  
V
+
V
2.5V  
OCM  
OUT, dm  
1.5V  
R
F
V
OP  
500  
200Ω  
2V  
200Ω  
R
G
500Ω  
V
Figure 5. Single-Ended Input Impedance  
ON  
V
V
+
IN  
V
2.5V  
OCM  
0V  
OUT, dm  
Equation 12 provides the single-ended input resistance, 267 Ω.  
Equation 13 indicates that the parallel resistance, RT, should be  
61.5 Ω to bring the 267 Ω input resistance down to 50 Ω.  
V
OP  
R
G
500Ω  
–2V  
R
F
5003.5V  
1.75V  
1
RT =  
= 61.5  
(13)  
1.25V  
0.75V  
2.5V  
1
1
50ꢀ 267ꢀ  
1.5V  
Figure 4. Example of Single-Ended Input to ADC Driver  
Figure 6 shows the circuit with source and termination resistances.  
The open-circuit voltage of the source, with its 50 Ω source  
resistance, is 2 V p-p. When the source is terminated in 50 Ω,  
the input voltage is reduced to 1 V p-p, which is also the  
differential output voltage of the unity-gain driver.  
200  
Although the input is single-ended, VIN, dm is equal to VIN.  
Because resistors RF and RG are equal and balanced, the gain  
is unity, and the differential output, VOP − VON, is equal to the  
input, that is, 4 V p-p. VOUT, cm is equal to VOCM = 2.5 V and,  
from the lower feedback circuit, input voltages VA+ and VA−  
are equal to VOP/2.  
50Ω  
200Ω  
2.5V  
V
ON  
V
+
R
Using Equation 3 and Equation 4, VOP = VOCM + VIN/2, an in-phase  
swing of 1 V about 2.5 V. VON = VOCM – VIN/2, an antiphase  
swing of 1 V about 2.5 V. Thus, VA+ and VA− swing 0.5 V  
about 1.25 V. The ac component of the current that must be  
supplied by VIN is (2 V – 0.5 V)/500 Ω = 3 mA, so the resistance  
to ground that must be matched, looking in from VIN, is 667 Ω.  
T
V
OCM  
2V p-p  
OUT, dm  
61.5Ω  
V
OP  
200Ω  
200Ω  
Figure 6. Single-Ended Circuit with Source and Termination Resistances  
This circuit may initially appear to be complete, but an unmatched  
resistance of 61.5 Ω in parallel with 50 Ω has been added to the  
upper RG alone. This changes the gain and single-ended input  
resistance and mismatches the feedback factors. For small gains, the  
change in input resistance is small and neglected for the moment,  
but the feedback factors must still be matched. The simplest way  
to accomplish this is to add resistance to the lower RG. Figure 7  
shows a Thevenin equivalent circuit in which the previously  
mentioned parallel combination acts as the source resistance.  
The general formula for determining this single-ended input  
resistance when the feedback factors of each loop are matched  
is shown in Equation 12, where RIN, se is the single-ended input  
resistance.  
RG  
RF  
RG + RF  
RIN, se  
=
(12)  
1−  
2×  
(
)
R
TH  
27.6  
This is a starting point for calculating the termination resistance.  
However, it is important to note that amplifier gain equations  
are based on the assumption of a zero impedance input source. A  
significant source impedance that must be matched in the presence  
of an imbalance caused by a single-ended input inherently adds  
resistance only to the upper RG. To retain the balance, this must  
be matched by adding resistance to the lower RG, but this affects  
the gain.  
V
TH  
1.1V p-p  
Figure 7. Thevenin Equivalent of Input Source  
With this substitution, a 27.6 Ω resistor, RTS, is added to the  
lower loop to match loop feedback factors, as seen in Figure 8.  
R
200  
F
R
R
200Ω  
TH  
27.6Ω  
G
Although it may be possible to determine a closed-form solution to  
the problem of terminating a single-ended signal, an iterative  
method is generally used. The need for it will become apparent  
in the following example.  
V
ON  
V
+
V
TH  
1.1V p-p  
V
2.5V  
OCM  
OUT, dm  
V
OP  
R
R
G
200Ω  
TS  
27.6Ω  
R
F
200Ω  
Figure 8. Balanced Single-Ended Termination Circuit  
Rev. 0 | Page 4 of 16  
 
 
 
 
 
