STMicroelectronics AN2653 Operational amplifier stability compensation method Datasheet

AN2653
Application note
Operational amplifier stability compensation methods
for capacitive loading applied to TS507
Introduction
Who has never experienced oscillations issues when using an operational amplifier? Opamps are often used in a simple voltage follower configuration. However, this is not the best
configuration in terms of capacitive loading and potential risk of oscillations.
Capacitive loads have a big impact on the stability of operational amplifier-based
applications. Several compensation methods exist to stabilize a standard op-amp. This
application note describes the most common ones, which can be used in most cases.
The general theory of each compensation method is explained, and based on this, specific
data is provided for the TS507. The TS507 is a high precision rail-to-rail amplifier, with very
low input offset voltage, and a 1.9 MHz gain bandwidth product, which is available in
SOT23-5 and SO-8 packages.
This document simplifies the task of designing an application that includes the TS507. It
spares you the time-consuming effort of trying numerous combinations on bench, and it is
also much more accurate than using Spice models which are not designed to study system
stability, even though they can give a general trend.
November 2007
Rev 1
1/22
www.st.com
Contents
AN2653
Contents
1
Stability basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2
Operational amplifier modeling for stability study . . . . . . . . . . . . . . . . . . . . 4
2
Stability in voltage follower configuration . . . . . . . . . . . . . . . . . . . . . . . 6
3
Out-of-the-loop compensation method . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4
5
3.1
Theoretical overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2
Application on the TS507 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
In-the-loop compensation method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.1
Theoretical overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.2
Application on the TS507 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Snubber network compensation method . . . . . . . . . . . . . . . . . . . . . . . 16
5.1
Theoretical overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2
Application on the TS507 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
7
Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2/22
AN2653
Stability basics
1
Stability basics
1.1
Introduction
Consider a linear system modeled as shown in Figure 1.
Figure 1.
Linear system with feedback model
Vin
A
Vout
ß
The model in Figure 1 gives the following equation:
A
V out = ----------------- ⋅ V in
1 + Aβ
A ---------------is named closed loop gain.
1 + Aβ
From this equation, it is evident that for Aβ = -1, the circuit is unstable (Vout is independent of
Vin).
Aβ is the loop gain.
To evaluate it, the loop is opened and -Vr/Vs is calculated as shown in Figure 2.
Figure 2.
Loop gain calculation
A
Vout
Vs
Vr
ß
Opening the loop leads to the following equation:
V
– ------r = Aβ
Vs
If a small signal Vs is sourced into the system, and if Vr comes back in phase with it with an
amplitude above that of Vs (which means that Aβ is a real number greater than or equal to 1)
then the system oscillates and is unstable.
This leads to the definition of the gain margin, which is the opposite of the loop gain (in dB)
at the frequency for which its phase equals -180°. The bigger the gain margin, the more
stable the system. In addition, the phase margin is defined as the phase of the loop gain
plus 180° at the frequency for which its gain equals 0 dB. Therefore, from the value of Aβ it
is possible to determine the stability of the system.
3/22
Stability basics
1.2
AN2653
Operational amplifier modeling for stability study
Figure 3 illustrates the definition of phase and gain margins in a gain configuration.
Illustration of phase and gain
margins
Figure 4.
TS507 open loop gain
TS507 Open Loop Gain
Loop Gain
-45
130
-90
100
-60
70
-90
-180
40
-120
-160
-225
10
-150
-200
-270
-20
-180
Frequency (Hz)
Gain
-30
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+08
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
1.E+00
-120
1.E+02
-135
1.E+01
Phase Margin
-80
1.E+00
Gain (dB)
-40
1.E-01
Gain Margin
0
TS507 :
Vcc = 5 V
Vicm = 2.5 V
T = 25 °C
1.E-02
0
Phase (°)
160
Gain (dB)
0
40
Phase (°)
Figure 3.
Frequency (Hz)
Phase
Gain
Phase
To apply this stability approach to operational amplifier based applications, it is necessary to
know the gain of the operational amplifier when no feedback and no loads are used. It is the
open loop gain (A(ω )) of the amplifier (shown in Figure 4 for the TS507). From this
parameter, it is possible to model the amplifier and to study the stability of any gain
configuration.
Figure 5.
Equivalence between schematics and block diagram
The loop gain is:
Rg
V
– ------r = A ( ω) ⋅ ------------------Rf + Rg
Vs
This equation shows the impact of the gain on the stability: if Rf/Rg increases, the closed
loop gain of the system increases and the loop gain decreases. Because the phase remains
the same, the gain margin increases and stability is improved.
In addition, if you consider the case of a second order system such as the one shown in
Figure 6, a decrease of the loop gain allows to pass the 0 dB axis before the second pole
occurs. It minimizes the effect of the phase drop due to this pole, and as a result, the phase
margin is higher. Therefore, a voltage follower configuration is the worst case for stability.
4/22
AN2653
Stability basics
Figure 6.
Impact of closed loop gain on stability
Gloop gain (dB)
Case 1
Case 2
0
f
Closed Loop Gain (Case1) < Closed Loop Gain (Case 2)
Another parameter that impacts stability is the amplifier output impedance Zo. Including
this parameter in the model of the amplifier leads to the model shown in Figure 7.
Follower configuration model with
capacitive load for loop gain
calculation
Figure 8.
