During system design we often need to verify how accurate the vendor supply models are. An opamp is extensively used in analog and mixed signal system design, this document tells you how to verify the opamp models supplied by the manufacturers.
4
By Soufiane Bendaoud Amplifiers Marketing Manager National Semiconductor Corp. Spice models have gained popu- larity over the past years. While IC manufacturers strive to provide their customers with accurate models, system designers dictate the trend of accuracy and in- novations in Spice macro model development. Many IC companies claim that their models are the best and that they offer revolutionary features. What they often fail to provide is a circuit that allows their cus- tomer to verify the accuracy of the macro model. Op-amp macro models are probably the most sought after. While they can be extremely helpful when accurate, they can also cause serious problems, es- pecially when they are not in the hands of experts. Most system design engi- neers take the time to test the op amp macro model by itself before implementing it in a more comprehensive circuit. I’ve seen people get erroneous simulation results and then call up the apps group of the IC manufacturer to tell them that the model they’ve developed is not functional. When the applications person tries to get some details about what’s not working in the model, the response is usually, “I used a different model from your com- petitor, and it works just fine in my circuit.” The truth is that all models are not the same, and some may not work in a particular setting. Instead of investigating op amp macro models’ shortcomings, it would be helpful to have a collec- tion of circuits that allows users to test any model, in other words, universal test circuits for op amp macro models. Parameters for test Macro models differ in their levels of complexity or simplicity. Much like the datasheets, the models should emulate parameters rel- evant to applications in which the op amp is thought to be ap- propriate. For example, if you use a rail to rail output op amp, you should be able to test and verify Figure 1: When testing open loop gain and phase, the user should choose a frequency range that goes beyond the unity gain bandwidth of the amplifier. Figure 2: When running the simulation for slew rate, make sure your setup is in transient. Figure 3: CMRR is especially important in non-inverting configurations because of modulation. Figure 4: PSRR is important in any application where the voltage supply is susceptible to any interference. Electronic Engineering Times-Asia | November 1-15, 2006 | eetasia.com Verify macro model accuracy with PSpice test circuits SPICE
Transcript
By Soufiane BendaoudAmplifiers Marketing ManagerNational Semiconductor Corp.
Spice models have gained popu-larity over the past years. While IC manufacturers strive to provide their customers with accurate models, system designers dictate the trend of accuracy and in-novations in Spice macro model development.
Many IC companies claim that their models are the best and that they offer revolutionary features. What they often fail to provide
is a circuit that allows their cus-tomer to verify the accuracy of the macro model.
Op-amp macro models are probably the most sought after. While they can be extremely helpful when accurate, they can also cause serious problems, es-pecially when they are not in the hands of experts.
Most system design engi-neers take the time to test the op amp macro model by itself before implementing it in a more comprehensive circuit. I’ve seen people get erroneous simulation
results and then call up the apps group of the IC manufacturer to tell them that the model they’ve developed is not functional. When the applications person tries to get some details about what’s not working in the model, the response is usually, “I used a different model from your com-petitor, and it works just fine in my circuit.”
The truth is that all models are not the same, and some may not work in a particular setting. Instead of investigating op amp macro models’ shortcomings, it
would be helpful to have a collec-tion of circuits that allows users to test any model, in other words, universal test circuits for op amp macro models.
Parameters for testMacro models differ in their levels of complexity or simplicity. Much like the datasheets, the models should emulate parameters rel-evant to applications in which the op amp is thought to be ap-propriate. For example, if you use a rail to rail output op amp, you should be able to test and verify
Figure 1: When testing open loop gain and phase, the user should choose a frequency range that goes beyond the unity gain bandwidth of the amplifier.
Figure 2: When running the simulation for slew rate, make sure your setup is in transient.
Figure 3: CMRR is especially important in non-inverting configurations because of modulation.
Figure 4: PSRR is important in any application where the voltage supply is susceptible to any interference.