APPLICATION NOTE  
AN-1026  
Note that the Thevenin voltage of 1.1 V p-p is larger than the  
properly terminated voltage of 1 V p-p, and the gain resistors  
are each increased by 27.6 Ω, decreasing the closed-loop gain.  
These opposing effects tend to cancel each other out for large  
resistors (>1 kΩ) and small gains (1 or 2), but are not entirely  
canceled out for small resistors or higher gains.  
A single iteration of the method described here works well for  
closed-loop gains of 1 or 2. For higher gains, the value of RTS  
gets closer to the value of RG, and the difference between the  
value of RIN, se calculated in Equation 18 and that calculated in  
Equation 12 becomes greater. Several iterations are required for  
these cases.  
The circuit in Figure 8 is now easily analyzed, and the differential  
output voltage is calculated in Equation 14.  
This should not be arduous: the recently released differential  
amplifier calculator tools, ADIsimDiffAmp™ and ADI Diff Amp  
Calculator™, do all the heavy lifting; they perform the above  
calculations in a matter of seconds. See www.analog.com for  
more information.  
200ꢀ  
227.6ꢀ  
VOUT,dm =1.1V p - p  
= 0.97 V p - p  
(14)  
The differential output voltage is not quite at the desired level of  
1 V p-p, but a final independent gain adjustment is available by  
modifying the feedback resistance as shown in Equation 15.  
INPUT COMMON-MODE VOLTAGE RANGE (ICMVR)  
ICMVR specifies the range of voltage that can be applied to the  
differential amplifier inputs for normal operation. The voltage  
appearing at those inputs can be referred to as ICMV, Vacm, or VA .  
This specification is often misunderstood. The most frequent  
difficulty is determining the actual voltage at the differential  
amplifier inputs, especially with respect to the input voltage.  
The amplifier input voltage (VA ) can be calculated when the  
variables VIN, cm, β, and VOCM are known, using the general  
Equation 19 for unequal βs or the simplified Equation 20 for  
equal βs.  
Desired VOUT,dm  
RF = 227.6 ꢀ  
=
1.1 V p- p  
(15)  
1.0 V p- p  
1.1 V p- p  
227.6ꢀ  
= 206.9 ꢀ  
Figure 9 shows the completed circuit, implemented with  
standard 1% resistor values.  
R
205Ω  
F
R
IN, se  
2β1β2VOCM + V β2  
(
1β1  
β1 + β2  
VOCM VICM  
)
+VIN β1(1β2 )  
IP  
Vacm orVA  
=
(19)  
(20)  
R
R
±
S
G
50Ω  
200Ω  
V
ON  
V
+
R
T
Vacm or V =VIN cm + β  
(
)
V
2.5V  
OCM  
A
,
2V p-p  
OUT, dm  
±
61.9Ω  
V
OP  
Note that VA is always a scaled-down version of the input signal  
(as shown in Figure 4). The input common-mode voltage range  
differs among amplifier types. Analog Devices, Inc., high speed  
differential ADC drivers have two configurations of input stages,  
centered and shifted. The centered ADC drivers have about 1 V of  
headroom from each supply rail (hence centered). The shifted input  
stages add two transistors to allow the inputs to swing closer to  
the –VS rail. Figure 10 shows a simplified input schematic of a  
typical differential amplifier (Q2 and Q3).  
R
G
R
200Ω  
TS  
28Ω  
R
F
205Ω  
Figure 9. Complete Single-Ended Termination Circuit  
Referring to Figure 9, the single-ended input resistance of the  
driver, RIN, se, has changed due to changes in RF and RG. The  
gain resistances of the driver are 200 Ω in the upper loop and  
200 Ω + 28 Ω = 228 Ω in the lower loop. Calculation of RIN, se  
with differing gain resistance values first requires two values of  
beta to be calculated, as shown in Equation 16 and Equation 17.  
A
Q2  
Q3  
–IN  
+IN  
Q1  
Q4  
200ꢀ  
RF + RG 405ꢀ  
RG  
β1 =  
β2 =  
=
= 0.494  
(16)  
(17)  
RG + RTS  
RF + RG + RTS 433ꢀ  
228ꢀ  
Figure 10. Simplified Differential Amplifier with Shifted ICMVR  
=
= 0.527  
The shifted input architecture allows the differential amplifier to  
process a bipolar input signal, even when the amplifier is powered  
from a single supply, making it well suited for single-supply  
applications with inputs at or below ground. The additional PNP  
transistors (Q1 and Q4) at the input shift the input to the differential  
pair up by one transistor Vbe. For example, with –0.3 V applied  
at –IN, Point A is 0.7 V, allowing the differential pair to operate  
properly. Without the PNPs (centered input stage), –0.3 V at  
Point A would reverse-bias the NPN differential pair and halt  
normal operation.  
The input resistance, RIN, se, is calculated as shown in Equation 18.  
RG  
β1  
(
β1 + β2  
)
= 271ꢀ  
RIN,se  
=
(18)  
(
β2 + 1  
)
This differs little from the original calculated value of 267 Ω  
and does not have a significant effect on the calculation of RT,  
because RIN, se is in parallel with RT.  
If a more exact overall gain is necessary, higher precision or  
series trim resistors can be used.  
Rev. 0 | Page 5 of 16  
 
 
 
