NFB and Electron Propagation (from Blind Testing)

[mikek]

Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Re: Harold S. Black.

I seem to remember getting a mediocre grade in Control Systems,
but I did escape knowing what phase margin is. (post 37)

I see what you mean about discussion, since you now have launched into multiple post monologs.

You know what happened to the last one to utter "Dave, This conversation can serve no useful purpose."

okay...lets talk....as i recall, you mentioned an amplifier having several poles worth of phase shift, and i suggested that a stable amplifier cannot have such for stability.....i took it as given that below ft you can not have more than a single pole roll-off if the thing is to be unconditionally stable....beyond ft, such singularities as may exist a of vanishingly low relevance......unless of course you have inardvetently generated an RHP zero in your foward path, in which case it would have to be moved well beyond ft, or cancelled to prevent a 'bump' in your gain-freq. response that may, again result in instability...


Now.....i hope this helps...?

 

[SY]

Having never taken Control Systems, I'm somewhat baffled. Would not an amp with multiple poles in its open loop response be stable as long as the dominant pole was removed by a decade (pulling a ratio out of my nether regions) from the next?

 

[mikek]

Having never taken Control Systems, I'm somewhat baffled. Would not an amp with multiple poles in its open loop response be stable as long as the dominant pole was removed by a decade (pulling a ratio out of my nether regions) from the next?

No...you ideally want to maintain a single pole roll-off to the unity gain intercept...this gives you a 90 degree phase margin....however, in practice, you can quite reasonably get away with a 45 degree phase margin at the unity gain intercept, (ft).

 

[SY]

No...you ideally want to maintain a single pole roll-off to the unity gain intercept...this gives you a 90 degree phase margin

Well, yes, I understand that, but you've still got an amp with more than one pole that's stable. It's just that those higher frequency poles lie well above the intercept frequency. What am I missing?

 

[Steve Eddy]

...this constant equating of EMF to electromagnetic waves is really depressing....cant you tell the difference?

What's really depressing is how completely out of your depth you seem to be here.

I'm quite aware of the difference. What you seem oblivious to is that EMF, in and of itself as it relates to this issue is merely incidental. EMF simply initiates the process. What happens after that initiation is the propagation of electromagnetic waves down the length of the line with a finite velocity. And you're not going to see any of that initiating EMF manifest itself at the end of the line until sometime later when the electromagnetic waves have reached the end of the line.

se

 

[Steve Eddy]

...wrong again.........i am refering to EMF (not 'electromagnetic fields') setting up the electron drift, .......the converse is not true however.....elementary my dear Watson...

First, Holmes never said "Elementary my dear Watson." So let's just get that myth out of the way right now so we can move on to your myth.

Again, EMF is only the initiator. What carries that process from the beginning of the line to the end of the line is the propagation of electromagnetic waves.

Ok, let's get you your EMF. A simple battery will do. The battery's EMF maintains a potential difference between its terminals which we call voltage.

So take your EMF, a switch, and a pair of wires. We don't really need to bother with a load but let's include one just to make things a bit more more understandable perhaps. So add a resistor to the list.

Hook them up until so you get something that looks like this:

<center>
<img src="http://www.q-audio.com/images/emf.jpg">
</center>

Now get yourself two volt meters.

Connect one volt meter at the nodes A+ and A-. Connect the other volt meter at the nodes B+ and B-.

Now flip the switch.

You have been arguing that when the first volt meter detects the voltage at nodes A+ and A-, the second volt meter will instantaneously detect voltage at nodes B+ and B-.

Now don't start going on about your meaning EMF and not voltage. EMF manifests itself as a potential difference (i.e. voltage) between two points.

In any case, that is not what will happen.