�Electronic Engineering Times-Asia | November 1-15, 2006 | eetasia.com
Verify macro model accuracy with PSpice test circuits
the output saturation voltage vs. the load current. Likewise, a low-noise amplifier should have a model that emulates at least the voltage noise.
Despite their differences, am-plifier macro models have some things in common. These are the most interesting parameters—they are usually the starting point of the simulation.
Open loop gain, phase margin—The open loop gain vs. frequency is probably the first test that en-gineers perform to evaluate the amplifier’s macro-model perfor-mance. This test is important be-cause it shows the DC gain, -3dB frequency, crossover frequency (gain bandwidth product in the
case of a voltage feedback ampli-fier) and the phase margin all in one simple little circuit.
Figure � shows the test cir-cuit. The RC network ensures that the output is biased at a suitable DC voltage. At higher frequencies, the capacitor shorts the inverting input to ground, placing the op amp in open loop. The capacitor is chosen to be large to provide an early roll off (f = 2πRC). Thus, even if the op amp you’re testing has a very low frequency dominant pole, you would be able to simulate, see a smooth transition and the 20dB per decade roll off.
When test ing open loop gain and phase, the user should choose a frequency range that
goes beyond the unity gain band-width of the amplifier.
When using a rail-to-rail out-put model, it is important to use the test circuit with the same load indicated in the datasheet. Otherwise, you will probably not get the result you would expect, especially the DC gain (AOL = gmRL).
Slew rate—This is another ele-ment defining the amplifier speed that every model should have. Slew rate is defined by the ratio of the tail current over the compensation capacitance.
Depending on the macro model, the capacitor that deter-mines the slew rate can be placed at the input or in a separate network.
Since we already know the relationship Idt = Cdv, we can simply use the circuit of Figure 2 and take the derivative of the output to get the slew rate. To do this, just use the insert command in the probe screen and type in the letter d before the output
voltage that appears.When running the simulation
for slew rate, make sure your setup is in transient. Ensure that the input signal also has fast enough rise time and fall time (edges) as to not limit the slew rate. On the other hand, the input signal frequency must be chosen according to the op amp’s speed. An input signal that’s too fast will cause convergence problems.
CMRR, PSRR—These two parame-ters are not always modeled, but they can be equally important. CMRR and PSRR are fairly easy to implement in a model, as they usually consist of a simple RC network, resistor divider and volt-age-controlled voltage source.
CMRR is especially important in non-inverting configurations because of modulation. On the other hand, PSRR is important in any application where the volt-age supply is susceptible to any interference.
The test circuits presented in Figures 3 and 4 allow the user to
Figure 5: When modeled correctly, the output impedance helps to get a more accurate settling-time behavior when the amplifier is driving capacitive loads.
Figure 6: To simulate the current-noise density, use the same circuit and place a 100kΩ resistor in series with the non-inverting terminal.
Figure 7: You can use any circuit previously described, and simply activate the voltage and current probes in PSpice to see them.
2 Electronic Engineering Times-Asia | November 1-15, 2006 | eetasia.com
simulate these two parameters. If they are modeled correct-ly, the pole and zero location should match the graphs in the datasheet.
Note that you can choose to invert either CMRR or PSRR on the plot screen by simply inserting a minus sign preceding the volt-age probe.
Output impedance—This is a specification often omitted from the datasheet altogether, but is sometimes necessary.
When modeled correctly, the output impedance helps to get a more accurate settling-time behavior when the amplifier is driving capacitive loads. Output impedance is also needed to calculate the proper component values when a compensation scheme is considered for stability purposes.
The test circuits provide the user with three curves for the
output impedance at different gains in the frequency domain. The output impedance is ob-tained by taking the ratio of the output voltage over the 1A cur-rent source.
The graph in Figure 5 shows the output impedance of the LMV791 to be approximately 100Ω.