AN-1026  
APPLICATION NOTE  
Table 1. High Speed ADC Driver Specifications  
Supply Voltage (V)  
ADC Driver  
Slew  
ICMVR  
VOCM  
Output  
Swing  
from  
BW  
Rate  
Noise  
ISUPPLY  
(mA)  
Part No.  
AD8132  
AD8137  
(MHz) (V/μs) (nV)  
5 V  
+5 V  
+3.3 V  
+3 V  
5 V +5 V  
+3.3 V  
+3 V  
Rails (V)  
350  
76  
1200  
450  
8
–4.7 to +3  
–4 to +4  
0.3 to 3  
1 to 4  
0.3 to 1.3  
1 to 2.3  
0.3 to 1  
1 to 2  
3.6  
4
1 to 3.7  
1 to 4  
N/A  
0.3 to 1  
1 to 2  
1
12  
8.25  
1 to 2.3  
Rail to  
rail  
3.2  
AD8138  
AD8139  
320  
410  
1150  
800  
5
–4.7 to +3.4 0.3 to 3.2 N/A  
–4 to +4 1 to 4 N/A  
N/A  
N/A  
3.8  
3.8  
1 to 3.8  
1 to 3.8  
N/A  
N/A  
N/A  
N/A  
1.4  
20  
25  
2.25  
Rail to  
rail  
ADA4927-1/ 2300  
ADA4927-2  
5000  
2800  
6000  
4700  
6800  
1.4  
3.6  
2.2  
2.6  
2.3  
–3.5 to +3.5 1.3 to 3.7 N/A  
–4.8 to +3.2 0.2 to 3.2 N/A  
N/A  
N/A  
3.5  
3.8  
1.5 to 3.5 N/A  
1.2 to 3.2 N/A  
N/A  
N/A  
1.2  
20  
9
ADA4932-1/ 1000  
ADA4932-2  
1
ADA4937-1/ 1900  
ADA4937-2  
N/A  
–4.7 to +3.4 0.3 to 3.4 N/A  
N/A 1.1 to 3.9 0.9 to 2.4 N/A  
0.3 to 3  
0.3 to 1.2 N/A  
N/A  
3.7  
1.2 to 3.8 1.2 to 2.1 N/A  
1.3 to 3.7 N/A N/A  
1.3 to 3.5 1.3 to 1.9 N/A  
0.9  
1.2  
0.8  
40  
37  
37  
ADA4938-1/ 1000  
ADA4938-2  
N/A  
ADA4939-1/ 1400  
ADA4939-2  
N/A  
AC-coupled, single-ended-to-differential applications are similar to  
their differential-input counterparts but have common-mode  
ripple, a scaled-down replica of the input signal, at the amplifier  
input terminals. An ADC driver with a centered input common-  
mode range places the average input common-mode voltage  
near the middle of its specified range, providing plenty of  
margin for the ripple in most applications.  
Table 1 provides a quick reference to many specifications of  
Analog Devices ADC drivers, including which drivers feature a  
shifted ICMVR and which do not.  
INPUT AND OUTPUT COUPLING: AC OR DC  
The need for ac or dc coupling can have a significant impact on  
the choice of a differential ADC driver. The considerations  
differ between input and output coupling.  
When input coupling is optional, it is worth noting that ADC  
drivers with ac-coupled inputs dissipate less power than similar  
drivers with dc-coupled inputs because no dc common-mode  
current flows in either feedback loop.  
An ac-coupled input stage is illustrated in Figure 11.  
R
F
C
IN  
R
G
AC coupling the ADC driver outputs is useful when the ADC  
requires an input common-mode voltage that differs substantially  
from that available at the output of the driver. The drivers have  
maximum output swing when VOCM is set near midsupply; this  
presents a problem when driving low voltage ADCs with very  
low input common-mode voltage requirements. A simple solution  
to this predicament is to ac-couple the connection between the  
driver output and the ADC input (see Figure 12), removing the  
dc common-mode voltage of the ADC from the driver output  
and allowing a common-mode level suitable for the ADC to be  
applied on its side of the ac coupling. For example, the driver  
could be running on a single 5 V supply with VOCM = 2.5 V and  
the ADC could be running on a single 1.8 V supply with a required  
input common-mode voltage of 0.9 V applied at ADC CMV.  
V
OCM  
R
C
G
IN  
R
F
Figure 11. AC-Coupled ADC Driver  
For differential-to-differential applications with ac-coupled inputs,  
the dc common-mode voltage appearing at the amplifier input  
terminals is equal to the dc output common-mode voltage  
because dc feedback current is blocked by the input capacitors.  
Also, the feedback factors at dc are matched and exactly equal to  
unity. VOCM, and consequently the dc input common mode, is very  
often set near midsupply. An ADC driver with a centered input  
common-mode range works well in these types of applications,  
with the input common-mode voltage near the center of its  
specified range.  
R
F
C
OUT  
R
G
R
R
OUT  
OUT  
ADC  
CMV  
TO  
ADC  
V
OCM  
R
G
C
OUT  
R
F
Figure 12. DC-Coupled Inputs with AC-Coupled Outputs  
Rev. 0 | Page 6 of 16  
 
 
 
 
APPLICATION NOTE  
AN-1026  
–20  
–30  
Drivers with shifted input common-mode ranges generally  
work best in dc-coupled systems operating on single supplies.  
This is because the output common-mode voltage is divided  
down through the feedback loops, and its variable components  
can get close to ground, which is the negative rail. With single-  
ended inputs, the input common-mode voltage gets even closer  
to the negative rail due to the input-related ripple.  
V
= 2V p-p  
OUT  
–40  
HD2 @ 10MHz  
HD3 @ 10MHz  
HD2 @ 30MHz  
HD3 @ 30MHz  
–50  
–60  
–70  
–80  
Systems running on dual supplies, with single-ended or differential  
inputs and ac coupling or dc coupling, usually work well with either  
type of input stage because of the increased headroom.  
–90  
–100  
–110  
–120  
Table 2 summarizes the most common ADC driver input  
stage types used with various input coupling and power supply  
combinations. However, these choices may not always be the  
best; each system should be analyzed on a case-by-case basis.  
1.4  
1.6  
1.8  
2.0  
2.2  
2.4  
2.6  
2.8  
3.0  
3.2  
3.4  
V
(V)  
OCM  
Figure 13. Harmonic Distortion vs. VOCM at Various Frequencies for the  
ADA4932 with a 5 V Supply  
Table 2. Coupling and Input Stage Options  
Figure 13 shows harmonic distortion vs. VOCM at various  
frequencies for the ADA4932, which is specified with  
a typical output swing to within 1.2 V of each rail (headroom).  
The output swing is the sum of VOCM and VPEAK of the signal (1 V).  
Note that the distortion starts to accelerate above 2.8 V  
(3.8 VPEAK, or 1.2 V below the 5 V rail). At the low end,  
distortion is still low at 2.2 V (–1 VPEAK). The same type of  
behavior appears in the discussions of bandwidth and slew rate.  
Input Coupling Input Signal Power Supplies Input Type  
Any  
AC  
DC  
AC  
DC  
Any  
Dual  
Either  
Centered  
Shifted  
Centered  
Centered  
Single-ended Single  
Single-ended Single  
Differential  
Differential  
Single  
Single  
OUTPUT SWING  
NOISE  
To maximize the dynamic range of an ADC, it should be driven  
to its full input range. However, care is needed when driving the  
ADC. If the ADC is driven too hard, the input may be damaged;  
if it is not driven hard enough, resolution is lost. Driving the  
ADC to its full input range does not mean that the amplifier  
output must swing to its full range. A major benefit of  
ADC imperfections include quantization noise, electronic—or  
random—noise, and harmonic distortion. Important in most  
applications, noise is usually the most important performance  
metric in broadband systems.  
All ADCs inherently have quantization noise, which depends  
on the number of bits (n); quantization noise can be decreased  
by increasing the number of bits (n). Because even ideal  
converters produce quantization noise, it is used as a benchmark  
against which to compare random noise and harmonic distortion.  
The output noise from the ADC driver should be comparable to  
or lower than the random noise and distortion of the ADC.  
Beginning with a review of the characterization of ADC noise  
and distortion, how to weigh ADC driver noise against the  
performance of the ADC will be shown.  
differential outputs is that each output must swing only half as  
much as a traditional single-ended output. The driver outputs  
can stay away from the supply rails, allowing decreased  
distortion. However, this is not the case for single-ended drivers.  
As the output voltage of the driver approaches the rail, the  
amplifier loses linearity and introduces distortion.  
For applications where every millivolt of output voltage is  
required, see Table 1 for ADC drivers that have rail-to-rail  
outputs with typical headroom ranging from a few millivolts to a  
few hundred millivolts, depending on the load.  
Quantization noise occurs because the ADC quantizes analog  
signals that have infinite resolution into a finite number of discrete  
levels. An n-bit ADC has 2n binary levels. The difference between  
one level and the next represents the finest difference that can  
be resolved; it is referred to as a least significant bit (LSB), or q,  
for the quantum level. One quantum level is therefore 1/2n of  
the range of the converter. If a varying voltage is converted by a  
perfect n-bit ADC, then converted back to analog and subtracted  
from the input of the ADC, the difference looks like noise. It  
has an rms value of  
q
1
RMS Quantization Noise =  
=
(21)  
12 2n 12  
Rev. 0 | Page 7 of 16  
 
