Bob Cordell Interview: Negative Feedback

Re: Re: Re: Re: Re: Re: Your Designs????

mikeks said:



Where, Rodolfo?

Where?


I will speak only for myself, and with a single example. Others may take care of their own controversies.

When I posted a brief analysis of normal negative feedback and a general form of EC, you immediately challenged my inclusion of an attenuator block 1/A'. Despite several subsequent posts where I tried to make clear this only added generality, you persisted in deleting it from further analysis.

Please, I do not want by any means to fire up now a discussion on past issues, I do not think it could be of interest, worse still, it will be of no use if you only persist on an deffensive attitude to prove yourself right by all means.

Rodolfo


Rodolfo
 
ingrast said:
When I posted a brief analysis of normal negative feedback and a general form of EC, you immediately challenged my inclusion of an attenuator block 1/A'.
Rodolfo

Rodolfo, the point I made was that block was not essential to the principle, something you agreed with.

In fact, i actually demonstrated that it was a source of confusion rather than enlightenment.
 
mikeks said:


Rodolfo, the point I made was that block was not essential to the principle, something you agreed with.

In fact, i actually demonstrated that it was a source of confusion rather than enlightenment.


As said Mike, I do not want to fire up a discussion on past issues so this will be my last reference to this particular issue.

One thing is to say certain part is not essential to the principles, and something very different is to say it is useless.

The 1/A' block sets the ideal EC gain to *exactly A'*, while deleting it sets said gain to unity. This last may be OK for some cases, but certainly implies a loss of generality (and usefulness).

I hardly can accept this is a source of confussion.

Rodolfo
 
Let's go back to feedback !

I have a question.

Let 's assume that you are a diyer and you need amplifiers for active speakers i.e. one amplifier per speaker.
Then the load of each speaker can be measured.

Why not use amplifiers with two pole compensation. This will give a lot of open loop gain in the audio band of interest ( so very high desensitivity).

The drawback of two pole compensation is its intolerance to parasitic poles, capacitor load and overload.

In this dedicated application, it is possible to avoid any clipping by design with enough headroom.
Because the load is well known, why not use dedicated passive correction networks ( zobel networks) at the output of each amp to make the load resistive ( like explained by Leach in his AES papers). Then a well designed two pole compensation could be very effective and would decrease a lot high frequency distortion products. Because transient thermal effects are generators of high frequency crosseover distortion bursts, this compensation method could help a lot.

Of course this is not a solution for a general purpose amplifier

What do you think?
 
JPV said:
Let's go back to feedback !

I have a question.

Let 's assume that you are a diyer and you need amplifiers for active speakers i.e. one amplifier per speaker.
Then the load of each speaker can be measured.

Why not use amplifiers with two pole compensation. This will give a lot of open loop gain in the audio band of interest ( so very high desensitivity).

The drawback of two pole compensation is its intolerance to parasitic poles, capacitor load and overload.

In this dedicated application, it is possible to avoid any clipping by design with enough headroom.
Because the load is well known, why not use dedicated passive correction networks ( zobel networks) at the output of each amp to make the load resistive ( like explained by Leach in his AES papers). Then a well designed two pole compensation could be very effective and would decrease a lot high frequency distortion products. Because transient thermal effects are generators of high frequency crosseover distortion bursts, this compensation method could help a lot.

Of course this is not a solution for a general purpose amplifier

What do you think?


You are absolutely right. In the amplifiers I designed for my Athena active loudspeakers, I did not incorporate any protection circuits at all, and did not incorporate an LR output networlk. I did still employ a rather conventional Zobel. When you know exactly what speakers you are driving, and are not having to drive them through a crossover network, you can take advanatage of it.

Cheers,
Bob
 
Re: Re: Re: Re: Perspective on Open Loop Bandwidth Comparisons

andy_c said:


Very true. In fact, I remember we were discussing common-mode effects on the input stage a while back. I've been looking at a design that uses two-pole compensation, with about 60 dB of feedback at 20 kHz. I wanted to see the contribution of the input stage to the overall distortion of the amp at 20 kHz. I came up with a sim of the input stage, with VAS and output stage ideal (but with compensation to simulate the load of the VAS on the input stage, and an input stage imbalance due to the bias current of the VAS). I was pretty disappointed to find that despite the very small differential input voltage, the input stage contribution to overall distortion was of the same order as the VAS. I ended up trying the trick of using a bootstrapped cascode for the input stage. The results of the sim amazed me. I've attached the LTSpice projects of both to this message so you can see for yourself. Check out the simulated THD of both configurations. This seems to indicate that common-mode effects dominate the input stage distortion if the feedback is high enough. What amazed me was how much. I still don't know if this is an Early effect thing, a capacitance thing or what. Of course, one could rightfully say that the sim does not represent reality either - but I suspect it's going to be a lot closer than the Gilbert approach that neglects common mode effects entirely.

