Fun with Feedback

What people are working on at the moment
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#106 Re: Fun with Feedback

Post by Cressy Snr »

Aye, that’s about the size of it.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#107 Re: Fun with Feedback

Post by Cressy Snr »

So to conclude:

1) Measurements matter so a scope and sig gen are required if you want to do a project like this one.
2) Global feedback in a single-ended valve amp is good when implemented properly.
3) There is a bit of engineering maths and AC theory to be got to grips with.

Great sound in single ended mode can be had from common or garden modern production, indirectly heated, triode strapped pentodes, when used with global feedback. Anyone who tells you it can't be done should be ignored.
Using KT150s, a potentially superb 25W SEUL or 15W triode amp could be built. Icon Audio do one I believe but I don't know what feedback scheme David Shaw uses apart from the fact that it is switchable by a toggle on the rear panel.

Building this thing has certainly expanded my engineering knowledge. With the advice from Nick and James, and some bloody hard work, it's given me a solid base to go from if I want to do anything else like that. Job's a good'n I think.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#108 Re: Fun with Feedback

Post by Cressy Snr »

Apart from the fact that the damned thing heats up the living room, which is the last thing we want in this weather :shock:
So for now it’s the solid state amp.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#109 Re: Fun with Feedback

Post by Mike H »

For what it's worth, everything I've done is better with NFB.

And I agree with you, my experience has been that the FB point has to be taken from the same place where the speaker is connected, if off another secondary tap, it's all gets a bit strange. Aka, 'summat's oop wi' it'. :D
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#110 Re: Fun with Feedback

Post by Cressy Snr »

Mike H wrote: Mon Jul 11, 2022 7:38 pm For what it's worth, everything I've done is better with NFB.

And I agree with you, my experience has been that the FB point has to be taken from the same place where the speaker is connected, if off another secondary tap, it's all gets a bit strange. Aka, 'summat's oop wi' it'. :D
Aye. The FB in my amp is taken from the rear of the speaker output terminals. That way, you are feeding back the same signal as the speaker is receiving. Taking it off another tap, means you are not.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#111 Re: Fun with Feedback

Post by steve s »

Cressy Snr wrote: Mon Jul 04, 2022 3:08 pm I know also that RD would have had a fit over the number of caps sprinkled around the circuit and would have given the two wrongs don't make a right' argument. Others will no doubt say 'if you had used linear tubes (expensive DHTs) in the first place then there wouldn't have been any need for nasty horrible feedback.'This is a valid point, but I don't have the resources any more for expensive DHTs and even more expensive interstage/parafeed/output transformers. This is a project of necessity as the mother of invention when it comes down to it.

This project's aim is to try to obtain a bit of single-ended magic on the cheap and if by necessity, I have to use inexpensive audio pentodes strapped as triodes, cheap ubiquitous voltage amps and drivers, and cheapo output transformers, then using negative feedback, properly applied is a damned good way of getting the kind of sound quality that punches way above its weight.

TBH, this is probably the only forum where I would have gotten away with doing what I've just done. I wouldn't have dared to put this project on DIYAudio for instance. I'd probably have been chewed up, spat out and run out of town whilst picking shotgun pellets out of my arse.
Steve, as no one has commented on this I though I would say a few words
Your last few weeks for me have been very interesting, and cemented further my limited understanding on the subject.
I was also firmly in the 'dont go there' feedback place.
..mainly for a what am I going to gain by doing that?
But you have also made me think from a getting the best from what you have perspective, which any manufacturer worth his salt should also be doing.


I have heard so much kit and have my idea of what reproduced music should sound like now, . I seriously don't know how to improve on some of the amps I build (within their power limitations of course)
Funny enough the mofo is not sounding as good to me as it was a year or so ago.
Some of my best triodes beat it in many respects, but not in the bass
So as ever my system is fraught with comprimise.

Leading on from the feedback talk a few months ago, and what has been said in this thread, the following is true...
Feed back hf correction is output transformer dependant or loudspeaker dependent when there's no transformers.

And the following are my assumptions..
To work perfectly the feed back would need to align perfectly with the signal at all frequencies
As that doesn't happen, the compensation caps improve the situation
And by careful calcs you can get it close with a test signal.
Again my understanding ?

