AN1685: Development of a Voltage Feedback Spice Op-Amp Macromodel

Application Note 1685
Author: Jian Wang and Tamara Schmitz
Development of a Voltage Feedback Spice Op-Amp
Macromodel
Introduction
The Input Stage
A voltage-feedback amplifier macromodel has been developed
that simulates the most common effects, such as transient
response, frequency response, voltage noise and input/output
slew rate limiting. Detailed descriptions of each stage in the
model will be presented with examples of model performance
and correlation to actual device behavior.
The input stage of the zero-drift amplifier is shown in Figure 2.
The 100µA current source 'I2' feeds the PMOS input pair.
Normally, I2 should be chosen less than the quiescent current.
Remember, the ISL28133's typical supply current (RL = open)
is only 18µA. However, a small I2 (~10µA) would make the
input voltage noise too large to emulate. This will be discussed
later in the noise analysis part. Choose I2 = 100µA and use I1
to compensate back to the total quiescent current. Cin1 and
Cin2 are the input common mode capacitance and Cdiff is the
input differential mode capacitance.
Macromodels are developed for the customer instead of
releasing full transistor schematics. Of course, the most
accurate simulations are conducted from fully-extracted 3-D
device models. Not only would it be impractical to share these
models because of the need to accommodate the numerous
simulation platforms, but also because of proprietary reasons.
One of the first op-amp macromodel techniques was
developed by Boyle in 1974 and used only two transistors, a
few diodes and linear elements [1]. Linear elements like
resistors, capacitors, inductor and voltage/current control
sources simulate much faster than active elements and are
used to provide poles, zeros and any gain. For a DC model, a
voltage-controlled voltage source can represent the amplifier
while resistances can be added to better represent the input
and output impedance. Capacitors, inductors, diodes and
transistors can then provide the proper AC response. If you
want more information on the development of simulation
models, see Alexander and Bowers [2] and [3]. We will follow
their model here.
The ISL28133
As an example, we are going to investigate the ISL28133. The
ISL28133 is a zero-drift operational amplifier with voltage
feedback topology. Intended for low frequency and power
precision applications, the gain bandwidth product is 400kHz,
the slew rate is 0.1V/µs and the supply current is 18µA. A
five-stage model represents the actual circuit, the block
diagram for which is shown in Figure 1. They are the input
stage, the gain stage, the frequency-shaping stage, the output
stage and the noise module.
FIGURE 2. INPUT STAGE
Input Stage
Frequency
Shaping
Stages
Gain Stage
Output
Stage
The Gain Stage (Figure 3)
This stage performs some important functions in the model:
1. This stage sets the DC gain of the part. All the subsequent
stages provide unity DC gain.
2. It provides slew rate limiting.
Noise
Module
3. It adds the dominant pole to the AC characteristic.
FIGURE 1. THE BLOCK DIAGRAM OF ISL28133 MACROMODEL
February 20, 2012
AN1685.0
1
4. It level shifts the signal from two voltages referred to the
supplies to a single voltage referred to the mid-point.
5. It limits the full scale output swing.
CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures.
Copyright Intersil Americas Inc. 2012. All Rights Reserved.
1-888-INTERSIL or 1-888-468-3774 | Intersil (and design) is a trademark owned by Intersil Corporation or one of its subsidiaries.
All other trademarks mentioned are the property of their respective owners.
Application Note 1685
FIGURE 3. GAIN STAGE
Referring to Figure 3, Ga is the gain of block G1 and G2. Gb is the
gain of block G3 and G4.
SlewRate =
I 2 • R3 • Ga • R5 • Gb
C1
(EQ. 1)
Changing the value of V3 and V4 limits the slew rate. Also, R8/C1
and R7/C2 decide the dominant pole of this model. E1 is used to
set the reference level at the middle of Vcc and Vee.
Frequency-Shaping Stages
The "telescopic" frequency shaping techniques used here are
very common in op-amp modeling. It is easy to add more poles
and zeros. Each frequency-shaping block provides unity gain. A
zero-pole pair is included in this model and shown in Figure 4.
(EQ. 2)
R10 = R11 = 1MΩ
R9 = R12 = R10 ⋅ (
L1 = L 2 =
fp
fz
− 1)
(EQ. 3)
FIGURE 4. ZERO-POLE PAIR STAGE
R9
2πf p
(EQ. 4)
A higher order pole stage is shown in Figure 5, G7/8, R13/14
and C3/4.
