The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Fully Differential FC opamp SC CMFB
https://designers-guide.org/forum/YaBB.pl?num=1071115088

Message started by RobertNew on Dec 10^th, 2003, 7:58pm

Title: Fully Differential FC opamp SC CMFB
Post by RobertNew on Dec 10^th, 2003, 7:58pm

Hi all analog designers,

I am currently designing a fully differential folded cascode (FC) opamp for switched capacitor (SC) circuits. It's well known that we need a CMFB circuit to properly define the output common mode voltage and switched capacitor (SC) CMFB is chosen to implement this CMFB circuit. My question is how do I simulate the gain, slew-rate of the FC opamp with this SC CMFB which is sampled-data in nature? Thanks for your kind help.

Regards,
Robert

Title: Re: Fully Differential FC opamp SC CMFB
Post by Paul on Dec 10^th, 2003, 11:39pm

Hi Robert,

for DC and small signal analysis, you may replace the switched cap CMFB by a resistive one, using equivalent resistors as given by the capacitor values and the switching frequency.

Paul

Title: Re: Fully Differential FC opamp SC CMFB
Post by Ken Kundert on Dec 11^th, 2003, 7:34am

SpectreRF can be used to compute the small-signal characteristics (gain, bandwidth, noise, etc.) of your SC-CMFB amplifier. To do so, perform a PSS analysis with only the clock applied, then perform a PAC, PXF, and/or PNoise analysis to compute the small-signal characteristics. This process is described more fully in Simulating switched-capacitor filters with SpectreRF at http://www.designers-guide.com/Analysis/.

To simulate the slew rate, you could use envelope following, but transient is probably best unless the slewrate is very slow relative to the clock frequency.

-Ken

Title: Re: Fully Differential FC opamp SC CMFB
Post by RobertNew on Dec 11^th, 2003, 7:20pm

Hi Paul and Ken,

Thank you for your replies.
Ken, could you suggest similar tools in Mentor Graphics that I can use for the steps you mention (such as PSS etc.)?

Cheers,
Robert

Title: Re: Fully Differential FC opamp SC CMFB
Post by Ken Kundert on Dec 11^th, 2003, 10:16pm

No, Mentor does not really offer anything suitable.

-Ken

Title: Re: Fully Differential FC opamp SC CMFB
Post by Paul on Dec 12^th, 2003, 12:47am

I am using Mentor Tools and I especially like the extended control language of Eldo, but that's a personal point of view I don't want to argue about. True that MGC doesn't offer the kind of simulation you ask for.

What you can do, other than computing the equivalent resistor value and replacing it in the schematic, is run a transient simulation up to reaching equilibrium (all nodes have reached their correct operating point). From that point you can run AC or noise simulations (not sure whether you can use harmonic balance from Eldo RF, I'm only using "plain" Eldo).

I copied the following section from the Eldo documentation:
--------
AC in the middle of a .TRAN
Whenever a .AC command, a .TRAN command, and a .OP command containing time specifications are in the .cir file, then an AC analysis will be performed at each of these time points. In the following example Eldo will perform AC, NOISE and TRAN analyses at time 0, time 5n, and time 7n:
.AC DEC 10 1 1e9
.NOISE v(5) vin 70
.TRAN 1n 20n
.OP 5n 7n
--------
I never used PAC, PXF in SpectreRF so I cannot really compare. For my analog (not RF) designs, this works fine. Does this answer your question?

Paul

Title: Re: Fully Differential FC opamp SC CMFB
Post by Ken Kundert on Dec 12^th, 2003, 3:06pm

Paul,
The approach you describe does not really work for switched-capacitor circuits. Consider a traditional SC filter. Generally, at no point in time there is no direct path from the input to the output. So if you run an AC analysis during one clock phase, you get 0 gain. Then if you run another AC analysis during the other phase you again get 0 gain. This approach tells you that there is no signal transmission from input to output, which is clearly wrong.

-Ken

Title: Re: Fully Differential FC opamp SC CMFB
Post by Paul on Dec 14^th, 2003, 11:42pm

Ken,

you are totally right that this doesn't work for SC filters. I do use this kind of simulation for SC amplifiers where the CM takes some clock periods for converging to its operating point, but where the voltage gain is defined by a capacitor ratio, i.e. there is a AC direct path from input to output.

It depends on Roberts topology whether this kind of simulation is usable for him or not. As far as I understood his first post, he only considers the bare amplifier and the SC CMFB. In that case I believe he can use this kind of simulation. Am I wrong?

Paul

Title: Re: Fully Differential FC opamp SC CMFB
Post by ravi on Dec 24^th, 2003, 4:00am

Hi Ken and Paul,
Is it ok to use TPUNCH and TRANAC options in spice. I don't know if it is available with all versions of spice but many commercial versions may be having.
TPUNCH punches the operating point at the specified time and TRANAC does the AC analysis at that operating point.
I don't know how far is it accurate. Anybody who know more about this can comment on this.

regards
Ravi

Title: Re: Fully Differential FC opamp SC CMFB
Post by Ken Kundert on Dec 24^th, 2003, 9:23am

Ravi,
I've never heard of TPUNCH and TRANAC, but if I understood your explanation, the transient and AC analyses in Spectre can be configured to do the same thing. I occasionally use such an analysis approach to provide some insight as to how the circuit is behaving, but I am always somewhat dubious of the results as this approach violates the basic assumptions behind operating points and AC analysis.

-Ken

Title: Re: Fully Differential FC opamp SC CMFB
Post by Stan on Dec 30^th, 2003, 6:51am

Ken and Paul,

Could you give some more detail on how to simulate the open loop gain and frequency responce of fully differential OPAMP with SC CMFB? If I replace the capacitors with equivalent resistors, will the gain and output swing be affected?
Thank you.

Stan

Title: Re: Fully Differential FC opamp SC CMFB
Post by The_One on Jan 1^st, 2004, 12:32am

Hi Stan

You can use the Spectre RF simulator....the analysis is called PSS and PAC analysis.

By the way, do you know how to design the SC CMFB? any general guidelines, eg. how to choose the correct value for the capacitors? thx

Title: Re: Fully Differential FC opamp SC CMFB
Post by Paul on Jan 1^st, 2004, 3:01am

Hi Stan,

you should have a look at the following post:
http://www.designers-guide.com/Forum/?board=analog_design;action=display;num=1068014457

If you use Cadence tools, as the One mentions, you don't need to replace the SC by equivalent resistors. Have a look at Ken's answer above. If you don't use Spectre, equivalent resistors may solve your simulation problem (also discussed above). If the capacitor values are chosen correctly, they will not considerably degrade your amplifier performance.

The One, I already answered this question in your other post:
http://www.designers-guide.com/Forum/?board=analog_design;action=display;num=1071817367
Maybe it is not clear?

Paul

Title: Re: Fully Differential FC opamp SC CMFB
Post by Shano on Jun 7^th, 2007, 10:20pm

Hello all experienced guys:

I am new here and new about SC_CMFB design.
I know this post is three years ago!!
I have some thoughts on this topic,
Could anyone comment on it and enlighten me

Robert said "designing a fully differential folded cascode (FC) opamp for switched capacitor (SC) circuits". The OPamp is used for discrete-time
applications. The OP gain which he concerned about is the gain in some phase rather than an averaged gain. If we use a resistor to replace the switched-capacitor or just do PSS/PAC simulation, we will get a lower gain.
In this case, we just do a transient simulations followed by AC simulation.
Then, we will get what we want. Is this right?

Shano