“Trust, but verify” SPICE model accuracy, part 2: input offset voltage vs. input common-mode voltage

It’s no secret that low-voltage rail-to-rail input operational amplifiers (op amps) are gradually taking the place of traditional high-voltage amplifiers in many precision applications. Rail-to-rail input amplifiers are extremely useful, since their linear input-voltage range spans the entire power-supply voltage range (or even beyond). They traditionally achieve this span through the use of two pairs of input transistors instead of one pair, but you should be mindful of new design challenges that this topology creates.

One challenge is the change in the op amp’s input offset voltage (V_OS) when the amplifier input stage crosses over from one pair of transistors to another. This phenomenon is often called input crossover distortion. V_OS is an important performance characteristic of a precision op amp, and many systems must calibrate out the initial offset voltage to meet their performance goals. Any changes to V_OS, whether caused by changes in input common-mode voltage (V_CM), temperature or other variables, are highly undesirable and can throw off a system’s total error performance. Figure 1 gives an example of V_OS changing dramatically with increased V_CM.

Figure 1: V_OS vs. V_CM

When using SPICE simulation for rail-to-rail input amplifier designs, it’s wise to check that the V_OS vs. V_CM behavior of your models matches the real devices. Figure 2 shows the recommended test circuit.

Figure 2: V_OS vs. V_CM test circuit

This simple circuit places the op amp in a unity-gain buffer configuration to prevent output swing limitation issues, then sweeps V_CM to determine the change in V_OS. To plot V_OS vs. V_CM, run a DC transfer characteristic while stepping V_CM across the entire supply voltage range and measure V_OS across the op amp input pins as shown in Figure 2.

Let’s use this method to test the response of the OPA388, a new zero-crossover precision amplifier from TI that uses a charge pump in its input stage to achieve true rail-to-rail performance using only a single transistor pair. This eliminates the input crossover distortion found in traditional rail-to-rail input op amps. See Figure 3.

Figure 3: V_OS vs. V_CM results of the OPA388

The simulated results match the responses of the three test devices given in the OPA388 data sheet very closely, with a change of less than 1μV over the entire V_CM range.

Let’s use the same test circuit to check the response of the OPA2325, another zero-crossover precision amplifier from TI. See Figure 4.

Figure 4: V_OS vs. V_CM results for the OPA2325

Again, the simulated results match the real silicon very well. Keep in mind that while the simulation model looks like it has higher offset than the real silicon, all of the test devices measured in this plot had a V_OS lower than the typical spec of 40μV, while the SPICE model was designed to match the typical.

Thanks for reading the second installment of the “Trust, but verify” blog series! In the next installment, I’ll discuss how to measure open-loop output impedance and small-signal step response to perform stability analysis. If you have any questions about simulation verification, log in and leave a comment, or visit the TI E2E™ Community Simulation Models forum.

Additional resources

• Download the TI tech note, “Zero-Crossover Amplifiers: Features and Benefits.”
• Watch these videos to learn more about V_OS vs. V_CM:
  • “TI Precision Labs – Op Amps: Low Distortion Design 2.”
  • “TI Precision Labs – Op Amps: Input and Output Limitations 1.”