 
 
AN-1026  
APPLICATION NOTE  
From this, the logarithmic (dB) formula for the signal-to-  
quantizing-noise ratio of an n-bit ADC over its Nyquist  
bandwidth can be derived (Equation 22); it is the best  
achievable SNR for an n-bit converter.  
1
SINAD(dB) = 20log10  
(24)  
THD + N  
If SINAD is substituted for the signal-to-quantizing-noise ratio  
(Equation 22), an effective number of bits (ENOB) that a converter  
can have if its signal-to-quantizing-noise ratio is the same as its  
SINAD (Equation 25) can be defined.  
Signal-to-Quantization-Noise Ratio (dB) = 6.02n + 1.76 dB (22)  
Random noise in ADCs, a combination of thermal, shot, and  
flicker noise, is generally larger than the quantization noise.  
Harmonic distortion, resulting from nonlinearities in the ADC,  
produces unwanted signals in the output that are harmonically  
related to the input signals. Total harmonic distortion and noise  
(THD + N) is an important ADC performance metric that  
compares the electronic noise and harmonic distortion with an  
analog input that is close to the full-scale input range of the ADC.  
Electronic noise is integrated over a bandwidth that includes the  
frequency of the last harmonic to be considered. In Equation 23,  
the total THD includes the first five harmonic distortion  
SINAD(dB) = 6.02(ENOB) + 1.76 dB  
(25)  
ENOB can also be expressed in terms of SINAD as shown in  
Equation 26.  
SINAD(dB) 1.76 dB  
ENOB =  
(26)  
6.02  
ENOB can be used to compare noise performance of an ADC  
driver with that of the ADC to determine its suitability to drive  
that ADC. A differential ADC noise model is shown in Figure 14.  
V
V
components, which are root-sum-squared along with the noise.  
nRG1  
nRF1  
R
R
F1  
G1  
THD + Noise =  
inIN+  
[
v2(rms) 2  
]
+
[
v3(rms) 2  
]
+
[
v4(rms) 2  
v1(rms) 2  
]
+
[
v5(rms) 2  
]
+
[
v6(rms) 2  
+ vn  
]
2
+
V
nIN  
V
[
]
no, dm  
inIN–  
(23)  
V
OCM  
The input signal is v1, the first five harmonic distortion products  
are v2 through v6, and the ADC electronic noise is vn.  
V
nCM  
R
R
F2  
G2  
V
nRG2  
V
nRF2  
The reciprocal of THD + noise, the signal-to-noise-and-distortion  
ratio (SINAD) is usually expressed in decibels (Equation 24).  
Figure 14. Noise Model of Differential ADC Driver  
The contributions to the total output noise density of each of  
the eight sources are shown in Equation 27 for the general case  
and when β1 = β2 β.  
2vnIN  
β1 + β2  
vnIN  
β
vno, dm due to vnIN  
vno, dm due to vnCM  
=
=
for β1 = β2 = β  
2vnCM  
(
β1 β2  
)
=
= 0 for β1 = β2 = β  
β1 + β2  
2inIN +  
(
1 β1  
)
RG  
1
vno, dm due to inIN +  
=
=
(
inIN +  
)
(
RF1 for β1 = β2 = β  
)
β1 + β2  
2inIN −  
(
1 β2  
)
RG  
2
vno, dm due to inIN −  
vno, dm due to vnRG1  
=
=
(
inIN −  
)
(
RF2  
)
for β1 = β2 = β  
β1 + β2  
(
2 4kTRG1  
)
(
1 β1  
)
RF1  
RG1  
=
=
4kTRG1  
4kTRG2  
for β1 = β2 = β  
β1 + β2  
(
2 4kTRG2  
)
(
1 β2  
)
RF2  
RG2  
vno, dm due to vnRG2  
vno, dm due to vnRF1  
vno, dm due to vnRF2  
=
=
for β1 = β2 = β  
β1 + β2  
2β1 4kTRF1  
β1 + β2  
=
=
4kTRF1 for β1 = β2 = β  
4kTRF2 for β1 = β2 = β  
2β1 4kTRF2  
β1 + β2  
=
=
(27)  
Rev. 0 | Page 8 of 16  
 