The second volt meter will not detect any voltage until the electromagnetic wave, initiated by the EMF/battery, propagates down the line and reaches nodes B+/B-. This delay will be a function of the propagation velocity of the electromagnetic wave and the length of the line. And this delay will be non-zero for any non-zero length of line.

se

 

[phase_accurate]

I think the delay, caused by the finite propagation of waves through a line, is still negligible, compared to all the poles introduced by parasitic capacitances etc, within a multistage amplifier using overall feedback.
If we could manage to have only one pole and it was ridiculously high, then we shold start to take the wave propagation delay into consideration.
Just keep in mind that for a path-length of 20 cm, 90 degrees of phase shift occur at 375 MHz.

I do by no means want to imply that for our real world amplifiers short signal paths wouldn't be a good thing though.

Regards

Charles

 

[Fred Dieckmann]

No 9000 computer has ever made a mistake or distorted information.

No...you ideally want to maintain a single pole roll-off to the unity gain intercept...this gives you a 90 degree phase margin....however, in practice, you can quite reasonably get away with a 45 degree phase margin at the unity gain intercept, (ft).

Now you you admit that the closed loop response can roll off at greater than a first order rate and be stable. You even demonstrate familiarity with phase margin. Phase margins of even less than 45 degrees are fairly common and achieving 90 degrees of phase margin is relatively rare in an audio power amp.

This seems to be quite a reversal of your firm position in post #32 where you stated

...no fred, you cannot have several poles worth of phase shift in a stable linear system...that is a contradiction in terms....

Maybe someone did revisit their control systems text after all ..............

 

[mikek]

Re: No 9000 computer has ever made a mistake or distorted information.

Now you you admit that the closed loop response can roll off at greater than a first order rate and be stable. You even demonstrate familiarity with phase margin. Phase margins of even less than 45 degrees are fairly common and achieving 90 degrees of phase margin is relatively rare in an audio power amp.

This seems to be quite a reversal of your firm position in post #32 where you stated

....there you go again...selectively editing my post, so that you can deliberately misrepresent it......at what point exactly does your 'several poles worth of phase shift become one and a bit poles???

....need i say more? ok....You can never have a two pole roll-offf for a stable amp....there is no 'multi' about it .period..

 

[mikek]

What's really depressing is how completely out of your depth you seem to be here.

I'm quite aware of the difference. What you seem oblivious to is that EMF, in and of itself as it relates to this issue is merely incidental. EMF simply initiates the process. What happens after that initiation is the propagation of electromagnetic waves down the length of the line with a finite velocity. And you're not going to see any of that initiating EMF manifest itself at the end of the line until sometime later when the electromagnetic waves have reached the end of the line.

se

...an electromotive force...or potential difference established across a conductor of whatever provenance, will not 'set up' an 'electromagnetic wave'.....this can be found in any elementary physics text...etc...etc...which i reckon is where i am going to leave it.....


see definition of electromagnetic wave...and to use your beloved phrase, ' learn something' :

http://www.its.bldrdoc.gov/fs-1037/dir-013/_1944.htm

and....


http://scienceworld.wolfram.com/physics/ElectromagneticWave.html



.....surely even you 'get it' now old chap??

 

[mikek]

Well, yes, I understand that, but you've still got an amp with more than one pole that's stable. It's just that those higher frequency poles lie well above the intercept frequency. What am I missing?

...if you have the second pole, (i.e, the first non-dominant pole), precisely at ft, you'll have precisely 45 degrees phase margin....

....Please note: this does not mean you have a double pole roll-off before ft!!, (single-pole Miller compensation assumed of course)...which is the interpretation Fred will almost certainly have....

 

[phase_accurate]

this does not mean you have a double pole roll-off before ft!!...

So what do you think you've got with two-pole compensation ?
I do generally agree with you that the roll-off should be first order when crossing ft but it doesn't have to be that way for the whole region below ft.
Me personally, I prefer a first order roll-off as well though. But that does not have to be everybody else's taste.