2
AN1685.0
February 20, 2012
Application Note 1685
is also called 1/f noise. Normally, the frequency where the flicker
noise curve crosses the white noise curve is defined as the corner
frequency. The small amount of flicker noise that remains is
modeled within the SPICE diode model. Referring to Figure 6,
in2 = 2qI d + KF •
I dAF
frequency
(EQ. 10)
Id is the DC diode current. AF and KF are the model parameters
of the SPICE diode and q is the charge of the electron. The flicker
noise exponent (AF) is set to 1 and the flicker noise coefficient
(KF) is set
KF =
Ea2
2R 2 • I d
(EQ. 11)
where Ea is the noise-voltage spectral density at 1Hz. The
simulated voltage noise will show the 1/f noise-voltage spectral
density with the correct corner frequency.
FIGURE 5. HIGHER ORDER POLE STAGE
Noise Simulation
The ISL28133 input current noise is very small (~70fA), so it is
neglected in this model. The voltage noise of the MOSFET can be
modeled like the following equations.
2 1
K
+
Vi 2 ( f ) = 4kT ( )
3 g m WLCox f
(EQ. 5)
gm ∝ I D
(EQ. 6)
ID is the drain current. High bias current in the model is needed
to emulate the low voltage noise. At the input stage, the tail
current is set high enough to generate the low input voltage
noise. Before the noise sources are added, the model has to be
rendered lower noise than the spec or typical performance noise
curve in the datasheet. The noise-voltage module of Figure 6
generates 1/f and white noise by using a 0.1V voltage source
biasing a diode-resistor series combination. White noise is
generated by the thermal noise-current.
in2 =
4kT
R
(EQ. 7)
where k is the Boltzmann's constant. So, the required value of
the resistor for a given noise-voltage spectral density is:
R=
en2
2 × 4kT
(EQ. 8)
1
f
β
Output Stage
After the frequency shaping-stages, the signal appears at Node
VV5, which is referenced to the midpoint of two supply rails. Each
controlled source can generate enough current to support the
desired voltage drop across its paralleled resistor. R15 and R16
are equal to twice of the open loop output resistance, so their
parallel combination gives the correct Zout. D5-D8 and G9/10
are used to force a current from the positive rail to the negative
rail to correct the real current sink or source in the supply pins.
G9 = G10 = G11 = G12 =
where en is the spectral density of the white noise voltage. The
design of the chopper stabilized amplifier greatly reduced 1/f
noise. 1/f noise (flicker noise) refers to the noise exhibiting
power spectral density inversely proportional to the frequency.
More generally, the noise with the spectral density
SN ∝
FIGURE 6. NOISE VOLTAGE MODULE
β >0
R15 = R16 = 2Z out
1
2Z out
(EQ. 12)
(EQ. 13)
(EQ. 9)
3
AN1685.0
February 20, 2012
Application Note 1685
FIGURE 10. LARGE SIGNAL STEP RESPONSE. THE SLEW RATE
SIMULATED IS 0.198V/µs AND THE ERROR IS 1%
FIGURE 7. OUTPUT STAGE
Simulation Results
Some SPICE simulation results are compared with the typical
performance curve from the datasheet in the following.
(Figure 8b is from the datasheet).
FIGURE 8. GAIN vs FREQUENCY vs LOAD CAPACITANCE. IT CANNOT
BE VERY ACCURATE BECAUSE THE PARASITIC
CAPACITANCE ON THE BOARD WASN’T INCLUDED IN
THE MODEL. THE ERROR IS LESS THAN 5%
FIGURE 11. INPUT NOISE VOLTAGE vs FREQUENCY. AT 1kHz, THE
SIMULATED INPUT NOISE VOLTAGE IS 64.9 nV/Hz, VERY
CLOSE TO THE VALUE IN THE DATASHEET 65nV/Hz. THE
SIMULATED CURVE CANNOT CATCH THE PEAK NEAR
10kHz
Conclusion
A truly comprehensive SPICE macromodel for a voltage feedback
amplifier is developed. This macromodel includes effects such as
transfer response, accurate AC response, DC offset and voltage
noise. It is easy to add more features like CMRR, PSRR, input
current-noise, etc. Also it is convenient to change the parameters
of the model to fit other voltage feedback amplifiers. Actually,
several of Intersil's voltage feedback amplifiers use the same
model topology.
