The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> No Peaking with Transient Response with PM of 66 degre
https://designers-guide.org/forum/YaBB.pl?num=1335730561

Message started by Ricky Chen on Apr 29^th, 2012, 1:15pm

Title: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 29^th, 2012, 1:15pm

I designed a 2-stage Miller compensated CLASS-AB OTA.
The exact schematic is attached.
I found the differential mode Phase margin using Middlebrook method.
I got a PM of 67 degree. The AC response is attached.
When I gave a step input I am NOT seeing any peaking at the 1st OTA output. Actually I should see a slight peaking with 67 PM.
Can anybody tell why there is NO peaking at the output.
The transient response is also attched

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 29^th, 2012, 1:16pm

The AC response

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 29^th, 2012, 1:18pm

The Transient response attached

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by buddypoor on Apr 29^th, 2012, 11:08pm

Hi Ricky,

1.) I suppose the shown ac magnitude respose is the loop gain, correct?
Why didn't you show the phase response?
2.) Are you sure that the loop gain was simulated correctly? According to the slope of the magnitude response at unity gain the phase margin should me much lower (perhaps 45-50 deg).
3.) Does the step response belong to the closed loop circuit?

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 29^th, 2012, 11:22pm

buddypoor wrote on Apr 29^th, 2012, 11:08pm:

1) I have shown the phase response. Just Scroll down the .jpeg file

2)I am assuring you it is correct. I have checked with 2 methods of Middlebrok and both gave the same results

3)Yes it is for the closed loop circuit ( The schematic is the entire circuit and the o/p is take at the o/p of the 1st OTA)

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 29^th, 2012, 11:36pm

Ricky Chen wrote on Apr 29^th, 2012, 11:22pm:

buddypoor wrote on Apr 29^th, 2012, 11:08pm:

Sorry the LOOP_GAIN plot has not same axis for both PHASE and Gain so I have attached the same with same axis in GAIN and PHASE

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by raja.cedt on Apr 30^th, 2012, 1:53am

@buddy: hello buddy..i couldn't understand this "According to the slope of the magnitude response at unity gain the phase margin should me much lower". could you please explain how to estimate phase margin from the slope?

@ ricky: in theory you get small overshoot (around 3%), but when you implement with transistors you could expect slight change in the gm or some small signal parameter, and may be your overdrive of the input pair lower and with 30mv swing it may change some %ge. So try to apply very small step like micro volts or increase the transistor overdrive.
From your step response i can imagine there is a RHZ..but why it is not in magnitude plot?
Thanks,
raj.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by buddypoor on Apr 30^th, 2012, 3:13am

@ricky: Sorry, I didn't see the phase (forgot to scroll). Please, can you confirm that you have used the "double-injection technique" ? (Because the term "Middlebrook" is somewhat ambigious).

@raja: In most cases (in particular for opamps without transfer function zeros and with a continuous gain decrease) the phase response is directly related to the gain response and, thus, must not be shown explicitely. In these cases, the phase margin can be estimated using the SLOPE of the gain function only. In particular, if the 2nd pole is approx. at the unity gain frequency (slope changes from -20 to -40 dB/dec) the PM will be approx. 40...50 deg.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 3:44am

buddypoor wrote on Apr 30^th, 2012, 3:13am:

1:- Middle-Brook theorem is NOT ambiguous at all.

2:- Yes it is 'double injection technique' i.e. I found RRi(injecting current) and RRv( injecting voltage) at the break point.
This is clear from the schematic I have shown.

3:- From gain plot it is very difficult to figure the PHASE AMrgin.
As it may have pole zero pairs.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 3:48am

raja.cedt wrote on Apr 30^th, 2012, 1:53am:

1:- Why you imagine RHZ from transient response ? Can you tell the reason behind it ?
2:- Do you expect a peaking in Magnitude plot to reflect RHZ ?

