2
\$\begingroup\$

Shockingly simple question but for some reason I cannot get a configuration to work. I already designed a folded cascode op-amp and I can measure the slew rate using the Cadence function "SlewRate" and the differential output (vout_p - vout_n) with a differential input pulse, but this is for an open loop configuration with a capacitive load, like how you would expect to measure gain and phase margin. I got around 50V/us, but for my design this number seems suspiciously high. The answer is easy for single ended op-amps, as you can just use negative feedback, but I have not found a definitive answer anywhere for the commonly accepted configuration. Surely my brain is tired and I am missing something obvious, but I'd rather figure it out before bed.

\$\endgroup\$
3
  • \$\begingroup\$ The slew rate is specified (and must be measured) for strong negative feedback only! \$\endgroup\$
    – LvW
    Commented Dec 4, 2022 at 10:45
  • \$\begingroup\$ That is what I thought, but I am not finding a consistent answer anywhere online that indicates how this should actually look as a circuit schematic. Do you have a diagram by chance? Are there standardized values for feedback resistors and load capacitor? \$\endgroup\$ Commented Dec 4, 2022 at 19:35
  • \$\begingroup\$ There is no specific diagram for slew rate measurements. For this purpose, simply provide heavy feedback (for example unity closed-loop gain) and watch the output for a 1-volt input step. \$\endgroup\$
    – LvW
    Commented Dec 5, 2022 at 9:07

1 Answer 1

1
\$\begingroup\$

This isn't really an easy question at all as it's not often shown in texts.

The first thing to make sure of in order to apply a similar test you would use in a single ended configuration, is you need establish a common mode feedback and make sure the common mode at the output is fixed (usually half supply). When you are using single ended DC feedback it is working well because your feedback fixes the output to a known fixed value where the amplifier is operating in a linear region. Same applies with fully differential, but the operating point is fixed by the CMFB loop (even if the main loop is open!).

You can also use open loop, similar to what you do with single ended configurations. Your expected slew rate can be estimated using I=CL*SR, where CL is the load capacitance (no load would just be output node capacitance) and I is the bias current of the op amp, or just measure SR=dV/dT. Since it is likely an OTA, a resistance load is not necessary. However, you can add capacitive feedback with very large resistors in parallel to clamp the dc feedback, if you want to measure under a closed loop load.

For Gain and PM in open loop, as long as you have op pt. fixed by CMFB, you can just use estimated load capacitance on each leg (like e.g. 1p) and measure the fully differential G/PM by subtracting both output AC signals and plotting differential G/PM.

\$\endgroup\$
2
  • \$\begingroup\$ @ Pat I am not sure if your above description/explanation is really in accordance with slew rate DEFINITION: For a correct slew rate measurement the amplifier must have strong negative feedback, Then, an applied input step must bring the first diff. stage into saturation - until the feedback effect will bring the whole amplifier back to linear operation after a certain delay. Within this delay time, we can observe a quasi-linear slope of the output voltage. This defines the amplifiers slew rate. \$\endgroup\$
    – LvW
    Commented Dec 5, 2022 at 9:21
  • \$\begingroup\$ Because of the variation in results that I was finding with different configurations, I ended up utilizing an ideal element available in the cadence library to convert the differential outputs to a single ended output. Then, I applied a step with negative feedback like I normally would for a single ended amp. The result matched my expectations. \$\endgroup\$ Commented Dec 5, 2022 at 18:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.