Voltage, current noise—This is one of the areas in which the creators of amplifier macro models have made progress. Some of today’s models allow you to simulate the voltage noise with its flicker noise component and current noise with excellent accuracy. Modeling noise into the macro model doesn’t take much more computing or simulation time. However, it can be a difficult task, at least until you have figured out the right equations that make the voltage-noise density curve mimic the datasheet graph with
the 1/f corner as well.You can easily test the volt-
age-noise density by taking the output of a voltage follower (with a voltage source of 0V) on a log-log scale. To simulate the cur-rent-noise density, use the same circuit and place a 100kΩ resistor in series with the non-inverting terminal. In the probe window, make sure to divide the result by 100E3 or whatever resistor value you choose.
Using a large resistor value makes the current noise domi-nate since it is coupled into the resistance. Thus, voltage and thermal noises become negli-gible compared to the current noise.
Make sure to specify the out-put voltage in the analysis setup window of PSpice. In Figure 6, we specify the output voltage as Vout and the input voltage as Vin, and check the box “noise-enabled.”
Input bias current input offset voltage—These parameters are probably the easiest to model. The input offset voltage can eas-ily be implemented as a voltage-controlled voltage source at the input whose value is taken from the datasheet.
In general, you won’t even have to use a specific circuit to test Vos and IB. You can use any circuit previously described, and simply activate the voltage and current probes in PSpice to see them. Figure 7 shows the input bias current at 1.5pA and the in-put offset voltage at 1.48mV. Note
that the supply current is also shown to be 1.15mA at ±2.5V.
Output saturation voltage—This parameter is sometimes known as the dropout voltage. It is par-ticularly important in rail-to-rail output models. This is because it indicates the output swing as a function of the load current and can help you choose the appro-priate op amp, especially when driving heavy loads or when dy-namic range is a concern.
The test circuit uses a simple DC sweep with two equal input voltages of opposite magnitude to replicate the sourcing and sink-ing of the load current.
Supply current vs. supply volt-age—The test circuit in Figure 8 sweeps the current across the supply and allows you to determine how much current is drawn from the amplifier at dif-ferent supply voltages. This test is particularly helpful for power-conscious applications.
The slope of the supply cur-rent curve can easily be added into the model.
Overshoot, transient response—This test circuit serves two purposes: testing the transient response (whether small signal or large signal) and the overshoot.
Overshoot is important be-cause it indicates how much ring-ing an amplifier has with a capaci-tive load. Overshoot is a measure of stability in time domain; it is the equivalent of what peaking is in
Figure 8: This test circuit sweeps the current across the supply and allows you to determine how much current is drawn from the amplifier at different supply voltages.
Figure 9: The first test circuit uses a voltage-controlled voltage source.
Figure 10: In the second test circuit, we sweep the voltage from -2.5V to 2.5V.
3Electronic Engineering Times-Asia | November 1-15, 2006 | eetasia.com
passive components to mimic the overshoot accurately. But generally, if the phase margin is accurate, the overshoot should come pretty close to what it should be.
You can test the transient response using the same test circuit without the 100pF ca-pacitor. Some datasheets indicate whether a small capacitance is
used as a load when measuring the small signal transient. In that case, simply use the same value of capacitance.
Common-mode voltage range—This parameter is important as it allows the user to see the head room and how far away your in-put signal needs to be from the supply.
The first test circuit in Figure 9 uses a voltage-controlled voltage
source. In the second test circuit in Figure �0, we sweep the volt-age from -2.5V to 2.5V.
Phase reversal—Phase reversal occurs in some amplifiers when the input signal exceeds the input common-mode voltage range. During a phase reversal, the output changes polarity and may cause damage to the op amp, resulting in system lockups.
The test circuit is a simple
voltage follower with a sine wave input of 6V. The output waveform indicates that the macro model, just like the op amp, doesn’t exhibit any phase reversal. It is clipped at ±2.5V.
The test circuits described are not meant to replace the evalua-tion of the device on the bench. Rather, they provide the user with the flexibility of making quick assessments with respect to the accuracy of the macro model.
4 Electronic Engineering Times-Asia | November 1-15, 2006 | eetasia.com