TS507 output impedance Zo
TS507 Output Impedance (Zo)
1.E+05
90
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
1.E+04
45
Phase (°)
1.E+07
1.E+06
1.E+05
-135
1.E+04
-90
1.E+00
1.E+03
1.E+01
1.E+02
-45
1.E+01
1.E+02
1.E+00
0
1.E-01
1.E+03
1.E-02
Impedance (Ohm)
Figure 7.
Frequency (Hz)
Impedance
Phase
Zo is neither constant over frequency nor purely resistive. Figure 8 shows how the output
impedance varies with the frequency in the case of the TS507. These variations complicate
the stability study.
Finally, to study the stability of an op-amp based system, two parameters need to be taken
into account in order to better fit reality: the amplifier open-loop gain and the amplifier output
impedance. Then, a calculation of the loop gain indicates how stable the system is.
5/22
Stability in voltage follower configuration
2
AN2653
Stability in voltage follower configuration
This section examines a voltage follower configuration because it is the worst case scenario
for stability (compared with a gain configuration).
Figure 9.
Voltage follower configuration
Figure 10. Closed loop gain measured for a
voltage follower configuration
Voltage Follower Configuration - Closed Loop Gain
20
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
Gain (dB)
10
0
Without CL
-10
-20
CL=550 pF
-30
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
-40
Frequency (Hz)
Gain without CL
Gain with CL = 550 pF
In voltage follower configuration, the loop gain is:
V
A ( ω)
– ------r = ----------------------------------------Vs
Zo
1 + ------- + jZ o C L ω
RL
The capacitive load adds a pole to the loop gain that impacts the stability of the system. The
higher the frequency of this pole, the greater the stability. In fact, if the pole frequency is
lower than or close to the unity gain frequency, the pole can have a significant negative
impact on phase and gain margins. It means that the stability decreases when the
capacitive load increases.
Without CL, the system is stable. However, Figure 11 and Figure 12 show, for the TS507, the
oscillations due to instability with and without an AC input signal for a capacitive load of
550 pF. The oscillation frequency is in line with the peaking frequency observed in a closed
loop gain configuration (approximately 1.9 MHz according to Figure 10).
6/22
AN2653
Stability in voltage follower configuration
Figure 11. Input and output signals measured Figure 12. Input and output signals measured
with grounded input
for an AC input signal
Voltage Follower Configuration Output Signal with Input Grounded
Voltage Follower Input and Output Signals
0.08
0.04
0.1
0.02
0
0.05
0
-0.02
-0.05
-0.04
-0.1
-0.06
0.00
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 550 pF
0.15
Amplitude (V)
Amplitude (V)
0.2
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 550pF
0.06
-0.15
0.50
1.00
1.50
Time (μs)
Output Signal
2.00
0
100
200
300
400
500
Time (μs)
Input Signal
Output Signal
Input Signal
To remove this instability and work with higher capacitive loads, many compensation
methods exist, and this application note examines some of them. By adding zeroes and
poles to the loop gain, stability can be improved.
However, compensation components have to be chosen carefully. A compensation scheme
can indeed improve stability, but can also lead the system to instability, depending on the
choice of component values. Similarly, a compensation configuration can work for a specific
load, but modifying this load can affect stability.
7/22
Out-of-the-loop compensation method
AN2653
3
Out-of-the-loop compensation method
3.1
Theoretical overview
A simple compensation method, using only one extra component, consists in adding a
resistor in series between the output of the amplifier and its load (see Figure 13). It is often
referred to as the out-of-the-loop compensation method because the additional component
(ROL) is added outside of the feedback loop. The resistor isolates the op-amp feedback
network from the capacitive load.
Figure 13. Out-of-the-loop compensation
schematics
Figure 14. Out-of-the-loop equivalent
schematics for loop gain
calculation
From Figure 14, the loop gain with this compensation method is:
R OL
A ( ω) ⋅ ⎛ 1 + ---------- + jR OL C L ω⎞⎠
⎝
Vr
RL
– ------ = --------------------------------------------------------------------------------R OL + Z o
Vs
1 + ----------------------- + j ( Z o + R OL )C L ω
RL
This compensation introduces a zero in the loop gain, just after the pole caused by the
capacitive load, at:
1
----------------------------------------------2π ⋅ R OL | | R L C L
This pole is also unfortunately shifted to lower frequencies at:
1
----------------------------------------------------------------2π ⋅ ( Z o + R OL ) | | R L C L
However, due to the zero, the effect of the pole is minimized and the stability is improved. To
obtain a good level of stability, ROL must be chosen such that the frequency of the zero
occurs at least one decade before unity-gain frequency. It then allows a significant shift of
the phase and therefore increase phase and gain margins.
The previous equation shows that if ROL >> Zo, then -Vr/Vs= A(ω ), and the circuit is stable.
In that case, pole and zero occur at the same frequency. However, the value of ROL is limited
by the load impedance, ROL and RL acting as a a divider bridge from the operational
amplifier output. Therefore, in order to minimize the error on Vout, ROL must be very small
compared to RL (for example, a maximum of 1%, but this criterion depends on the required
accuracy).
Finally, this compensation method is effective, but the drawback is a limitation on the
accuracy of Vout depending on the resistive load value.
