Fold cascode OP AMP Design Problem(P.E.Allen)

[Joestar]

op amp follower problems

When i design a Folded Cascode OP AMP(Follow the step by the P.E.Allen 's book),The curve of Magnitude-frequence appear a peak at the vicinity of the second pole.Firstly I thought it maybe cause by the the right half plane zero,but I couldnot solute the issue after added the compensation capacitor and resistance.So i think the problem may be cause by the bias circuit of the cascode active load . who can interpret the bias circuit and the phenomena of the curve.

The attachment is the curve of Mag-Fre

 

[borodenkov]

folding the cascode

you forgot the attachment..

 

[carpa]

op amp design gain peaking

Check bias first. Is the output DC magnitude the same as you calculated?

 

[rambus_ddr]

p.e.allen

I think it may be the non-dominant pole is too close the unit gain frequency. You can increase this pole or decrease the the unit gain frequency.

 

[jodenma]

opamp pole

in spice add options:
.option unwrap

and run ,maybe you can resolve this problem.

 

[Joestar]

cascode opamp poles

you forgot the attachment..

Sorry, I didnot finish the upload owing to the file is too large

 

[Joestar]

p e allen 2001

Check bias first. Is the output DC magnitude the same as you calculated?

yeah it is right

 

[rambus_ddr]

folded cascode operational amp

I think it is not the wave of the non-dominant pole close to the unit gain frequency.

 

[cadb0y]

fold cascode op

Possible due to R/C isolation resistor you are using to break loop in DC/AC.

 

[electronrancher]

miller folded cascode opamp

your amp is perfect. i would say it is texbook implementation of a folder. zero placement looks just about perfect, only a tiny amount of gain peaking before rolloff.

the peak you asked about is a natural characteristic of the amp - there are many high-frequency poles and zeroes located near each other. avoid putting more zeroes into your system else you can peak the gain back over the 0db point, causing conditional stability.

 

[sutapanaki]

fold cascode bias circuit design

First, is there a second stage in this folded cascode opamp - the one that you use Miller compensation around it? If there is, it is known that a 2 stage folded cascode opamp can have a gain peaking around the unity gain frequency due to a pair of complex conjugate poles. You may want to check a paper by Milles Copland on that matter. It is an old one and I don't remember the title, but I think it is cited in P.Allen book.
If your opamp is not a 2 stage one you should be able to compensate it well just by the load capacitor. Run a small signal step responce to see if it oscillates. Run also a large signal step responce to see if there are additional instability problems.

 

[jodenma]

fold cascode bias circuit

i have another question: for the almost OPA,the first stage gain should be ? 20db~40db?if the gain is greater than 40db will the result be poor?

 

[Joestar]

two-stage folded cascode operational amplifier

First, is there a second stage in this folded cascode opamp - the one that you use Miller compensation around it? If there is, it is known that a 2 stage folded cascode opamp can have a gain peaking around the unity gain frequency due to a pair of complex conjugate poles. You may want to check a paper by Milles Copland on that matter. It is an old one and I don't remember the title, but I think it is cited in P.Allen book.
If your opamp is not a 2 stage one you should be able to compensate it well just by the load capacitor. Run a small signal step responce to see if it oscillates. Run also a large signal step responce to see if there are additional instability problems.

You are right .The OPAMP is two stage folded cascode OPAMP,the second one's output is connected to it's input for Miller conpensation,How could i set about to eliminate the phenomenon

 

[sutapanaki]

allen op amp

First, make sure that you have the phenomena. I'm not sure that you have it looking at your ac responce - your gain indeed picks up at the end, but then stays flat at this high frequencies, which looks fishy for me - no circuit can have flat responce at high frequencies. But then on the other hand the phase changes with more than 180 degrees, which suggest some sourse of resonance. What I'd do is do the ac simulation for larger frequency range to see if this flat portion is still there. If it is that may suggest something wrong with the models. Then, find the paper I mentioened before and see what they say there. It is not simple to explain in few words.

 

[fehler]

fold cascode + bias

When i design a Folded Cascode OP AMP(Follow the step by the P.E.Allen 's book),The curve of Magnitude-frequence appear a peak at the vicinity of the second pole.Firstly I thought it maybe cause by the the right half plane zero,but I couldnot solute the issue after added the compensation capacitor and resistance.So i think the problem may be cause by the bias circuit of the cascode active load . who can interpret the bias circuit and the phenomena of the curve.

The attachment is the curve of Mag-Fre

Hi, is your opamp single-ended? It seems that there are a
complex-pole-complex-zero-pair in your design. I dont think it will exist in
fully differential design. So, if you are single ended, please check with a
smaller load cascode transistor L, and to verify the simulation results.

If your design is fully differential, please add isolating caps in bias, and
try to find the source of this complex-pole-complex-zero-pair.

 

[segabird]

cascode load

When i design a Folded Cascode OP AMP(Follow the step by the P.E.Allen 's book),The curve of Magnitude-frequence appear a peak at the vicinity of the second pole.Firstly I thought it maybe cause by the the right half plane zero,but I couldnot solute the issue after added the compensation capacitor and resistance.So i think the problem may be cause by the bias circuit of the cascode active load . who can interpret the bias circuit and the phenomena of the curve.

The attachment is the curve of Mag-Fre

Hi, is your opamp single-ended? It seems that there are a
complex-pole-complex-zero-pair in your design. I dont think it will exist in
fully differential design. So, if you are single ended, please check with a
smaller load cascode transistor L, and to verify the simulation results.

If your design is fully differential, please add isolating caps in bias, and
try to find the source of this complex-pole-complex-zero-pair.

How can I find the source of this complex-pole-complex-zero-pair in the single-end circuit? or in second stage full-differential op?

 

[sutapanaki]

hspice +unwrap spectre

do a pole-zero analysis. depending on the software you use, it should have this option. Eldo has it, I think hspice has it, very probable that spectre has it too.

 

[segabird]

op amp active cascode

do a pole-zero analysis. depending on the software you use, it should have this option. Eldo has it, I think hspice has it, very probable that spectre has it too.

yes,
the software can list the frequency of the pole.
But the circuit node corresponding to the pole will not be listed in the log file.
So it is different for us to finding the parasitic poles.

 

[sutapanaki]

fold-cascode

This is normal. Only some of the first order poles can explicitly be associated with real circuit nodes. And zeroes are not associated with nodes at all. Also, complex conjugate poles are not associated with nodes. These are combinations of conductances and capacitances. I believe your purpose was yo see if there were pairs of complex poles and zeroes that could result in your gain peaking and then staying flat.

 

[fehler]

fold cascode op + output stage

[quote="segabird
How can I find the source of this complex-pole-complex-zero-pair in the single-end circuit? or in second stage full-differential op?[/quote]

hello, this pair exists in the cascode load of your single ended opamp. But,
if your opamp is fully differential, there will not be such a pair. In my opinion,
if the cascode transistors L the larger, the more possible that you have such
a pair. The manner is just like the mirror pole described by Razavi's book.
The difference is that a pole gives a mirror zero. A complex pole gives a complex
zero. You can derive it.

By the way, I dont think that Miller compensation will create such a pair.