The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> STB and transient analyses inconsistent
https://designers-guide.org/forum/YaBB.pl?num=1269782046

Message started by michaelFCT on Mar 28^th, 2010, 6:14am

Title: STB and transient analyses inconsistent
Post by michaelFCT on Mar 28^th, 2010, 6:14am

Hi,

I designed a 2 stage amplifier (with compensation) and when simulated (setup in attachment) using tran, it shows that the output common-mode of the amplifier is unstable.
I then used STB (substituting the SC-CMFB for an equivalent RC circuit) and PSTB, using the actual SC-CMFB, and placed the cmdmprobe (cmdm=1) at the output of the amplifier, thus breaking all loops in the circuit. STB and PSTB analyses always return a PM > 65º.
The DC OP point has been preserved, and the circuit is also unstable for Vin=vss (without stimulation).

Why can't I detect this instability? Is there something wrong with the setup?

Thanks in advance.

Title: Re: STB and transient analyses inconsistent
Post by sheldon on Mar 28^th, 2010, 7:18pm

Michael,

Observations:
1) I am not sure what you are doing with the input source, should the
negative reference be the vcmi?
2) You don't show all the signs on the input source so it is not clear
how you have configured it.
3) Neither of common-mode feedback blocks appear to be connected
to vcmi, how is the common-mode level defined?
4) Why is the common-mode level output from the SC-CMFB
connected to the CT-CMFB?

Best Regards,

Sheldon

Title: Re: STB and transient analyses inconsistent
Post by michaelFCT on Mar 29^th, 2010, 1:52am

Answering Sheldon's questions,

1) What I'm performing (or trying to achieve) with the vcvs is the feedback of vod with feedback factor (β) = 1. vx = vin - vod, this is the diff. input of the balun.

2) Now shown in setup (below).

3) Sorry about that. Vcmi is connected to the CMFB blocks (now shown in setup).

4) The architecture of the opamp needs 2 CMFB circuits. We are using one SC-CMFB circuit to adjust the 2nd stage and a CT-CMFB to adjust the 1st stage.

Thanks for the feedback,
Michael.

Title: Re: STB and transient analyses inconsistent
Post by subgold on Mar 29^th, 2010, 7:01am

a little more detail would help others to answer your question, such as:

how the schematic of the amp looks like, especially how you feedback the cm signal and the 2 cmfb blocks?

how the cm probe looks like?

etc.

anyway, i expect the cm loop to be unstable, since your cm signal is going through 4 stages! if you want to regulate each stage of your amp separately, why dont you just configure a local cmfb of your 1st stage? that way, the stability much easier to achieve.

Title: Re: STB and transient analyses inconsistent
Post by michaelFCT on Mar 29^th, 2010, 7:59am

Sorry about the lack of detail, but I am unable to disclose any information about the amplifier and the CT-CMFB.

The SC-CMFB can be found in "Choksi, O.; Carley, L.R.; , "Analysis of switched-capacitor common-mode feedback circuit," Circuits and Systems II: Analog and Digital Signal Processing, IEEE Transactions on , vol.50, no.12, pp. 906- 917, Dec. 2003 doi: 10.1109/TCSII.2003.820253" (Fig. 10).

By cm probe do you mean the "CMDMPROBE" block? If yes, this block is from Cadence's analogLib. If no, I don't know what you mean.

I have already sized the transistors for this amp. and got it work as I wanted (DM and CM were stable). I then tried a different sizing and got the ringing CM.
I want to know, why I can't detect it with STB analysis. It obviously is because of the sizing, but I must be able to detect the instability with an AC simulation of some kind.

Michael.

Title: Re: STB and transient analyses inconsistent
Post by subgold on Mar 29^th, 2010, 12:42pm

michaelFCT wrote on Mar 29^th, 2010, 7:59am:

ok, i am not sure if what i am about to say is your case, just want to give you a hint.

in your very first post you said "thus breaking all loops in the circuit", but you should be a little cautious on this claim. you have to distinguish between many cases:

1) there are multiple feedback loops in parallel,
2) there is a global feeback, composed by some inner feeback loops,
3) there are cross-coupled feedback loops
4) more complicated situation (e.g. feedforward involved)
5) ....

the claim you referred is always valid only for case 1, and you circuit seems to be case 2. i am afraid what you got from stb analysis is just the stablity summary of the inner loop. but obviously there are lots of simple ways to verify this. for example you can check what you get when you move the position of the probe, or simulate the inner loop alone (with the correct loading), etc.

Title: Re: STB and transient analyses inconsistent
Post by michaelFCT on Mar 29^th, 2010, 2:26pm

Hi subgold, thanks for your feedback.

I think you are correct that my circuit is case 2. I think by placing the cmdmprobe at the output before the global feedback (vod), I've broken this loop also. Could there be another critical point? Could the loop of the compensation capacitor cause this instability? How do I check for this?

I'm not sure now, but I think I've already simulated without global feedback and got the exact same result. This could be due to the fact that I'm only feeding back the differential signal, so in terms of common-mode the global feedback loop is open.

Michael.

Title: Re: STB and transient analyses inconsistent
Post by subgold on Mar 30^th, 2010, 6:22am

michaelFCT wrote on Mar 29^th, 2010, 2:26pm:

hi michael,

i suppose what you are interested in is the cm stability, right? so lets focus on that.

the cm signal in your circuit undergoes two feedback loops (neglect the miller cap for a moment):

1) an outer loop, composed by your amp, SC-cmfb, CT-cmfb
2) an inner loop, composed by the 2nd-stage of your amp, SC-cmfb
(notice this is what i guess, since you didn't disclose how you feed back vcm_1stage, vcm_2stage into your amp, probably the inner loop consists of the 1st-stage as well)

anyway, in order to have stable cm regulation, you need to guarantee the stability of both loops, because your cm regulation configuration applies to case 2, only one break point check is not sufficient.

by the position of your probe, i expect what you've got from your stb analysis is just the result of the inner loop. you have to check that of the outer loop as well. you can put a simple probe (e.g. vdc) at the node vcm_1stage to do that.

Title: Re: STB and transient analyses inconsistent
Post by michaelFCT on Mar 31^st, 2010, 4:28am

Hi subgold,

Yes I am interested in the CM stability of the amplifier (cmdm=1 in the CMDMPROBE).
About your "guessing", you are correct (sorry again for the lack of info.).

As stated in the stb analysis manual, if you break all loops at one critical point (my case = output of amplifier) the stb analysis should return the stability of the least stable loop (I think). This is what I've done. So I think one break point should be sufficient.

Michael.