8/22
AN2653
3.2
Out-of-the-loop compensation method
Application on the TS507
This compensation method brings very good results in terms of stability, improving strongly
the phase and gain margins. Table 1 and Table 2 show the results obtained for different load
conditions, in the case of voltage follower and gain configurations. Note that ROL is limited to
1% of RL even though better results can be obtained with higher values of ROL.
Table 1.
Results of out-of-the-loop compensation for different load conditions in the case of a
voltage follower configuration for TS507
RL = 1 kΩ
CL
ROL
(Ω)
fu/fz(1)
RL = 10 kΩ
Mg(2)
(dB)
(degree)
Mϕ(2)
-4.1
-28.5
-2.5
-16.8
-22.2
-78.4
-14
-32.4
-34.1
-84.4
17.1
6.8
ROL
(Ω)
fu/fz(1)
RL = 100 kΩ
Mg(2)
(dB)
(degree)
Mϕ(2)
-5
-34.1
16
26.9
-22.9
-79.5
23
37
-34.4
-84.6
23.4
39.4
ROL
(kΩ)
fu/fz
Mg(2)
(dB)
(degree)
Mϕ(2)
-5.1
-34.4
22.4
52.1
-23
-79.6
22.6
52.3
-34.5
-84.6
22.6
52.3
1 nF
10
0.11
100
1.13
1
11.3
10 nF
10
1.13
100
11.3
1
112.3
100 nF
10
11.3
100
113.3
1
1126
1. fu/fz cells are shaded when the value is lower than 10, which is not the best case due to ROL limitation.
2. Negative values indicate instability.
Table 2.
Results of out-of-the-loop compensation for different load conditions in the case of a
gain configuration of either -10 or +11 (Rg = 100 Ω and Rf = 1 kΩ) for TS507
RL = 1 kΩ
CL
ROL
(Ω)
fu/fz(1)
RL = 10 kΩ
Mg(2)
(dB)
Mϕ(2)
(degree)
17.6
84.7
19
84.7
-0.6
-16.1
7.2
81.2
-13
-69.2
38
41
ROL
(Ω)
fu/fz(1)
RL = 100 kΩ
Mg(2)
(dB)
Mϕ(2)
(degree)
16.8
85.1
36.9
85.1
-1.3
-25.7
43.9
81.4
-13.3
-69.8
44.3
80.6
ROL
(kΩ)
fu/fz
Mg(2)
(dB)
Mϕ(2)
(degree)
16.7
85.2
43.4
85
-1.4
-25.9
43.4
84.8
-13.3
-69.9
43.4
84.8
1 nF
10
0.11
100
1.13
1
11.3
10 nF
10
1.13
100
11.3
1
112.6
100 nF
10
1.
11.3
100
113.3
1
1126
fu/fz cells are shaded when the value is lower than 10, which is not the best case due to ROL limitation.
2. Negative values indicate instability.
As expected, Table 1 and Table 2 show that the higher the value of ROL, the better the
compensation (because the best ROL is always its maximum value RL/100).
These results also show that, for a voltage follower configuration, this compensation method
does not work with low RL (and low CL), because the zero frequency cannot be one decade
before the unity-gain frequency of the open loop gain. In the case of the TS507, it works well
only if the ROL.CL product is above 10-6.
9/22
Out-of-the-loop compensation method
AN2653
Figure 15 and Figure 16 show the loop gain and closed loop gain respectively. These curves
are plotted for RL = 10 kΩ and CL = 1 nF.
Figure 15. Loop gain
Figure 16. Measured closed loop gain
Voltage Follower Configuration - Closed Loop Gain
Compensation with the Out-of-the-Loop Technique
0
With Out-of-the-Loop Compensation Technique
-10
-20
Without Compensation
-30
1.E+07
1.E+06
1.E+05
1.E+03
1.E+04
-40
Frequency (Hz)
Gain without Compensation
Phase without Compensation
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
10
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
1.E+00
1.E-01
20
Gain (dB)
0
-30
-60
-90
-120
-150
-180
-210
-240
-270
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
Phase (°)
150
130
110
90
70
50
30
10
-10
-30
1.E-02
Gain (dB)
Voltage Follower Configuration - Loop Gain
Compensation with the Out-of-the-Loop Technique
Frequency (Hz)
Gain with ROL = 100 Ω
Phase with ROL = 100 Ω
Gain without Compensation
Gain with ROL = 100 Ω
Both figures further demonstrates the stability improvement.
Note that the fact that Zo is almost a self at high frequencies (for the TS507) explains the
presence of peaking in the loop gain curve, depending on the load capacitor. This is
because the denominator is equal to
R OL ⎛ L ( ω)
1 + ---------- + j ----------- + R OL C L ω⎞ – L ( ω)C L ω2
⎝ R
⎠
RL
L
with Zo = jL(ω )ω.
It leads to a resonance frequency of approximately
1
------------------------------------2π ⋅ L ( ω)C L
For the peaking frequency
1
f = -----------------------------------2π ⋅ L ( ω)C L
the damping is given by the term:
L
( ω)- + R C ω
---------OL L
RL
When there is no compensation, it is only:
L
( ω)---------RL
With the compensation, at the resonance frequency,
L
( ω)- « R C ω
---------OL L
RL
therefore the peaking is attenuated.
To help implement the compensation, the abacus given in Figure 17 to Figure 20 provide the
ROL value to choose for a given CL and phase/gain margins. These abacus are plotted in
the case of a voltage follower configuration and a gain configuration of -10 or +11, with a
load resistor of 10 kΩ..