Regards

Charles

 

[mikek]

So what do you think you've got with two-pole compensation ?
I do generally agree with you that the roll-off should be first order when crossing ft but it doesn't have to be that way for the whole region below ft.
Me personally, I prefer a first order roll-off as well though. But that does not have to be everybody else's taste.

Regards

Charles

You are absolutely correct, phase accurate,.......i routinely use two pole compensation....i took it as given that such a two-pole roll-off is understood not to be maintained up to, and beyond Ft.......

 

[Fred Dieckmann]

pole vaulting and jumping to conclusions

...if you have the second pole, (i.e, the first non-dominant pole), precisely at ft, you'll have precisely 45 degrees phase margin....

....Please note: this does not mean you have a double pole roll-off before ft!!, (single-pole Miller compensation assumed of course)...which is the interpretation Fred will almost certainly have....

Yes, you can get the contribution from the second pole roll off before the closed loop get becomes zero. That is why you can have a phase margin less even than 90 degrees. Zero phase margin occurs when the phase shift approaches 180 degrees while the closed loop gain is greater than one. Negative feedback then becomes positive feedback and you have an oscillator. You can still have a low phase margin and have stability, but you will get some ringing with respect to the transient response. Most audio power amplifiers do not have the luxury of achieving 90 degrees of phase margin. Design one, simulate one, and even measure one and you will see for yourself. You sure must have had an easy grader for Control Systems. I am pleased to see that your spelling has improved and that you now capitalize my name.

A few words of advice from the last one who thought he knew everything.

--"Well, I don't think there is any question about it. It can only be attributable to human error. This sort of thing has cropped up before and it has always been due to human error."

--"Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill and think things over."

 

[SY]

OK, Mike, I admit to being the world's worst writer and fairly incapable of understanding. So, to make up for my own deficiencies, I'll try asking the question again in such a way that only requires "yes-no" responses, which even a moron like me can understand.

1. An amp has two poles, one at 1 kHz, and another at 1 MHz. There are no zeroes added to correct those poles. When we close the loop, the intercept frequency on the Bode plot (is that ft? sorry, I don't know the jargon) is 10 KHz. Does that amp have two poles?

2. Is that amp stable?

3. An amp has two poles, a lowpass rolloff at 1 kHz and a high-pass rolloff at 20 Hz. That's two poles, if my fingers don't lie. Again, no zeroes added deus ex machina style. The intercept frequencies on the Bode plot when the loop is closed are 2 Hz and 20 kHz. Does this amp have two poles?

4. Is THIS amp stable?

 

[Fred Dieckmann]

Is THIS amp stable?

You will need to state the open loop DC gain and The closed loop DC gain as well to figure out the phase margin. The condition stable or unstable is a little ambiguous you can have an amp with very small phase margin that will not oscillate but have a very bad transient response. That why 45% phase margin is a target
and an absolute. You have a very valid question just state the gains and that should be enough information.

 

[mikek]

OK, Mike, I admit to being the world's worst writer and fairly incapable of understanding. So, to make up for my own deficiencies, I'll try asking the question again in such a way that only requires "yes-no" responses, which even a moron like me can understand.

1. An amp has two poles, one at 1 kHz, and another at 1 MHz. There are no zeroes added to correct those poles. When we close the loop, the intercept frequency on the Bode plot (is that ft? sorry, I don't know the jargon) is 10 KHz. Does that amp have two poles?

2. Is that amp stable?

3. An amp has two poles, a lowpass rolloff at 1 kHz and a high-pass rolloff at 20 Hz. That's two poles, if my fingers don't lie. Again, no zeroes added deus ex machina style. The intercept frequencies on the Bode plot when the loop is closed are 2 Hz and 20 kHz. Does this amp have two poles?

4. Is THIS amp stable?

1. yes...if by intercept frequency you mean unity-gain frequency....


2.yes.....

3. ....again.....if by 'intercept freq.' you mean the frequency at which the gain falls to unity, then you cannot have two such 'intercept' freqs., within the conditions stipulated by your question.

