[SOLVED] SC CMFB Two stage amplifier

[jgk2004]

Hello all,

When looking at all switch cap cmfb, I have noticed that most books/papers usually just use it around a single stage amplifier. When looking at the SC CMFB network, it is non-inverting, thus you can use it around one stage, the network drives the gate of a transistor, then inverts gate to drain, thus negative feedback. But when you have a two stage design, I can't seem to find any papers talking about SC-CMFB. Since if you drive the first stage with the SC network, you would have positive feedback in your CMFB loop(network non-inverting, stage 1 invert, stage 2 invert = loop positive feedback!). Thus is there an inverting CMFB network? I can't seem to find one or figure out how to make it inverting... I would assume, I would just do a CMFB around the first stage, then another one around the second stage, then adjusts the bandwidths of each loop so they don't race, thus make it stable.
Am I missing anything here?

Kind Regards,
JGK

 

[sutapanaki]

The SC CMFB is there to just substitute the averaging resistors that detect the common-mode voltage. And yes, it is non-inverting. But if you want to wrap it around two stages, you can combine it with just a simple diff pair, say, with diode loads.
Having a CMFB that measures the 2nd stage output and feeds to the 1st stage has its challenges too. When you consider the external feedback of the amplifier (the one you normally use in differential mode), in common-mode it turns into positive feedback one and it can latch the amplifier to either of the rails.

 

[jgk2004]

Hello Sutapanaki,
Thanks for the feedback and comments. I will for sure try the diff pair with diodes to add another stage to invert back. I was not considering this, since it adds another pole in the feedback loop and thus to push it out, I would need to burn current in this diff pair which would stink.

As for your other statement, I am a little confused. So in normal CT opamps/ct-cmfb, its very common to measure second stage, and feedback to first stage. This of course is a third order feedback in the cmfb loop which you need to consider, but that is fine. As for the "external feedback", do you mean outside the opamp so example, you do an inverting amplifier and you have your R2 in feedback? If so, your right, that is positive feedback but the diff tail helps reject that, but correct me if I am wrong, that's always there, even if you do multi loop- or feed around one stage.. Thats why when you break your loop or STB/middlebrook, you have to break that loop and the inner loop to measure both loops in parallel for the true stability.

So anyone else have any papers/references on nested SC-cmfb in two stage or higher amplifiers? Any help would be great!

JGK

 

[sutapanaki]

Adding an extra diff pair, diode-loaded stage for the CMFB should not be problematic. It does add an extra pole, but because of the diode loads, that pole is at high frequencies and could always be tackled so that stability is insured.

Yes, I meant the external feedback, the Rf around the amplifier. It is true that the tail current source degenerates that positive feedback in normal operation but troubles usually happen at power up. Doesn't always happen and not in all corners but when it happens it latches and you can't get the circuit out of that state. People sometimes put startup circuit to make sure that latch state never happens. It is less troublesome, when you have CMFB for each stage separately.

 

[jgk2004]

Hello Sutapanaki,
Thanks for the feedback.

Your correct with the diode loaded stage, not a problem, just more current.

Super good point on the latch... I never considered it like you state it but your right. I have always looked at that as startup issues, which as you stated you can put in the startup circuit to stop that, which I have normally done in the past. I have never considered it as a possible issue during normal operation since I knew my input driver and everything was CT based. I guess with SC circuits it can happen more often.

So in answering my first question, so it is totally normal/common to put in single stage sc-cmfb around each stage of the amplifier? In CT cmfb, I have done for years. Like when doing it sc-cmfb, I don't see any problems corners/MC. I was just being conservative and wanted to ask since I haven't read it anywhere...

JGK

 

[sutapanaki]

Yes, correct