APPLICATION NOTE  
AN-1026  
The total output noise voltage density, vno, dm, is calculated by  
computing the root-sum square of these components. Entering  
the equations into a spreadsheet is the best way to calculate the  
total output noise voltage density. The ADI Diff Amp Calculator,  
which quickly calculates noise, gain, and other differential ADC  
driver behaviors, is also available at www.analog.com.  
SUPPLY VOLTAGE  
Considering supply voltage and current is a quick way to narrow  
the choice of ADC drivers. Table 1 provides a compact reference  
to ADC driver performance with respect to power supply. The  
supply voltage influences bandwidth, signal swing, and ICMVR.  
Weighing the specifications and reviewing the trade-offs are  
important to differential amplifier selection.  
ADC driver noise performance can now be compared with the  
ENOB of an ADC. Take for example, a differential driver with a gain  
of 2 for the AD9445 ADC on a 5 V supply with a 2 V full-scale input;  
it processes a direct-coupled broadband signal occupying a 50 MHz  
(−3 dB) bandwidth, limited with a single-pole filter. From the data  
sheet listing of the ENOB specifications for various conditions,  
ENOB = 12 bits for a Nyquist bandwidth of 50 MHz.  
Power-supply rejection (PSR) is another important specification.  
The role of power supply pins as inputs to the amplifier is often  
ignored. Any noise on the power supply lines or coupled into  
them can potentially corrupt the output signal.  
For example, consider the ADA4937-1 with 50 mV p-p at 60 MHz  
of noise on the power line. Its PSR at 50 MHz is −70 dB. This  
means that the noise on the power supply line reduces to  
approximately 16 μV at the amplifier output. In a 16-bit system  
with a 1 V full-scale input, 1 LSB is 15.3 μV; the noise from the  
power supply line therefore swamps the LSB.  
The ADA4939 is a high performance, broadband differential ADC  
driver that can be direct coupled. It is a good candidate to drive  
the AD9445 with respect to noise. The data sheet recommends  
RF = 402 Ω and RG = 200 Ω for a differential gain of approximately  
2. A total output voltage noise density for this configuration is  
9.7 nV√Hz.  
To improve this situation, add series SMT ferrite beads, L1 and  
L2, and shunt bypass capacitors, C1 and C2 (see Figure 15).  
First, calculate the system noise bandwidth, BN, which is the  
bandwidth of an equivalent rectangular low-pass filter that  
outputs the same noise power as the actual filter that determines  
the system bandwidth for a given constant input noise power  
spectral density. For a one-pole filter, BN is equal to π/2 times  
the 3 dB bandwidth, as shown in Equation 28.  
V+  
L1  
U1  
C1  
AD8138AR  
10nF  
3
8
2
1
5
–OUT  
π
2
V
OCM  
6
B =  
50 MHz = 78.5 MHz  
(28)  
N
4
+OUT  
Next, integrate the noise density over the square root of the system  
bandwidth to obtain the output rms noise (see Equation 29).  
C2  
10nF  
L2  
v
no, dm (rms) = (9.7 nV/√Hz)(√78.5 MHz)= 86 μV rms (29)  
The amplitude of the noise is presumed to have a Gaussian  
distribution; therefore, using the common 3σ limits for the  
peak-to-peak noise (noise voltage swings between these limits  
about 99.7% of the time), the peak-to-peak output noise is  
calculated as  
V–  
Figure 15. Power Supply Bypassing  
At 50 MHz, the ferrite bead has an impedance of 60 Ω and the  
10 nF (0.01 ꢁF) capacitor has an impedance of 0.32 Ω. The  
attenuator formed by these two elements provides 45.5 dB of  
attenuation (see Equation 32).  
v
no, dm (p-p) ≈ 6(86 μV rms) = 516 μV p-p  
(30)  
Now compare the peak-to-peak output noise of the driver with  
1 LSB voltage of the AD9445 LSB, based on an ENOB of 12 bits  
and full-scale input range of 2 V, as calculated in Equation 31.  
0.32  
0.32 + 60  
Divider Attenuation = 20log  
= −45.5 dB (32)  
The divider attenuation combines with the PSR of –70 dB to  
provide approximately 115 dB of rejection. This reduces the  
noise to approximately 90 nV p-p, well below 1 LSB.  
One LSB = 2 V/212 = 488 μV  
(31)  
The peak-to-peak output noise from the driver is comparable to  
the LSB of the ADC with respect to 12 bits of the ENOB; the  
driver is therefore a good choice to consider in this application  
from the standpoint of noise. The final determination must be  
made by building and testing the driver and ADC combination.  
Rev. 0 | Page 9 of 16  
 