10/22
AN2653
Out-of-the-loop compensation method
Figure 17. Gain margin abacus in the case of a Figure 18. Phase margin abacus in the case of
voltage follower configuration
a voltage follower configuration
Voltage Follower Configuration - Phase Margin Abacus
Applied for Out-of-the-Loop Compensation Method
Voltage Follower Configuration - Gain Margin Abacus
Applied for Out-of-the-Loop Compensation Method
100
100
STABLE
10
ROL (Ω)
1
1
UNSTABLE
0.1
UNSTABLE
0.1
0.01
CL (F)
R1 (0 dB)
R1 (4 dB)
R1 (8 dB)
1.E-04
1.E-05
1.E-06
1.E-07
1.E-08
1.E-10
1.E-04
1.E-05
1.E-06
1.E-07
1.E-08
1.E-09
1.E-10
1.E-11
0.01
1.E-09
ROL (Ω)
10
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
STABLE
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL (F)
R1 (12 dB)
R1 (0 °)
R1 (16 dB)
R1 (10 °)
R1 (20 °)
R1 (30 °)
R1 (40 °)
Figure 19. Gain margin abacus in the case of a Figure 20. Phase margin abacus in the case of
gain configuration of -10 or +11
a gain configuration of -10 or +11
Gain Configuration of either -10 or +11 - Phase Margin Abacus
Applied for Out-of-the-Loop Compensation Method
Gain Configuration of either -10 or +11 - Gain Margin Abacus
Applied for Out-of-the-Loop Compensation Method
100
100
STABLE
ROL (Ω)
UNSTABLE
0.1
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
STABLE
10
1
1
UNSTABLE
0.1
0.01
CL (F)
R1 (0 dB)
R1 (10 dB)
1.E-04
1.E-05
1.E-06
1.E-07
1.E-08
1.E-09
1.E-10
1.E-04
1.E-05
1.E-06
1.E-07
1.E-08
1.E-09
1.E-10
0.01
1.E-11
ROL (Ω)
10
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL (F)
R1 (20 dB)
R1 (30 dB)
R1 (0 °)
R1 (20 °)
R1 (40 °)
R1 (60 °)
11/22
In-the-loop compensation method
AN2653
4
In-the-loop compensation method
4.1
Theoretical overview
Figure 21 shows a commonly used compensation method, often called in-the-loop, because
the additional components (a resistor and a capacitor) used to improve the stability are
inserted in the feedback loop.
Figure 21. In-the-loop compensation
schematics
Figure 22. In-the-loop equivalent schematics
for loop gain calculation
The loop gain in this configuration, corresponding to Figure 22, is the following:
A ( ω) ⋅ [ 1 + jR IL C IL ω]
V
– ------r = ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------Z
Z o + R IL
Vs
1 + -------------------- + j ( Z o + R IL )C L ω + jR IL ⎛⎝ 1 + ------o-⎞⎠ C IL ω – Z o R IL C IL C L ω2
R
R
L
L
It adds a zero and splits the pole caused by the capacitive load into two poles in the loop
gain. This compensation method allows, by a good choice of compensation components, to
compensate the original pole (caused by the capacitive load), and then to improve stability.
The main drawback of this circuit is the reduction of the output swing, because the isolation
resistor is in the signal path.
Note that, for the following cases, RIL is limited to 10% of RL (or Rf // RL in the case of a gain
configuration) even if better results can be obtained with higher RIL values.
But because the feedback loop is taken directly on Vout, the RIL / RL divider bridge does not
create inaccuracy on Vout as it does with the out-of-the-loop method.
4.2
Application on the TS507
In the case of the TS507, the first pole of the loop gain caused by the feedback occurs
around:
Z o + R IL
1 + -------------------RL
1
---------------------------------------------------------------------------------------------------- ≈ ---------------------------------------------------------------------|
|
Z
⋅
(
R L ) ( C IL + C L )
2π
R
o⎞
IL
⎛
2π ⋅ ( Z o + R IL )C L + R IL 1 + ------- C IL
⎝
R ⎠
L
12/22
AN2653
In-the-loop compensation method
The second one occurs at higher frequencies where its impact on stability is limited. The
goal of the first pole is to decrease the loop gain to get closer to 0 dB, just before the zero,
occurring at
1
------------------------------2π ⋅ R IL C IL
whose goal is to minimize the phase shift caused by the pole. The stability is increased as
the loop gain crosses the 0 dB axis with a limited phase shift. It minimizes the effect of the
second pole caused by the feedback, which is also pushed toward higher frequencies.
Although this compensation method may seem difficult to set up, it brings very good results,
as shown in Table 3 and Table 4, for the TS507 operational amplifier.
Table 3.
