Two Stage Op Amp Compensation

[tia_design]

I designed a low noise Op Amp. The compensation scheme I choose is cascoded compensation (Ahuja style). The problem is the second pole is always less than unit gain bandwidth.
Since this is low noise op amp, so the gm(transconductance) of input pair is very big. I have to increase the gm of second stage to increase the second (non-dominant)pole, but still this pole is less than UGW. I tried the different value of compensation capacitor and load capacitor, but it doesn't work.

Any idea is appreciate.

 

[gszczesz]

Ahuja wrote his paper describing the compensation technique around the second stage, but you shouldn't be restricting yourself to wrapping the feedback around the second stage. Can you put the feedback right into the first stage?

Your problem obviously is that you are designing an LNA. The common-gate amplifier introduced with the Ahuja technique increases NF. Wrapping it back to the input will be costly in terms of noise. You may be forced not to use Ahuja and use a simple output-input feedback compensation because it is less noise.

Greg

 

[hitlgc]

I tried the different value of compensation capacitor and load capacitor

what is the relationship between compensation capacitor and load capacitor?

 

[jiangwp]

Can you deduce the first stage input transistor and load transistor length.

If so, maybe , you can try it .

The same principle is fit for the second stage.

Maybe , the middle of gnd compenstion cascode transitor is needed to modify.

 

[layes2]

do not cancel Lhp pole just move it to infinite

 

[opamp741]

Try this method, [I use this for miller compensated 2 stage amps]

Problem : find out optimum gm1 and gm2 along with Cc for a given close loop freq and phase margin requirement:

Solution :
gm1 = UGB * Cc---------------(1)
gm2 = Kx * UGB * (C1+C2+C1.C2/Cc) ---------------(2)

[Kx = P2/UGB : we use Kx=2 for phase margin of 70 degree : u can also do simple simulation of 2 pole system to find out Kx for ur PM requirement]

model C1 C2 as K1*gm1+K2*gm2+K3 and put it in eqn 2

So now u hav 2 equations with 3 variables. Plot gm1, gm2 and gm1+gm2 as a function of Cc. The point where u get min(gm1+gm2) is ur optimum point.

see the attached for an example..