 
AN-1026  
APPLICATION NOTE  
The HD2 at 50 MHz is approximately −88 dBc, relative to a  
2 V p-p input signal. To compare the harmonic distortion level  
to 1 ENOB LSB, this level must be converted to a voltage as shown  
in Equation 33.  
HARMONIC DISTORTION  
Low harmonic distortion in the frequency domain is important  
in both narrow-band and broadband systems. Nonlinearities in  
the drivers generate single-tone harmonic distortion and multitone,  
intermodulation distortion products at amplifier outputs.  
88  
20  
HD2 =  
(
2 V p - p  
)
10  
80 ꢁV p - p  
(33)  
The same approach used in the noise analysis example can be  
applied to distortion analysis, comparing the harmonic distortion  
of the ADA4939 with 1 LSB of the AD9445s ENOB of 12 bits  
with a 2 V full-scale output. One ENOB LSB is 488 μV in the  
noise analysis.  
This distortion product is only 80 μV p-p, or 16% of 1 ENOB  
LSB. Thus, from a distortion standpoint, the ADA4939 is a  
good choice to consider as a driver for the AD9445 ADC.  
Because ADC drivers are negative feedback amplifiers, output  
distortion depends on the amount of loop gain in the amplifier  
circuit. The inherent open-loop distortion of a negative feedback  
amplifier is reduced by a factor of 1/(1 + LG), where LG is the  
available loop gain.  
The distortion data in the specifications table of the ADA4939  
is given for a gain of 2, comparing second and third harmonics  
at various frequencies. Table 3 shows the harmonic distortion  
data for a gain of 2 and differential output swing of 2 V p-p.  
Table 3. Second and Third Harmonic Distortion of the ADA4939  
The input (error voltage) of the amplifier is multiplied by a large  
forward voltage gain, A(s), then passes through the feedback  
factor, β, to the input, where it adjusts the output to minimize the  
error. Therefore, the loop gain of this type of amplifier is A(s) × β;  
as the loop gain (A(s), β, or both) decreases, harmonic distortion  
increases. Voltage feedback amplifiers, such as integrators, are  
designed to have large A(s) at dc and low frequencies, and then  
roll off as 1/f toward unity at a specified high frequency. As A(s)  
rolls off, loop gain decreases and distortion increases. Therefore,  
the harmonic distortion characteristic is the inverse of A(s).  
Parameter  
Harmonic Distortion (dBc)  
HD2 @ 10 MHz  
HD2 @ 70 MHz  
HD2 @ 100 MHz  
HD3 @ 10 MHz  
HD3 @ 70 MHz  
HD3 @ 100 MHz  
−102  
−83  
−77  
−101  
−97  
−91  
The data shows that harmonic distortion increases with frequency  
and that HD2 is worse than HD3 in the bandwidth of interest  
(50 MHz). Harmonic distortion products are higher in frequency  
than the frequency of interest, so their amplitude can be reduced  
by system band-limiting. If the system had a brick-wall filter at  
50 MHz, then only the frequencies higher than 25 MHz are of  
concern because all harmonics of higher frequencies are eliminated  
by the filter. Nevertheless, the system was evaluated up to 50 MHz  
because any filtering that is present may not sufficiently suppress  
the harmonics, and distortion products can alias back into the  
signal bandwidth. Figure 16 shows the harmonic distortion vs.  
frequency of the ADA4939 for various supply voltages with a  
2 V p-p output.  
Current feedback amplifiers use an error current as the feedback  
signal. The error current is multiplied by a large forward  
transresistance, T(s), which converts it to the output voltage,  
then passes through the feedback factor, 1/RF, which converts  
the output voltage to a feedback current that tends to minimize  
the input error current. The loop gain of an ideal current feedback  
amplifier is therefore T(s) × (1/RF) = T(s)/RF. Like A(s), T(s) has  
a large dc value and rolls off with increasing frequency, reducing  
loop gain and increasing the harmonic distortion.  
Loop gain also depends directly upon the feedback factor, 1/RF.  
The loop gain of an ideal current feedback amplifier does not  
depend on a closed-loop voltage gain; therefore, harmonic  
distortion performance does not degrade as the closed-loop  
gain increases. In a real current feedback amplifier, loop gain  
does have some dependence on the closed-loop gain but not  
nearly to the extent that it does in a voltage feedback amplifier.  
This makes a current feedback amplifier, such as the ADA4927,  
a better choice than a voltage feedback amplifier for applications  
requiring high closed-loop gain and low distortion.  
–60  
V
= 2V p-p  
OUT, dm  
–65  
–70  
HD2, V (SPLIT SUPPLY) = ±2.5V  
S
HD3, V (SPLIT SUPPLY) = ±2.5V  
S
HD2, V (SPLIT SUPPLY) = ±1.65V  
S
HD3, V (SPLIT SUPPLY) = ±1.65V  
S
–75  
–80  
–85  
HD2 –88dBc @ 50MHz  
–90  
–95  
–100  
–105  
–110  
1
10  
100  
FREQUENCY (MHz)  
Figure 16. Harmonic Distortion vs. Frequency  
Rev. 0 | Page 10 of 16  
 
 
 
APPLICATION NOTE  
AN-1026  
Figure 17 shows how well distortion performance holds up as  
the closed-loop gain increases for the ADA4927.  
Slew rate (a large signal parameter) refers to the maximum rate of  
change the amplifier output can track the input, without excessive  
distortion. Consider the sine wave output at the slew rate.  
–40  
V
,
= 2V p-p  
OUT dm  
VO = VP sin 2πft  
(34)  
–50  
–60  
–70  
–80  
The derivative (rate of change) of Equation 34 at the zero crossing,  
the maximum rate, is  
dv  
dt  
max = 2πfVP  
(35)  
–90  
–100  
–110  
–120  
–130  
where:  
dv/dt max is the slew rate.  
VP is the peak voltage.  
f is the full power bandwidth (FPBW). Solving for FPBW,  
G = 1  
G = 10  
G = 20  
Slew Rate  
FPBW =  
1
10  
100  
1k  
(36)  
FREQUENCY (MHz)  
2πVp  
Figure 17. Distortion vs. Frequency and Gain  
Therefore, when selecting an ADC driver, it is important  
to consider the gain, bandwidth, and slew rate (FPBW) to  
determine if the amplifier is adequate for the application.  
BANDWIDTH AND SLEW RATE  
Bandwidth and slew rate are especially important in ADC  
driver applications. Typically, the small signal bandwidth is the  
bandwidth of a device, whereas the slew rate measures the maximum  
rate of change at the amplifier output for large signal swings.  
STABILITY  
Stability considerations for differential ADC drivers are the same as  
for op amps. The key specification is phase margin. The phase  
margin of a particular amplifier configuration can be determined  
from the data sheets; however, in a real system, parasitic effects in the  
PC board layout can reduce it significantly.  
Effective usable bandwidth (EUBW), an acronym analogous to  
ENOB, describes bandwidth. Many ADC drivers and op amps  
boast wide bandwidth specifications; however, not all that band-  
width is usable. For example, −3 dB bandwidth is a conventional  
way to measure bandwidth, but it does not mean that all the  
bandwidth is usable. The amplitude and phase errors of the  
−3 dB bandwidth can be seen a decade earlier than the actual  
break frequency. An excellent way to determine the usable  
bandwidth is to consult the distortion plots on the data sheet.  
The stability of a negative voltage, feedback amplifier depends on  
the magnitude and sign of its loop gain, A(s) × β. The differential  
ADC driver is a bit more complicated than a typical op amp circuit  
because it has two feedback factors. Loop gain is seen in the  
denominators of Equation 7 and Equation 8. Equation 37 describes  
the loop gain for the unmatched feedback factor case (β1 ≠ β2).  
Figure 18 shows that to maintain greater than −80 dBc for  
second and third harmonics, the ADC driver should not be used  
for frequencies greater than 60 MHz. Because each application  
is different, the system requirements are a guide to the appropriate  
driver with sufficient bandwidth and adequate distortion  
performance.  
A(s)  
(
β1 + β2  
2
)
Loop Gain =  
(37)  
With unmatched feedback factors, the effective feedback factor  
is simply the average of the two feedback factors. When they are  
matched and defined as β, the loop gain simplifies to A(s) × β.  
–50  
For a feedback amplifier to be stable, its loop gain must not be  
allowed to equal −1 or its equivalent, an amplitude of +1 with  
phase shift of −180°. For a voltage feedback amplifier, the point  
where the magnitude of loop gain equals 1 (that is, 0 dB) on its  
open-loop gain frequency plot is where the magnitude of A(s)  
equals the reciprocal of the feedback factor. For basic amplifier  
applications, the feedback is purely resistive, introducing no  
phase shifts around the feedback loop. With matched feedback  
factors, the frequency independent reciprocal of the feedback  
factor, 1 + RF/RG, is often referred to as the noise gain. If the  
constant noise gain in decibels is plotted on the same graph as  
the open-loop gain, A(s), the frequency where the two curves  
intersect is where the loop gain is 1, or 0 dB. The difference  
between the phase of A(s) at that frequency and −180° is defined as  
the phase margin; for stable operation, it should be greater than  
or equal to 45°.  
HD2, G = 1, R = 200Ω  
F
HD3, G = 1, R = 200Ω  
F
HD2, G = 2, R = 402Ω  
–60  
–70  
F
HD3, G = 2, R = 402Ω  
F
–80  
–90  
–100  
–110  
–120  
1
10  
100  
FREQUENCY (MHz)  
Figure 18. Distortion Curves for ADA4937 Current Feedback ADC Driver  
Rev. 0 | Page 11 of 16  
 