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 3:57am

buddypoor wrote on Apr 30^th, 2012, 3:13am:

Dear Buddy,

Why you are deriving that the 2nd pole is near UGB of the Loop-Gain ?
If you carefully look at the Gain plot, the 2nd pole is around 1.38 GHz which is 2.3 times than the Unity LOOP Gain frequency 613MHz.
I guess 2.3 times can easily give you 66 degree PM.
Please NOTE the axis is Logarithmic !!

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 4:14am

[quote author=raja.cedt link=1335730561/0#6 date=1335775995]@buddy: hello buddy..i couldn't understand this

@ ricky: in theory you get small overshoot (around 3%), but when you implement with transistors you could expect slight change in the gm or some small signal parameter, and may be your overdrive of the input pair lower and with 30mv swing it may change some %ge. So try to apply very small step like micro volts or increase the transistor overdrive.

I actually did with 1uV input. No peaking is SHOWN.
I wonder why you tell that transistor is overdriven.
Pleas NOTE:- The input is being applied at the input-R of the OTA not at the OTA input. ( please see the schematic). Also the OTA's o/p stage is a CLASS-AB so slew is NOT that a problem and feed-back makes the input of the OTA stand-still(virtual Ground).

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by raja.cedt on Apr 30^th, 2012, 4:17am

hello ricky,
when you have RHZ step response starts in opposite direction. As you can see in your step response first started in -ve direction and it entered into +ve direction.

Thanks,
raj.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by buddypoor on Apr 30^th, 2012, 6:13am

1:- Middle-Brook theorem is NOT ambiguous at all.

I wrote that the TERM "Middlebrook" is not sufficient because his theorem (The general feedback theorem, GFT) consists of several parts (some with simplifications).

2:- Yes it is 'double injection technique' i.e. I found RRi(injecting current) and RRv( injecting voltage) at the break point.
This is clear from the schematic I have shown.

But it was not clear from the drawing - more than that, I must apologize for not beeing familiar with your simulation program (incl. schematic entry).

3:- From gain plot it is very difficult to figure the PHASE AMrgin.
As it may have pole zero pairs.

May answer to raja was a general one and not specifically related to your drawing.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by buddypoor on Apr 30^th, 2012, 6:16am

Why you are deriving that the 2nd pole is near UGB of the Loop-Gain ?
If you carefully look at the Gain plot, the 2nd pole is around 1.38 GHz which is 2.3 times than the Unity LOOP Gain frequency 613MHz.
I guess 2.3 times can easily give you 66 degree PM.
Please NOTE the axis is Logarithmic !!

As mentioned before, my statement regarding the UGB was part of my answer to raja and not in response to your drawing.

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 7:29am

raja.cedt wrote on Apr 30^th, 2012, 4:17am:

hello ricky,
when you have RHZ step response starts in opposite direction. As you can see in your step response first started in -ve direction and it entered into +ve direction.

Thanks,
raj.

Dear Raj,

Thanks a lot for pointing the very important point.
In the schematic there is a 8pF cap between in and OUT of the OTA.
So this is probably creating a feed-forward path and hence RHZ.
When I removed this cap I didn't get any reverse direction movement in the transient.

Again thanks a lot for pointing this.

Kind Regards,
Ricky

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by Ricky Chen on Apr 30^th, 2012, 7:34am

buddypoor wrote on Apr 29^th, 2012, 11:08pm:

Hi Ricky,

2.) Are you sure that the loop gain was simulated correctly? According to the slope of the magnitude response at unity gain the phase margin should me much lower (perhaps 45-50 deg).

You have mentioned here that the PM shouldn't be much higher from 45 degree. I guess you have thought the 2nd pole is very close to the Unit loop gain.
This actually baffled and scared me about my design and simulation method

Title: Re: No Peaking with Transient Response with PM of 66 degre
Post by raja.cedt on Apr 30^th, 2012, 8:54am

hello rycky,
i forgot to mention some thing..when you have 67deg phase margin step response slope at t=0 will be lesser than first order and in theory you get 0. But your slope is very, for me it gives first order feeling.

Thanks,
Raj.