Results of in-the-loop compensation for different load conditions in the case of a
voltage follower configuration for TS507
RL = 1 kΩ
CL
RIL(1)
(Ω)
CIL
(nF)
RL = 10 kΩ
Mg(2)
(dB)
(degree)
Mϕ(2)
-4.1
-28.5 °
4.7
24.5 °
-22.2
-78.4 °
6
21.9 °
-34.1
-84.4 °
6.5
34.3 °
RIL(1)
(kΩ)
CIL
(nF)
RL = 100 kΩ
Mg(2)
(dB)
(degree)
Mϕ(2)
-5
-34.1 °
15.2
53.9 °
-22.9
-79.5 °
13.6
61.2 °
-34.4
-84.6 °
6.5
66.9 °
RIL(1)
(kΩ)
CIL
(nF)
Mg(2)
(dB)
(degree)
Mϕ(2)
-5.1
-34.4 °
24.3
71.9 °
-23
-79.6 °
13.3
79.2 °
-34.5
-84.6 °
6.2
70.6 °
1 nF
100
1
1
0.4
10
0.2
10 nF
100
2
1
1.26
5
1.26
100 nF
79.4
7.9
0.5
6.3
0.63
6.3
1. RIL cells are shaded when its value is clamped to RL/10.
2. Negative values indicate instability.
Table 4.
Results of in-of-the-loop compensation for different load conditions in the case of a
gain configuration of either -10 or +11 (Rg = 100 Ω and Rf = 1 kΩ) for TS507
RL = 1 kΩ
CL
RIL(1)
(Ω)
CIL
(pF)
RL = 10 kΩ
Mg(2)
(dB)
Mϕ(2)
(degree)
17.6
84.7 °
39
89 °
-0.6
-16.1 °
40.2
78.6 °
-13
-69.2 °
44.5
66.2 °
RIL(1)
(Ω)
CIL
(pF)
RL = 100 kΩ
Mg(2)
(dB)
Mϕ(2)
(degree)
16.8
85.1 °
39
88.1 °
-1.3
-25.7 °
40.3
84.9 °
-13.3
-69.8 °
44.5
65.6 °
RIL(1)
(Ω)
CIL
(pF)
Mg(2)
(dB)
Mϕ(2)
(degree)
16.7
85.2 °
39
88 °
-1.4
-25.9 °
40.3
84.8 °
-13.3
-69.9 °
44.5
65.6 °
1 nF
100
126
100
126
100
126
10 nF
39.8
251
31.6
316
31.6
316
100 nF
10
631
10
631
10
631
1. RIL cells are shaded when its value is clamped to (RL// Rf)/10.
2. Negative values indicate instability.
In a gain configuration, when considering the loop gain, the output is loaded by a resistive
load of RL // (Rf + Rg), where Rf and Rg are the resistors used for the gain. If Rf + Rg << RL,
the loop gain and therefore the stability parameters are the same whatever the value of RL.
This is visible in Table 4 where Rf + Rg = 1.1 kΩ with RL = 10 kΩ and RL = 100 kΩ..
13/22
In-the-loop compensation method
AN2653
Table 5 and Table 6 help you to choose the best compensation components for different
ranges of load capacitors (and with RL = 10 kΩ) in voltage follower configuration and in a
gain configuration of either -10 or +11.
However, each case of load can be improved by choosing specific components (seeTable 3
and Table 4).
Table 5.
Best compensation components for different load capacitor ranges in voltage
follower configuration for TS507 (with RL = 10 kΩ)
Load capacitor range
RIL (kΩ)
CIL (pF)
Minimum gain
margin (dB)
Minimum phase
margin (degree)
10 pF to 100 pF
1
251
16.8
54.9 °
100 pF to 1 nF
1
251
15.8
42.1 °
1 nF to 10 nF
1
631
10.9
27 °
10 nF to 100 nF
1
2500
3.8
18.4 °
Table 6.
Best compensation components for different load capacitor ranges in a gain
configuration of either -10 or +11 (Rg = 100 Ω and Rf = 1 kΩ) for TS507 (with RL = 10 kΩ)
Load capacitor range
RIL (Ω)
CIL (pF)
Minimum gain
margin (dB)
Minimum phase
margin (degree)
10 pF to 100 pF
1000
40
39.2
88.8 °
100 pF to 1 nF
39.8
63
37.2
86.8 °
1 nF to 10 nF
63
251
36.5
70.7 °
10 nF to 100 nF
15.8
631
39.1
63.1 °
These tables are very valuable because almost all the follower and gain configuration
applications requiring compensation have capacitive loads in the range of 100 pF to 1 nF.
Thus, a simple combination of (RIL, CIL), depending on RL can cover all these cases with a
very good stability.
The loop gain shown in Figure 23, plotted for a voltage follower configuration with CL = 1 nF
and RL = 10 kΩ, shows the instability without compensation. This can also be observed in
Figure 24 with the peaking present on closed loop. Both figures show the benefits of
compensation.
Figure 23. Loop gain
Figure 24. Measured closed loop gain
Voltage Follower Configuration - Closed Loop Gain
Compensation with the In-the-Loop Technique
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
10
With In-the-Loop Compensation Technique
-20
Without Compensation
-30
Frequency (Hz)
Gain without Compensation
Phase without Compensation
14/22
Gain with RIL = 1 kΩ and CIL = 470 pF
Phase with RIL = 1 kΩ and CIL = 470 pF
Gain without CL
Gain with RIL = 1 kΩ and CIL = 470 pF
1.E+07
1.E+06
1.E+05
-40
1.E+04
Frequency (Hz)
0
-10
1.E+03
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
1.E+00
1.E-01
20
Gain (dB)
0
-30
-60
-90
-120
-150
-180
-210
-240
-270
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
Phase (°)
150
130
110
90
70
50
30
10
-10
-30
1.E-02
Gain (dB)
Voltage Follower Configuration - Loop Gain
Compensation with the In-the-Loop Technique
AN2653
In-the-loop compensation method
The zero and the pole introduced by the compensation are visible on the loop gain.