 
 
AN-1026  
APPLICATION NOTE  
Figure 19 illustrates the unity-loop gain point and phase margin  
for the ADA4932 with RF/RG = 1 (noise gain = 2).  
The loop gain is 0 dB where the 300 Ω feedback resistance  
horizontal line intersects the transimpedance magnitude curve.  
At this frequency, the phase of T(s) is approximately −135°,  
resulting in a phase margin of +45°. Phase margin and stability  
increase as RF increases and decrease as RF decreases. Current  
feedback amplifiers should always use purely resistive feedback  
with sufficient phase margin.  
80  
60  
40  
20  
90  
GAIN  
R
R
F
45  
1 +  
= 2  
(6dB)  
G
PHASE  
0
LOOP  
GAIN = 0dB  
–45  
–90  
–135  
–180  
–225  
–270  
PC BOARD LAYOUT  
6
0
When a stable ADC driver is designed, it must be realized on a  
PC board. Some phase margin is lost because of the parasitic  
elements of the board, which must be kept to a minimum. Of  
particular concern are load capacitance, feedback loop inductance,  
and summing node capacitance. Each of these parasitic reactances  
adds lagging phase shift to the feedback loops, thereby reducing  
phase margin. A design may lose 20° or more of phase margin  
due to poor PC board layout.  
–20  
–40  
–60  
–80  
PHASE  
MARGIN 70°  
1k  
10k  
100k  
1M  
10M  
100M  
1G  
10G  
FREQUENCY (Hz)  
Figure 19. ADA4932 Open-Loop Gain Magnitude and Phase vs. Frequency  
With voltage feedback amplifiers, it is best to use the smallest  
possible RF to minimize the phase shift due to the pole formed  
by RF and the summing node capacitance. If large RF is required,  
that capacitance can compensate with small capacitors, CF, across  
each feedback resistor with values such that RFCF equals RG  
times the summing node capacitance.  
Further examination of Figure 19 shows that the ADA4932 has  
approximately 50° of phase margin at a noise gain of 1 (100%  
feedback in each loop). Although it is not practical to operate ADC  
drivers at zero gain, this observation shows that the ADA4932  
can operate stably at fractional differential gains (for example,  
RF/RG = 0.25, noise gain = 1.25). This is not true for all differential  
ADC drivers. Minimum stable gains can be seen in all ADC  
driver data sheets.  
PC board layout is necessarily one of the last steps in a design.  
Unfortunately, it is also one of the most overlooked steps in a  
design, even though high speed circuit performance is highly  
dependent on layout. A high performance design can be  
compromised, or even rendered useless, by a sloppy or poor  
layout. Although all aspects of proper high speed PC board  
design cannot be covered here, a few key topics will be addressed.  
Phase margin for current feedback ADC drivers can also be  
determined from open-loop responses. Instead of forward gain,  
A(s), current feedback amplifiers use forward transimpedance,  
T(s), with an error current as the feedback signal. The loop gain  
of a current feedback driver with matched feedback resistors is  
T(s)/RF; therefore, the magnitude of the current feedback amplifier  
loop gain is equal to 1 (that is, 0 dB) when T(s) = RF. This point  
can be easily located on the open-loop transimpedance and phase  
plot, in the same way as for the voltage feedback amplifier. Plotting  
the ratio of a resistance to 1 kΩ allows resistances to be expressed  
on a log plot. Figure 20 illustrates the unity-loop gain point and  
phase margin of the ADA4927 current feedback, differential  
ADC driver with RF = 300 Ω.  
Parasitic elements rob high speed circuits of performance.  
Parasitic capacitance is formed by component pads and traces  
and ground or power planes. Long traces without ground planes  
form parasitic inductances, which can lead to ringing in transient  
responses and other unstable behaviors. Parasitic capacitance is  
especially dangerous at the summing nodes of an amplifier  
because it introduces a pole in the feedback response, causing  
peaking and instability. One solution is to make sure that the  
areas beneath the ADC driver mounting and feedback  
component pads are clear of ground and power planes  
throughout all layers of the board.  
1k  
100  
10  
50  
MAGNITUDE  
PHASE  
0
To minimize undesired parasitic reactances, keep all traces as  
short as possible. Outer layer 50 Ω PC board traces on FR-4  
contribute roughly 2.8 pF/inch and 7 nH/inch. These parasitic  
reactances increase by about 30% for inner layer 50 Ω traces.  
Additionally, make sure that there is a ground plane under long  
traces to minimize trace inductance. Keeping traces short and small  
helps to minimize both parasitic capacitance and inductance and  
maintains the integrity of the design.  
–50  
–100  
1
–135  
–150  
PHASE  
MARGIN 45°  
0.3  
0.1  
–180  
–200  
10  
100  
1k  
10k 100k  
1M  
10M 100M 1G  
10G  
FREQUENCY (Hz)  
Figure 20. ADA4927 Open-Loop Gain Magnitude and Phase vs. Frequency  
Rev. 0 | Page 12 of 16  
 
