Feedback Delay Time

Hi everybody ! Just a thought about NFB! The signal takes some time from the input to the output of a amplifier ,and then taken from the output to the input, with te mission to correct the linearity of this signal, but because of the delay it has (maybe a few microseconds) than the efect could be worse than without NFB? I much prefer the sound without feedback. Maybe because i dont use lots of watts....
I would like to know what do think about this....
Thanks
Silvino
 
Last edited by a moderator:
Hi everybody ! Just a thought about NFB! The signal takes some time from the input to the output of a amplifier ,and then taken from the output to the input, with te mission to correct the linearity of this signal, but because of the delay it has (maybe a few microseconds) than the efect could be worse than without NFB? I much prefer the sound without feedback. Maybe because i dont use lots of watts....
I would like to know what do think about this....
Thanks
Silvino

Or it is a silly question ?
 

20to20

Member
2010-06-23 9:25 pm
Or it is a silly question ?

Think about the speed of light when you try to imagine any ''delay." The only real concern is of phase shift if the signal is passed through stages that create phase shifting. Otherwise, there is no delay in the signal back to the input. A few inches of wire is no match for the speed of light compared to the frequency of an audio wave.
 

20to20

Member
2010-06-23 9:25 pm
Okay, but from the RCA socket to the OPT the signal passes through at least 2 or 3 tubes, and it takes some time to make that way, or not?

Electricity travels at the speed of light. The audio signal is a slow wave on a river of electrons. The wave form in the first tube is at the output transformer and back to the first tube again while the wave is still in the first tube. It's all electrical pressure that is equal throughout the circuit if phase shifting is not present.
 

20to20

Member
2010-06-23 9:25 pm
The worst offender is the output transformer itself.

I wouldn't call the OPT an ''offender'' because without the properties a transformer provides we couldn't do want we want with electricity. It provides us with an inverted phase so we can tap it for the negative feedback if we want to. Very convienient place to grab all the distortion produced and an inverted phase, too.
 

Arnulf

Member
2009-02-02 9:41 am
I wouldn't call the OPT an ''offender'' because without the properties a transformer provides we couldn't do want we want with electricity. It provides us with an inverted phase so we can tap it for the negative feedback if we want to. Very convienient place to grab all the distortion produced and an inverted phase, too.

I believe what Yvesm was trying to say is that it is usually the OPT in the entire tube amplifier chain that brings by far the most significant portion of phase distortions to the signal, thereby delaying certain part of the audio spectrum, possibly resulting in audible delay (such as oscillations :D ).
 

20to20

Member
2010-06-23 9:25 pm
I believe what Yvesm was trying to say is that it is usually the OPT in the entire tube amplifier chain that brings by far the most significant portion of phase distortions to the signal, thereby delaying certain part of the audio spectrum, possibly resulting in audible delay (such as oscillations :D ).

Well, when you consider that most decent OPTs have a <1% distortion rating and look at the data for power tubes with 2-10% the real offender is the tube. If you consider any phase shift by the OPT a distortion then I'd call that nitpicking because our ears and brain don't care what phase the signal is in compared to the input as long as both channels are equal, if it's stereo. The OPT phase shift in and of itself, is meaningless.
 

Arnulf

Member
2009-02-02 9:41 am
Well, when you consider that most decent OPTs have a <1% distortion rating and look at the data for power tubes with 2-10% the real offender is the tube. If you consider any phase shift by the OPT a distortion then I'd call that nitpicking because our ears and brain don't care what phase the signal is in compared to the input as long as both channels are equal, if it's stereo. The OPT phase shift in and of itself, is meaningless.

I think you're thinking of a different thing here: this is not about the whole audio band and not about distortion in general; it's about the phase shifts at the bottom of the band that can result in oscillations when NFB is applied around the amplifier (with enough phase shift it turns into PFB). These shifts accumulate over AC-coupled stages and OPT contributes some shift as well. I believe this is what Yvesm was trying to say (and my understanding of the original question on this thread was the same - namely "is there enough delay in the signal to cause problems when feedback is used ?").

RDH4 goes pretty deep into this subject, with few examples with actual figures for easier understanding, you might want to check it out if you're interested.
 
It seems to me that when you create an amplifier where the delay in the feedback path is on the same order of magnitude as the rise time of the expected signal, then you have a problem.

Also, you don't have to worry so much about signal reflections and impedance matching in audio since rise times are much slower than it takes for the signal to propagate.
 
I think you're thinking of a different thing here: this is not about the whole audio band and not about distortion in general; it's about the phase shifts at the bottom of the band that can result in oscillations when NFB is applied around the amplifier (with enough phase shift it turns into PFB). These shifts accumulate over AC-coupled stages and OPT contributes some shift as well. I believe this is what Yvesm was trying to say (and my understanding of the original question on this thread was the same - namely "is there enough delay in the signal to cause problems when feedback is used ?").

RDH4 goes pretty deep into this subject, with few examples with actual figures for easier understanding, you might want to check it out if you're interested.
Yes Arnulf, that is exactly what i wanted to say (is there enough delay in the signal to cause problems when NFB is used?)with different words!
So, CAN the DELAY cause problems?
I think it can cause serious problems if not very well done....
 

Arnulf

Member
2009-02-02 9:41 am
Yes Arnulf, that is exactly what i wanted to say (is there enough delay in the signal to cause problems when NFB is used?)with different words!
So, CAN the DELAY cause problems?
I think it can cause serious problems if not very well done....

Of course it can. I very warmly recommend RDH4 (it is avaliable for download online because it seems the copyright has long since expired), it might look scary at first but if you focuns on the section that interests you and read it over few times untils it sinks in, you're going to get a very good idea about how feedback works (with regards to phase - delay). This book is an encyclopedia of tubes, almost 1500 pages of nothing but tube goodness in 4th edition, awesome stuff :)

Google for Radiotron Designer's Handbook, 4th edition !
 

forr

Member
2004-12-01 6:46 pm
Next door
Maybe threre's some confusion about the word delay
A time delay is contant at all frequencies, it is very short in amps.
A phase delay, albeit expressed in angles, can be seen as a time delay varying with frequency.

See Bob Cordell's book, "Timeliness of Correction", p505,
about the fallacy of the concept of feedback seen as an "attempt to correct errors after they have happenes".
 

20to20

Member
2010-06-23 9:25 pm
All of you who keep referring to and insisting on the exsistance of a ''delay'' in the feedback, still don't get it. There is no delay. The phase shift of any particular frequency at the OPT is just an attenuation of the feedback level of that frequency. The feedback line can be tailored with components to equalize the levels and compensate for frequencies that can cause oscillation in an unstable amp. But that has nothing to do with the fact that there is no delay in the feedback signal returned to the input. That's why a square wave is used to see any inefficiency. You have to use a quick rise time wave just to see anything.
 

Arnulf

Member
2009-02-02 9:41 am
This could be a matter of semantics - I'm not a native English speaker so I can't quite verbalize everything I want to say precisely the way I would want to. Other members may be experiencing same problems. This is an English-language forum so we do our best to adapt.

As the OP has indicated, his concern was whether feedback can result in enough delayed signal to cause problems (such as oscillations). The answer is: "Yes, at the edge of the band in question, not across the entire band in question".