Introducing first the pole at
1
---------------------------------------------------------------------- = 125kHz
2π ⋅ ( R IL | | R L ) ( C IL + C L )
leads to a gain fall (with a slope of -40 dB/decade with compensation) which allows to come
closer to unity gain. The zero, occurring at
1
-------------------------------- = 400kHz
2π ⋅ R IL C IL
leads to an upturn of the phase so that when unity gain is reached, the effect of the first
poles are limited in terms of phase shifting. Thus, the circuit is stable with a good phase
margin. Furthermore, it leads to an excellent gain margin, because the gain keeps falling
whereas the phase increases due to the zero, before finally decreasing to reach the -180°
point.
15/22
Snubber network compensation method
AN2653
5
Snubber network compensation method
5.1
Theoretical overview
Figure 25 shows another way to stabilize an operational amplifier driving a capacitive load.
The snubber network compensation method consists in adding an RC series circuit
connected between the output and the ground. It is particularly recommended for lower
voltage applications, where the full output swing is needed.
Figure 25. Snubber network schematics
Figure 26. Snubber network equivalent
schematics for loop gain
calculation
Introducing a second load resistor RSN in the circuit decreases the resistive load, and as a
result, pushes the pole caused by the capacitive load to higher frequencies, from
1
-----------------------------------------------2π ⋅ ( Z o | | R L )C L
to
1
-------------------------------------------------------------------2π ⋅ ( Z o | | R L | | R SN )C L
Therefore, stability is increased.
Furthermore, adding a serial capacitor CSN with RSN removes the impact of RSN in DC.
On one hand, CSN must be big enough to consider that its impedance is small compared to
RSN at the frequencies where RSN plays its role of stabilizer.
On the other hand, RSN being very small, CSN must be small enough because when the
frequency increases, the system becomes limited by the current flowing through RSN and
CSN (depending on the output voltage swing).
V
I = -------------------------------j R SN – ------------C SN ω
Therefore, in the following examples, CSN is limited to 100 nF in order to not limit the
frequency range.
16/22
AN2653
Snubber network compensation method
In fact, this compensation introduces a zero and an additional pole into the loop gain:
A ( ω) ⋅ ( 1 + jR SN C SN ω)
V
– ------r = --------------------------------------------------------------------------------------------------------------------------------------------------------------------------Zo
Z
Vs
1 + ------- + jZ o ( C L + C SN )ω + j ⎛ 1 + ------o-⎞ R SN C SN ω – Z o R SN C L C SN ω2
⎝
RL
R L⎠
Because Zo is not a pure resistance over frequency, choosing the minimum RSN is not
always the best case.
5.2
Application on the TS507
For the TS507, according to the abacus in Figure 27 and Figure 28, this compensation
method in the case of a voltage follower configuration, works only if the capacitive load is
less than 1 nF, in order to obtain (at least) a phase margin of 20°.
Figure 27. Gain margin abacus in a voltage
follower configuration
Figure 28. Phase margin abacus in a voltage
follower configuration
Voltage Follower Configuration - Load Abacus
Phase Margin
Voltage Follower Configuration - Load Abacus
Gain Margin
1.E-06
1.E-06
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
1.E-07
1.E-08
CL (F)
CL (F)
UNSTABLE
UNSTABLE
1.E-08
1.E-09
1.E-09
1.E-10
1.E-10
1.E-11
1.E-12
1.E+00
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
1.E-07
1.E+01
1.E+02
1.E+03
1.E+04
STABLE
1.E-11
STABLE
1.E+05
1.E+06
1.E-12
1.E+00
1.E+07
1.E+01
1.E+02
1.E+03
RL (Ω)
0 dB
10 dB
1.E+04
1.E+05
1.E+06
1.E+07
RL (Ω)
20 dB
30 dB
0°
C1 (10 °)
20 °
C1 (30 °)
40 °
Figure 29 and Figure 30 are the same abacus in case of a gain of either -10 or +11.
Figure 29. Gain margin abacus in a gain
configuration of either -10 or +11
Figure 30. Phase margin abacus in a gain
configuration of either -10 or +11
Gain Configuration of either -10 or +11 - Load Abacus
Phase Margin
Gain Configuration of either -10 or +11 - Load Abacus
Gain Margin
1.E-06
1.E-06
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
UNSTABLE
1.E-07
1.E-08
CL (F)
CL (F)
1.E-08
1.E-09
1.E-10
1.E-09
1.E-10
1.E-11
1.E-12
1.E+00
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
UNSTABLE
1.E-07
1.E+01
1.E+02
1.E+03
1.E+04
STABLE
1.E-11
STABLE
1.E+05
1.E+06
1.E+07
1.E-12
1.E+00
1.E+01
1.E+02
RL (Ω)
0 dB
10 dB
20 dB
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
RL (Ω)
30 dB
40 dB
50 dB
0°
20 °
40 °
60 °
80 °
You can use the abacus in Figure 27 to Figure 30 to determine the best RSN value.
17/22
Snubber network compensation method
AN2653
Table 7 to Table 10 give the results of compensation in a follower configuration and several
gain configurations for different load conditions.
Table 7.