 
 
APPLICATION NOTE  
AN-1026  
Power supply bypassing is another key area of concern for layout;  
make sure that the power supply bypass capacitors, as well as the  
VOCM bypass capacitor, are located as close to the amplifier pins  
as possible. In addition, using multiple bypass capacitors on the  
power supplies helps to ensure that a low impedance path is  
provided for broadband noise. Figure 21 shows a typical differential  
amplifier schematic with power supply bypassing and a low-pass  
filter on the output. The low-pass filter limits the bandwidth and  
noise entering the ADC. Ideally, the power supply bypassing  
capacitor returns are close to the load returns; this helps to reduce  
circulating currents in the ground plane and improves ADC driver  
performance (see Figure 22).  
Use of ground plane, and grounding in general, is a detailed and  
complex subject and beyond the scope of this application note.  
However, a few key points are as follows (see Figure 22):  
Connect the analog and digital grounds together at one  
point only. This minimizes the interaction of analog and  
digital currents flowing in the ground plane, which  
ultimately leads to noise in the system.  
Terminate the analog power supply into the analog power  
plane and the digital power supply into the digital power  
plane.  
For mixed-signal ICs, terminate the analog ground returns  
into the analog ground plane and the digital ground returns  
into the digital ground plane and tie the two planes  
together using only one small connection to minimize the  
mixing of digital and analog currents (see Figure 23.)  
R4  
+V  
S
C2  
C3  
Refer to A Practical Guide to High-Speed Printed-Circuit-Board  
Layout for a detailed discussion about high speed printed  
circuit board (PCB) layout at www.analog.com.  
R2  
R
R6  
R7  
S
C6  
C7  
V
TO  
ADC  
OCM  
R
U1  
T
C1  
The information in this application note is intended to help you  
think about the many considerations that must be taken into  
account when you design with ADC drivers. Understanding  
differential amplifiers and paying attention to the details of ADC  
driver design at the outset of a project minimizes future problems,  
lowers risk, and ensures a robust design.  
R3  
R1  
C4  
C5  
–V  
S
R5  
Figure 21. ADC Driver with Power Supply Bypassing and Output Low-Pass Filter  
R2  
R4  
C2  
R6  
R7  
R
T
C3  
C4  
C6  
C7  
U1  
C1  
C5  
R1  
R3  
R5  
(a)  
(b)  
Figure 22. Component Side (a), Circuit Side (b)  
V
V
D
A
D
V
V
A
ANALOG  
CIRCUITS  
DIGITAL  
CIRCUITS  
MIXED  
SIGNALS  
AGND  
DGND  
SYSTEM  
STAR  
GROUND  
A
A
A
D
D
D
ANALOG  
CIRCUITS  
DIGITAL  
CIRCUITS  
ANALOG SUPPLY  
DIGITAL SUPPLY  
Figure 23. Mixed Signal Grounding  
Rev. 0 | Page 13 of 16  
 
 
 
AN-1026  
NOTES  
APPLICATION NOTE  
Rev. 0 | Page 14 of 16  
APPLICATION NOTE  
NOTES  
AN-1026  
Rev. 0 | Page 15 of 16  
AN-1026  
NOTES  
APPLICATION NOTE  
©2009 Analog Devices, Inc. All rights reserved. Trademarks and  
registered trademarks are the property of their respective owners.  
AN08263-0-11/09(0)  
Rev. 0 | Page 16 of 16  

相关型号:

AN-1027

Ultra Low Power Microcontroller Backup Using the EnerChip
CYMBET

AN-1028

Maximum Power Enhancement Techniques for SOT-223 Power MOSFETs
FAIRCHILD

AN-1029

Maximum Power Enhancement Techniques for SO-8 Power MOSFETs
FAIRCHILD

AN-1030

EVAL-08 Frequently Asked Questions & Troubleshooting
CYMBET

AN-1032

Performance Restrictions Associated with 3.5 Watts SO-8 Power MOSFETs
FAIRCHILD

AN-1036

Using the EnerChip CC in Energy Harvesting Designs
CYMBET

AN-1039

Correcting Imperfections in IQ Modulators to Improve RF Signal Fidelity
ADI

AN-1040

EVAL Kit GUI Software and TI Firmware Readme
CYMBET

AN-1041

EnerChip CC Li-Ion Battery Charger with Integrated Energy Storage
CYMBET

AN-1042

EnerChip CC Backup Power for Epson RX-8564 Real-Time Clock
CYMBET

AN-1043

EnerChip CC as Backup Power for a DS1390 Real-Time Clock
CYMBET

AN-1045

EnerChip CC Backup Power for Micro Crystal RV-2123 RTC
CYMBET