...i am assuming of course, that you have no inverted poles?
...see below for explanation...no magic involved....

http://ece-www.colorado.edu/~ecen2260/Bode/Bodenotes.pdf

and...

http://ece-www.colorado.edu/~ecen2260/Bode/Ch8.pdf

 

[Steve Eddy]

...an electromotive force...or potential difference established across a conductor of whatever provenance, will not 'set up' an 'electromagnetic wave'.....this can be found in any elementary physics text...etc...etc...which i reckon is where i am going to leave it.....


see definition of electromagnetic wave...and to use your beloved phrase, ' learn something'

Are you for real?

Yes, let's look at the definition of electromagnetic wave:

<i>A wave produced by the interaction of time-varying electric and magnetic fields.</i>

You can't establish a voltage across a conductor without producing time-varying electric and magnetic fields, unless the voltage across the conductor has ALWAYS existed and NEVER changes.

Going back to the previous example, before you flip the switch, the voltage across A+ and A- is zero. Which means that electric and magnetic fields are also zero. When you flip the switch, then the voltage goes from zero to something other than zero and the electric and magnetic fields in the wire also go from zero to something other than zero.

So now you have time-varying electric and magnetic fields which propagate down the line in the form of an electromagnetic wave.

If you're truly just being naive here, that's one thing. If you're being disingenuous and running around kicking people in the shins just to try and get a rise out of them, then I suggest you either stop or find some other board to post on.

se

 

[Pjotr]

Amazing to see that “experts” on control theory are quarrelling about 1 or 2 poles concerning loop stability. Dig up your old textbooks and read them over.

And Fred, Sy has given enough numbers to dig out the open loop DC gain your self, assuming Sy meant by intercept point the crossing of the bode plot at 0 dB. He can’t mean anything else because he has given only two pole frequencies.

It a pity that many interesting points in this weird discussion stay hidden. One to mention: What are the sonic implications of having flat open loop gain up to the end of the audio band, say 20 kHz? Is his necessary and why? This is along ongoing discussion. Up to now I have seen no consensus about.

 

[Fred Dieckmann]

Its not the wave OR the water

...an electromotive force...or potential difference established across a conductor of whatever provenance, will not 'set up' an 'electromagnetic wave'.....this can be found in any elementary physics text...etc...etc...which i reckon is where i am going to leave it.....


see definition of electromagnetic wave...and to use your beloved phrase, ' learn something'

Are you for real?

Yes, let's look at the definition of electromagnetic wave:

<i>A wave produced by the interaction of time-varying electric and magnetic fields.</i>

You can't establish a voltage across a conductor without producing time-varying electric and magnetic fields, unless the voltage across the conductor has ALWAYS existed and NEVER changes.

Going back to the previous example, before you flip the switch, the voltage across A+ and A- is zero. Which means that electric and magnetic fields are also zero. When you flip the switch, then the voltage goes from zero to something other than zero and the electric and magnetic fields in the wire also go from zero to something other than zero.

So now you have time-varying electric and magnetic fields which propagate down the line in the form of an electromagnetic wave.

If you're truly just being naive here, that's one thing. If you're being disingenuous and running around kicking people in the shins just to try and get a rise out of them, then I suggest you either stop or find some other board to post on.

se

Careful or we will he will get out the rubber hose to use on you with the water analogy...........

As my favorite cosmic cowboy Jimmie Dale Gilmore sang on SPINNING AROUND THE SUN :

"I felt like an endless ocean, rolling through the fog
Full emotion drifting like a weather beaten log
I even thought that I out-thought her
Till she said babe, you're just a wave, you're not the water"

Spinning round the fun,
Fred

P.S. your right Pjotr, I missed the intercept inclusion. An even more interesting question is the effect on small phase margins on dynamic stability with reactive speaker and cable loads which is a real live issue and not a theoretical construct.