how to determine pole & zero in the folded-cascode amp

[020170]

folded cascode pole

this is folded cascode amp using latch circuit

**broken link removed**

I have many wonder about this circuit

1. how to determin zero and pole?

2. parastic capacitances are so complex, but I cannot ignore it because they make poles. in designing amp, Would I ignore them?

3. this is the location of pole and zero

**broken link removed**

it is stange... pole and zero cancel each other. but -92.5561x is not canceling each other. yes this is 3dB frequency.

why did pole and zero cancel each other?

why did not -92.5561x cancel with corresponding zero?[/img]

 

[MLR67]

easy way to determine zeros and poles

Hello

1>Why do you have this positive feedback with transistors M7 and M9 on the current source ?
2>Why are M15 and M18 diode connected on your design ?

 

[Vamsi Mocherla]

determining amp poles

I think that it is perfectly normal for the poles and zeroes to cancel each other because they form the roots of the denominator and the numerator of the polynomial. I am sorry, but can you be more specific on the question because I am not sure what you are asking for?

 

[020170]

only r and c complex pole

is it normal that pole and zero cancel each other?

but, why did not 3db frequecy cancel that coressponding frequency?

when designing folded cascode amp, it is important to determine pole and zero location as amp is needed to operate in high frequency region.

so I want to know that how to determine pole and zero in detail not intuitive.

that is, my question is like that.

"how to determine transfer function of amp?"

simple amp model is properly used to determine the transfer function. but

it is not accurate. I cannot determine the transfer function by using this method.

in folded cascode amp, pole and zero location is so complex.

how to determine transfer function of folded cascolde amp?

without knowing exact transfer function, designing folded cascode amp is impossible.

thanks.

 

[jliang]

determine poles complex

I think you can do w = 1/RC for input, output and all the mid points to find the poles. R = looking in resistance at that point, same for C.

Corret me if I am wrong. I would like to know too, and especially zero's since they don't care too much about zero in the courses I have taken so far except for PM

 

[020170]

folded cascode pole zero analysis

the equation 1/(RC * 2*pi) is just approximation method. as the circuit is complexed, this method is useless. only simple amp is applied by the equation

In zero, the problem is too difficult.

I have to know location of zero. because zero is important to adjust freququency compensation. in more complex circuit, zero is property which determine phase margin.

when I calculated pole frequency in folded cascode , I made a conclusion like that.

although I consider all parasitic capatitance included in circuit, I got to wrong pole frequency. in my opinion, the pole location did not depend on the R and C on the only one node. so the signal path must be considered.

 

[pseudockb]

cmos + folded cascode + small signal model

By looking at the zeroes and poles location given by 020170, there is no zero at 92.5561x. Therefore the pole at 92.5561x will not be cancelled away. If you really want to know the exact expression of all the zeroes and poles (including the parasitic ones), I think the only way is to perform the brute force method by drawing the small-signal equivalent circuit of the half circuit and analyse from there. But that is a very tedious job.

 

[noiseless]

poles and zero + how to determine

first, I know it is peak to peak input OTA. what i do not understand is that why the positive feedback at the p-side?

As to the pole-zero cancelling, what i suggest is using small signal model to do the analysis.

 

[ezt]

poles in a folded cascode

Hi.
First, I think your circuit won't work the same for signals close to Vdd and to Ground, for you have used positive-feedback for PMOS loads but that's not the same for NMOS loads.
As I see, all poles and zeroes of your circuit are real and you have no complex ones. And you know, that's good! About poles and zeroes with the same frequency, I think they are computational poles and zeroes. Believe me, they always exist in all circuits, so I really insist on my opinion about these poles and zeroes to be computational. Don't worry they have nothing to do with your circuit or freq response. It seems all poles and zeroes of your circuit probably before fu, excluding 92MHz pole, are almost at the same frequency and hence they compensate each other. So I guess the freq response of the whole circuit should be very close to single-pole response. And that was predictable, for simple folded-cascode opamps have only one dominant pole before fu (unity gain freq). And one other thing. As I know positive-feedback helps the frequency response of circuit to become closer to its single-pole counterpart.

Regards,
EZT

 

[arsenal]

folded cascode frequency response

first, u use postive feedback,therefore you have to take care of the gm of m6 and m7, otherwise it wont work.
2ndly,why u use diode connected nmos as loading(m15)? it will decrease the ro

pls explain it,thanks