Hi, I have been designing a simple two stage op-amp using Sky130. So far I have been able to achieve about 56 dB of gain. My concern now is how I can ensure that my bias voltages remain constant - namely, the bias voltage at the output of the first-stage OTA and the bias voltage at the output node of the common-source second stage. I have a vague understanding of common-mode feedback. Is CMFB the solution to ensuring that the bias voltages remain where I want them to be? And if so, any advice about how CMFB can be implemented into my circuit would be greatly appreciated. I have read Razavi's sections on CMFB but I am unsure as to how CMFB will look like in my circuit. I am hoping to tape-out this 2-stage op amp with my undergraduate project team at Cornell University sometime in the Spring (via ChipIgnite). Thanks once again. Side note, if you spot any areas of concern within my circuit (all bias voltages are shown, as well as the tail current), please let me know (we also have a constant gm current reference that will replace the ideal current source shown here)! I am eager to learn how to better improve my circuit / my Analog ICs understanding.

Are you talking about the voltage labeled Vmid?

You need to use output feedback in your testbenches. See those testbenches here: https://www.edn.com/designing-with-a-complete-simulation-test-bench-for-op-amps-part-1-output-impedance/ https://www.edn.com/designing-with-a-complete-simulation-test-bench-for-op-amps-part-2-small-signal-bandwidth/ https://www.edn.com/designing-with-a-complete-simulation-test-bench-for-op-amps-part-3-input-referred-errors/

@Jon Ho To somewhat answer your CMFB question. You need to make sure that the bias is stable and is done by CMFB feedback without impacting your core. In your design, CMFB is not needed as for the first stage, the output dc voltage is set by the tail current source. As for the output stage, since you use the OPA in some feedbac configuration, the output dc will be set via the feedback externally and hence dont need CMFB either. For CMFB generation, a true differential inputs are needed. CMFB voltage will be the average of these two which can be derived by high value resistors. This then will be compared with a fixed Vref using another OPA, the output of the OPA will control the tail current.

Thanks for the insights guys. @ArunAshok I've actually implemented the CMFB loop that you mentioned since I made the post. Is this what you're referring to? So far, the CMFB loop I have seems to work in the sense that the bias voltage at output is set exactly to my reference of 1V.

@Jon Ho Are you simulating open loop or close loop gain ? In OPA test bench, you might be connecting your vout to either vp or vn depending on the feedback network. So setting vp or vn externall will also set the vout dc bias point. Hence I said you 'dont' need cmfb for vout node here.

I am currently simulating open-loop gain. @ArunAshok

It is not a practical amplifier. It could be a band-pass filter. It has no low frequency gain. And, this way, it can be unstable.