ISL28133 Macromodel Netlist
* ISL28133 Macromodel
* Revision B, July 2009 by Jian Wang
* This model simulates AC characteristics, Voltage Noise,
Transient Response
* Connections: +input -input +Vupply -Vsupply Vout
.subckt ISL28133
3
2 7
4
6
*Input Stage
FIGURE 9. FREQUENCY RESPONSE OF CLOSED LOOP GAIN WITH
DIFFERENT GAIN. AT GAIN = 100, THE BANDWIDTH IS
3.94kHz AND THE ERROR IS LESS THAN 5%. AT LOW
GAIN, THE BANDWIDTH IS EXPANDED BECAUSE OF THE
ZERO POLE PAIR
C_Cin1
8 0 1.12p
C_Cin2
2 0 1.12p
C_Cd
8 2 1.6p
R_R1
9 10 10
R_R2
10 11 10
R_R3
4 12 100
R_R4
4 13 100
M_M1
12 8 9 9 pmosisil
+ L=50u
4
AN1685.0
February 20, 2012
Application Note 1685
+ W=50u
C_C4
M_M2
13 2 11 11 pmosisil
4 VV5 0.12p
R_R13
4 VV5 1meg
+ L=50u
R_R14
VV5 7 1meg
+ W=50u
*Output Stage
I_I1
4 7 DC 92uA
G_G9
I_I2
7 10 DC 100uA
G_G10
21 4 6 VV5 0.0000125
22 4 VV5 6 0.0000125
*Gain stage
D_D5
4 21 DY
G_G1
4 VV2 13 12 0.0002
D_D6
4 22 DY
G_G2
7 VV2 13 12 0.0002
D_D7
7 21 DX
R_R5
4 VV2 1.3Meg
D_D8
7 22 DX
R_R6
VV2 7 1.3Meg
R_R15
4 6 8k
D_D1
4 14 DX
R_R16
6 7 8k
D_D2
15 7 DX
G_G11
6 4 VV5 4 -0.000125
V_V3
VV2 14 0.7Vdc
G_G12
7 6 7 VV5 -0.000125
V_V4
15 VV2 0.7Vdc
*Voltage Noise
*SR limit first pole
D_DN1
102 101 DN
G_G3
4 VV3 VV2 16 1
D_DN2
104 103 DN
G_G4
7 VV3 VV2 16 1
R_R21
0 101 120k
R_R7
4 VV3 1meg
R_R22
0 103 120k
R_R8
VV3 7 1meg
E_EN
8 3 101 103 1
C_C1
VV3 7 12u
V_V15
102 0 0.1Vdc
C_C2
4 VV3 12u
V_V16
104 0 0.1Vdc
D_D3
4 17 DX
.model pmosisil pmos (kp=16e-3 vto=10m)
D_D4
18 7 DX
.model DN D(KF=6.4E-16 AF=1)
V_V5
VV3 17 0.7Vdc
.MODEL DX D(IS=1E-18 Rs=1)
V_V6
18 VV3 0.7Vdc
.MODEL DY D(IS=1E-15 BV=50 Rs=1)
*Zero/Pole
.ends ISL28133
E_E1
16 4 7 4 0.5
G_G5
4 VV4 VV3 16 0.000001
G_G6
7 VV4 VV3 16 0.000001
L_L1
20 7 0.3H
R_R12
20 7 2.5meg
R_R11
VV4 20 1meg
L_L2
4 19 0.3H
R_R9
4 19 2.5meg
R_R10
19 VV4 1meg
*Pole
G_G7
4 VV5 VV4 16 0.000001
G_G8
7 VV5 VV4 16 0.000001
C_C3
VV5 7 0.12p
5
AN1685.0
February 20, 2012
Application Note 1685
References
[1] BOYLE, G.R., "Macromodeling of integrated circuit
operational amplifiers”, IEEE J. 1974, SC-9.
[2] Derek Bowers, Mark Alexander, Joe Buxton, "A
Comprehensive Simulation Macromodels for 'Current
Feedback' Operational Amplifiers,” IEEE Proceedings, Vol.
137, April 1990 pp.137-145
[3] Mark Alexander, Derek Bowers, "AN-138 SPICE-Compatible
Op Amp Macro-Models", Analog Devices Inc., Application
Note 138.
[4] "AN-840 Development of an Extensive SPICE Macromodel
for 'Current-Feedback' Amplifiers", National Semiconductor
Corp., Application Note 840.
Intersil Corporation reserves the right to make changes in circuit design, software and/or specifications at any time without notice. Accordingly, the reader is
cautioned to verify that the Application Note or Technical Brief is current before proceeding.
For information regarding Intersil Corporation and its products, see www.intersil.com
6
AN1685.0
February 20, 2012