Sorry for the earlier confusion - the 318 thing threw me off.


Hi Andy,

I like this front end you've designed here, do you have any references for the bootstrapped cascode?

Pete B.
 
Re: Re: Re: Re: Re: Perspective on Open Loop Bandwidth Comparisons

PB2 said:



Hi Andy,

I like this front end you've designed here, do you have any references for the bootstrapped cascode?

Pete B.


Yes, the common mode effects in the input stage can be quite important, but they vary a lot depending on topology. For example, if you use a cascode on the input stage, but don't bootstrap it, but the load on the input stage has good common-mode rejection, that can work very well. Input stage loads where the output of the input stage is taken single-ended from only one side will be much more susceptible to such common-mode effects. These considerations are why I use the differential current source load on the input stage in my MOSFET amplifier. That load has a very high differential impedance, but a very low common-mode impedance.

Of course, some people circumvent the input stage common mode issue by going to an inverting topology where the input is a virtual ground, but this has other problems and disadvantages.

Cheers,
Bob
 
andy_c said:

PMA then asked me to provide some formulas to show this, but at the time I hadn't written it up. I ended up writing up some preliminary stuff for a web page, but by the time I did so, the conversation had drifted to other subjects.

I just uploaded that info to my web site. It can be found here. This page is not linked anywhere from my site, so you can only get to it directly with the above URL.


Hi Andy,

I am sorry to reply so late, I have been very busy in my professional projects last months.
Anyway, I have downloaded your new paper now. Thank you very much, it is written in a very clear way.

Regards,
Pavel
 
JPV [/i] [B] The drawback of two pole compensation is its intolerance to parasitic poles said:


Not if it is properly designed.

This is interesting!
How to avoid dynamic behaviour of poles if the amp goes close to clipping or is at slew limit, so kind of clips?
And how many zeros is it best to intentionally introduce? one? two?
'Properly designed' can mean anything but 'wrong designed'.
Elaborate please.

regards
Adam
 
mikeks said:


Good idea!



Not if it is properly designed.


darkfenriz said:



This is interesting!
How to avoid dynamic behaviour of poles if the amp goes close to clipping or is at slew limit, so kind of clips?
And how many zeros is it best to intentionally introduce? one? two?
'Properly designed' can mean anything but 'wrong designed'.
Elaborate please.

regards
Adam


A bit surprised by the comments

😕

What I was expecting are comments on my proposed 'good' design.

two pole compensation requires a good estimation of the dynamic of the signal to avoid any saturation. In the application mentionned it is easy to design the required headroom because the required performance is known ( max spl of speaker) .

two pole compensation is sensitive to extra poles introduced by the load or by the feedback network. The solution proposed is to avoid any load problem by making it resistive through the use of passive correction networks. The methodology to design these networks is well explained in Leach AES papers.
The loop beeing resistive, the only care there will be to avoid parasitic capacitance in the layout of the pcb.

So if it is well designed in this way, is it ok and is it worthwhile?

As I understand from andy-c, two-pole compensation will not improve the front end common mode distortion. My question is: is feedback able to reduce common mode generated distortion?

In my case, the objective is to reduce the transient crossover distortion.

Another question to M Cordell, Curl and Pass:

If, based on your experience, you would have to rank the different distortion mecanisms with respect to their influence on sound quality, what would be your ranking ( providing that the design was done with normal good engineering):
front end common mode distortion
front end differential mode distortion
vas distortion
static cross-over disrortion ( bias)
transient thermal crossover distortion

Thanks
Jean-Pierre
 
darkfenriz said:
And how many zeros is it best to intentionally introduce? one? two?

One forward path LHP zero and one feedback path zero.

JPV said:
....two pole compensation requires a good estimation of the dynamic of the signal to avoid any saturation......

JPV said:

.....two pole compensation is sensitive to extra poles introduced by the load or by the feedback network.

If the forward path zero is located at a sufficiently low frequency, then double-pole compensation, in the absence of the mandatory Thiele load-stabilizing network, will be just as stable (or not) as single-pole Miller compensation.
 
Hi Mikeks,
thanks for the links, the first has over 70pages on compensation!!!

I'll get around to reading them shortly, understanding may just take a little longer.

But, I would like to hear/see your take on compensation in general and the two pole version in particular.