Results of snubber network compensation for different load conditions in the case of
a voltage follower configuration for TS507
RL = 1 kΩ
CL
RSN
(Ω)
CSN
(nF)
RL = 10 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
-4.1
-28.5 °
7.3
16.9 °
-22.2
-78.4 °
-7.5
-18.3 °
RSN
(Ω)
CSN
(nF)
RL = 100 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
-5
-34.1 °
7.7
16.7 °
-22.9
-79.5 °
-7.6
-18.5 °
RSN
(Ω)
CSN
(nF)
Mg(1)
(dB)
Mϕ(1)
(degree)
-5.1
-34.4 °
7.7
16.8 °
-23
-79.6 °
-7.7
-18.5 °
1 nF
34.4
100
31.6
100
31.6
100
10 nF
13.5
100
13.5
100
13.5
100
1. Negative values indicate instability.
Table 8.
Results of snubber network compensation for different load conditions in the case of
a gain configuration of either -1 or +2 (Rf = Rg = 1 kΩ) for TS507
RL = 1 kΩ
CL
RSN
(Ω)
CSN
(nF)
RL = 10 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
2.4
28.7 °
11
37.6 °
-15.7
-70.9 °
-1.5
-4 °
RSN
(Ω)
CSN
(nF)
RL = 100 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
1.5
19.4 °
10.9
37.6 °
-16.5
-72.4 °
-1.6
-4.1 °
RSN
(Ω)
CSN
(nF)
Mg(1)
(dB)
Mϕ(1)
(degree)
1.4
19.6 °
10.6
37.6 °
-16.6
-72.6 °
-1.6
-4.1 °
1 nF
58.9
100
56.2
100
58.9
100
10 nF
13.5
100
13.5
100
13.5
100
1. Negative values indicate instability.
Table 9.
Results of snubber network compensation for different load conditions in the case of
a gain configuration of either -10 or +11 (Rg = 100 Ω and Rf = 1 kΩ) for TS507
RL = 1 kΩ
CL
RSN
(Ω)
CSN
(nF)
RL = 10 kΩ
Mg(1)
(dB)
(degree)
Mϕ(1)
17.6
84.7 °
21.6
81.8 °
-0.6
-16.1 °
11.3
60.6 °
-13
-69.2 °
-5.2
-33.3 °
RSN
(Ω)
CSN
(nF)
RL = 100 kΩ
Mg(1)
(dB)
(degree)
Mϕ(1)
16.8
85.1 °
21.6
81.8 °
-1.3
-25.7 °
11.3
60.7 °
-13.3
-69.8 °
-5.4
-31.7 °
RSN
(Ω)
CSN
(nF)
Mg(1)
(dB)
(degree)
Mϕ(1)
16.7
85.2 °
21.6
81.8 °
-1.4
-25.9 °
11.3
60.6 °
-13.3
-69.9 °
-5.4
-31.7 °
1 nF
171.1
100
149.1
100
146.8
100
10 nF
26.1
100
25.7
100
25.5
100
100 nF
17.1
100
1. Negative values indicate instability.
18/22
15.3
100
15.3
100
AN2653
Snubber network compensation method
Table 10.
Results of snubber network compensation for different load conditions in the case of
a gain configuration of either -30 or +31 (Rg = 100 Ω and Rf = 3 kΩ) for TS507
RL = 1 kΩ
CL
RSN
(Ω)
CSN
(nF)
RL = 10 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
26
88.2 °
29.8
87.3 °
7.9
87.9 °
19.6
82.7 °
-4.2
-44.7 °
3.4
63.8 °
RSN
(Ω)
CSN
(nF)
RL = 100 kΩ
Mg(1)
(dB)
Mϕ(1)
(degree)
25.2
88.4 °
29.8
87.3 °
7.2
88 °
19.2
82.7 °
-4.5
-45.8 °
3.3
64 °
RSN
(Ω)
CSN
(nF)
Mg(1)
(dB)
Mϕ(1)
(degree)
25.1
88.4 °
29.8
87.3 °
7.1
88.1 °
19.1
82.8 °
-4.5
-45.9 °
3.3
64.1 °
1 nF
198
100
168.5
100
166
100
10 nF
30.7
100
30
100
30
100
100 nF
16.6
100
16.5
100
16.5
100
1. Negative values indicate instability.
The results are almost the same whether RL = 1 kΩ, 10 kΩ or 100 kΩ because in all cases, at
the frequency range that has a significant impact on stability, the resistive load on the
amplifier is RSN // RL ≅ RSN.
From Table 7, you can see that, in the case of a voltage follower configuration, this
compensation method doesn’t work for capacitive loads higher than 1 nF because CSN is
limited to 100 nF. Figure 27 and Figure 28 show that lower RSN values would give better
results. However, in this case, these abacus are not valid because the CSN impedance is not
negligible compared with RSN at the frequency range where RSN plays its role of stabilizer.
For RL = 10 kΩ and CL = 1 nF, the snubber network compensation method gives the loop
gain and closed loop gain shown in Figure 31 and Figure 32 respectively.
