Noise cancellation in two-stage Opamp with CMFB

[analogman]

two stage rolloff circuit

Many books say thermal noise in two-stage Opamp is dominated by
that in first-stage. Because that in sedond-stage is divided by
first-stage gain and second-stage gain.

And I did noise calcalation of Two-stage Opamp with CMFB by
using simulator and hand-calculation. But these results have
big differences in value. But I'm not able to understand why.

Do I need to consider contribution of CMFB?
I use some registers and transistors for CMFB.
I want some advise and suggestion.

Sorry for my poor english... Thank you!!

 

[noiseless]

Noise in Two-stage Opamp

don't understand your meaning "difference in value". You mean the books are wrong, or what else

 

[ocarnu]

Re: Noise in Two-stage Opamp

Registers in CMFB? How? Usually the noise of CMFB appears as, well, common mode, so it shouldn't bother you in hand calculations. How did you computed by hand, using Af, Kf parameters from models? Did you use the same formulas as the the simulator does? Hspice has more options and models for noise, you have to specify the parametres in the models AS WELL AS the formulas. Check any spice manual, noise section.
And start with the small simple circuit and see if the computed and simulated values are close.

 

[analogman]

Re: Noise in Two-stage Opamp

>noiseless, ocarnu
Thank you for replying!!

In Razabi's book (Design of Analog CMOS ICs), thermal noise in two-Stage
Opamp is described as eq.(9.52). My two-stage opamp is almost the same as
the circuit (figure 9.65). My first stage is terescopic, but thermal noise can be
calculated in a similar manner with eq.(9.52).
But simulation result indicates 1.5 times value as hand caluclation as.

I used registers in CMFB to generate the common-mode voltage of the
output signals.

I refered following pdf to design.
**broken link removed**

And I consider only thermal noise for noise calculation(in similar with eq.(9.52))
So Af, Kf parameters set zero in my models. And simulation results(Spectre) indicate
only thermal noise contribution.

I want more adivise or suggestion.
Thank you !!

 

[noiseless]

Noise in Two-stage Opamp

I cannot access the pdf file. would you upload it?

 

[FOX]

Re: Noise in Two-stage Opamp

I see fine the file pdf "Telescopic_ppt" with acrobat reader 4.

 

[noiseless]

Re: Noise in Two-stage Opamp

1. double check the simulaiton gm is the same as your hand calculation.

2. consider the effective length is not the specified as supposed

3. on the ppt you attached, what confused is why using open loop to simulate settling time? The result is wrong ! the open loop w3dB is very small, which means the time constant is very large. How can the settling time be 50ns!

 

[analogman]

Re: Noise in Two-stage Opamp

Thank you , noiseless
I considered your suggestions carefully.

1. I checked the simulation gm was the same as my handcalculation.
Next I checked simulation model , and I found thermal noise model
was different from my hand calculation. I should have considered
not only gm but also gds, gmb.

Consequently, difference between simulation and hand calculation
was smaller.
(OPamp noisepower : simulation 116.6a , hand calculation 78a→100a)
But some difference remains.

2. You mean BSIM3 thermal noise model ? I use modified SPICE2
thermal noise model. So the effective length isn't specified, I think.

3. I think open loop w3db is very small (3-4kHz) too. So settling time
can be 250us !! I did'nt realize... thank you !
And I don't know how I simulate fully-differential opamp with other
than this ppt describing...

If anyone knows about above ideas, suggests me please

 

[willyboy19]

Re: Noise in Two-stage Opamp

You guys all miss the most important point of your discrepancy between noise simulation and calculation: the contribution from the second stage does become significant at high frequency when the gain of the first stage roll off to 0dB! It is actually caused by the zero caused by the compensation cap and lead-compensation resistor. In other words, while it is still true that the noise contribution from the second-stage does get reduced by the first stage gain when referred at the input, you need to consider the roll-off of the first stage gain at mid-high frequencies.

You can find this yourself if you apply an AC source at the input of your second stage and plot the corresponding input and output noise.

 

[noiseless]

Noise in Two-stage Opamp

analogman, did you give me a helped point? just kidding

 

[analogman]

Re: Noise in Two-stage Opamp

Noiseless, do you mean I should give(donate) you a helped points?

I didn't know that rule. I'm sorry.
And I don't know how many points I should give.
So I give you a few points.

Thank you.

 

[noiseless]

Noise in Two-stage Opamp

analogman,

I am just kidding. Appreciate any way.

Are you Fudan graduate student?

 

[analogman]

Re: Noise in Two-stage Opamp

Noiseless,

I thank for your suggestion.

I'm not Fudan graduate student, I'm a Japanese student.
But Fudan's papers teach me much practical knowledge
I can't learn by some books.