Figure 31. Loop gain
Figure 32. Measured closed loop gain
Voltage Follower Configuration - Closed Loop Gain
Compensation with the Snubber Network Technique
Gain with RSN = 30 Ω and CSN = 100 nF
Phase with RSN = 30 Ω and CSN = 100 nF
0
With Snubber Network Compensation Technique
-10
-20
Without Compensation
-30
1.E+07
1.E+06
1.E+05
1.E+04
-40
1.E+03
Frequency (Hz)
Gain without Compensation
Phase without Compensation
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
10
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
1.E+00
1.E-01
20
Gain (dB)
0
-30
-60
-90
-120
-150
-180
-210
-240
-270
TS507 :
Vcc = 5 V
Vicm = 2,5 V
T = 25 °C
RL = 10 kΩ
CL = 1 nF
Phase (°)
150
130
110
90
70
50
30
10
-10
-30
1.E-02
Gain (dB)
Voltage Follower Configuration - Loop Gain
Compensation with the Snubber Network Technique
Frequency (Hz)
Gain without Compensation
Gain with RSN = 30 Ω and CSN = 100 nF
This compensation has the main advantage of neither reducing the output swing nor the
gain accuracy, unlike the two other compensation methods. Nevertheless, for the TS507,
this method is limited to capacitive loads lower than 1 nF in a voltage follower configuration,
and the stability improvement provided by the compensation is not as good as with the two
other methods.
19/22
Conclusion
6
AN2653
Conclusion
Based on the results of the three compensation methods described in this application note,
it can be stated that the best compensation solution for the TS507 is the in-the-loop method.
This is the most complex solution to understand, but a pick-up table is provided in order to
choose the most appropriate components for your application. The main drawback of this
compensation method is a limited output swing.
The out-of-the-loop compensation method is easy to implement because it requires only
one extra component. The way it works is also easy to understand. However, its main
limitation is an inaccuracy on the output voltage because the load is part of a divider bridge.
An abacus is provided to choose the component you need for your application.
Finally, the snubber method is easy to understand and use. It does not have the drawbacks
of the first two solutions. But because the loop gain cannot be considered as a pure thirdorder system (because of the open loop gain of the amplifier and the variations of the output
impedance which is not purely resistive and constant over frequency), it does not lead to
great improvements, and it is limited to load capacitors lower than 1 nF in the case of the
TS507. An abacus is also provided for this compensation method. Another drawback of this
solution is that it potentially limits the frequency range of the application for large output
signals due to a strong current flowing through the compensation elements. This
compensation method is not really useful for the TS507.
In conclusion, for the TS507, whenever possible, we recommend to use the in-the-loop
compensation method.
Note, that this document provides typical values for the TS507 at ambient temperature.
Therefore, your chosen solution must in any case be checked on bench.
20/22
AN2653
7
Revision history
Revision history
Table 11.
Document revision history
Date
Revision
7-Nov-2007
1
Changes
Initial release.
21/22
AN2653
Please Read Carefully:
Information in this document is provided solely in connection with ST products. STMicroelectronics NV and its subsidiaries (“ST”) reserve the
right to make changes, corrections, modifications or improvements, to this document, and the products and services described herein at any
time, without notice.
All ST products are sold pursuant to ST’s terms and conditions of sale.
Purchasers are solely responsible for the choice, selection and use of the ST products and services described herein, and ST assumes no
liability whatsoever relating to the choice, selection or use of the ST products and services described herein.
No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted under this document. If any part of this
document refers to any third party products or services it shall not be deemed a license grant by ST for the use of such third party products
or services, or any intellectual property contained therein or considered as a warranty covering the use in any manner whatsoever of such
third party products or services or any intellectual property contained therein.
UNLESS OTHERWISE SET FORTH IN ST’S TERMS AND CONDITIONS OF SALE ST DISCLAIMS ANY EXPRESS OR IMPLIED
WARRANTY WITH RESPECT TO THE USE AND/OR SALE OF ST PRODUCTS INCLUDING WITHOUT LIMITATION IMPLIED
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE (AND THEIR EQUIVALENTS UNDER THE LAWS
OF ANY JURISDICTION), OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
UNLESS EXPRESSLY APPROVED IN WRITING BY AN AUTHORIZED ST REPRESENTATIVE, ST PRODUCTS ARE NOT
RECOMMENDED, AUTHORIZED OR WARRANTED FOR USE IN MILITARY, AIR CRAFT, SPACE, LIFE SAVING, OR LIFE SUSTAINING
APPLICATIONS, NOR IN PRODUCTS OR SYSTEMS WHERE FAILURE OR MALFUNCTION MAY RESULT IN PERSONAL INJURY,
DEATH, OR SEVERE PROPERTY OR ENVIRONMENTAL DAMAGE. ST PRODUCTS WHICH ARE NOT SPECIFIED AS "AUTOMOTIVE
GRADE" MAY ONLY BE USED IN AUTOMOTIVE APPLICATIONS AT USER’S OWN RISK.
Resale of ST products with provisions different from the statements and/or technical features set forth in this document shall immediately void
any warranty granted by ST for the ST product or service described herein and shall not create or extend in any manner whatsoever, any
liability of ST.
ST and the ST logo are trademarks or registered trademarks of ST in various countries.
Information in this document supersedes and replaces all information previously supplied.
The ST logo is a registered trademark of STMicroelectronics. All other names are the property of their respective owners.
© 2007 STMicroelectronics - All rights reserved
STMicroelectronics group of companies
Australia - Belgium - Brazil - Canada - China - Czech Republic - Finland - France - Germany - Hong Kong - India - Israel - Italy - Japan Malaysia - Malta - Morocco - Singapore - Spain - Sweden - Switzerland - United Kingdom - United States of America
www.st.com
22/22