But at low frequencies that correction may be a few cycles behind, it is a lot more cycles behind at higher frequencies, so the actual results could be actually more programme dependent ?
And linked of course with the loudspeakers ability to reveal the signal

My last point is..when the feed back does not line up correctly, it is still added or subtracted from the original signal.

And depending on its actual phase at the time when it reunited.. that is going to slightly ( or a lot depending on the phasing), change the original signal?
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
User avatar
Nick
Site Admin
Posts: 15708
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#112 Re: Fun with Feedback

Post by Nick »

But at low frequencies that correction may be a few cycles behind
No, to put it mildly you are misthinking. At the highest frequency you are interested in (say 20kHz) if your delay is more than a few percent of a single cycle then you have problems that you probably cant fix. So at low frequencies the delay should be so tiny that its effectively 0.

Again as I said at length in the thread about asymmetry you need to think about the now and the almost now, not think about a sequence of cycles, but small time slices within the smallest of the cycles in the signal.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Nick
Site Admin
Posts: 15708
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#113 Re: Fun with Feedback

Post by Nick »

Feed back hf correction is output transformer dependant or loudspeaker dependent when there's no transformers.
In the absence of a transformer the load should have little or no effect on the feedback, at least assuming a lowish output impedance, but certainly not in the way a transformer will do.

Just to add to this, the phrase "the load should have little or no effect on the feedback" is true for amps that are basically stable. Its an argument in favor of those sorts of amps.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#114 Re: Fun with Feedback

Post by steve s »

Nick wrote: Tue Jul 12, 2022 10:46 am
But at low frequencies that correction may be a few cycles behind
No, to put it mildly you are misthinking. At the highest frequency you are interested in (say 20kHz) if your delay is more than a few percent of a single cycle then you have problems that you probably cant fix. So at low frequencies the delay should be so tiny that its effectively 0.

Again as I said at length in the thread about asymmetry you need to think about the now and the almost now, not think about a sequence of cycles, but small time slices within the smallest of the cycles in the signal.
Thanks nick, I was reading about delays though recording digital processors a few weeks ago so I may well be confused
But that prompts another question
Your mention of positive feedback .. that must be caused by over correction ?
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
User avatar
Nick
Site Admin
Posts: 15708
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#115 Re: Fun with Feedback

Post by Nick »

Your mention of positive feedback .. that must be caused by over correction
Not sure what you mean, but negative feedback happens if you take your input signal, amplify it, take the output signal, multiply it by -1 then add a fraction of that to the input signal. The -1 means its negative feedback. However at a given frequency, say 200kHz (to avoid you getting worried about delays that span cycles of audio) a half cycle takes 2.5us. If the delay through the amp is getting to that duration, then the -1 becomes +1 and so the negative feedback becomes positive.

Your mention of "over correction" makes me unsure again how you are thinking about this. I think you are maybe thinking too much like its a mechanical servo system with overshoot. There are similarities but I think the analogy is maybe taking you somewhere else.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#116 Re: Fun with Feedback

Post by JamesD »

Please remember that all the views of and types of feedback are all expressed within the context of the system it is being used in - there is no universal panacea... global feedback, interstage feedback, local feedback, negative feedback, positive feedback, etc. all have their place in ones system building armoury. They can all make things better or worse from a perceived sound perspective depending on the system and how they are implemented within the amplifier and system context. For instance, I have had examples of amplifiers and loudspeakers that sound worse when feebdack was derived from the loudspeaker terminal and sounded better when derived from the collector of the output transistor rather than its emitter... or from the otherside of an output zobel network - normally when GFB is derived in a system it does work best when taken from the loudspeaker terminal but not always - system context plays its role and it is relevant to keep that in mind when making or reading statements on how feedback is a cure all as it is't in many cases and particularly in low power SE DHT amplifiers it can remove the things from the sound that make us love them in the first place!

In Steve's system where he is using IHT valves with lots of system gain to burn under GFB and driving conventional speakers it is entirely appropriate and, as he has demonstrated, the right solution to making it sound terrific when optimised for that system....

ciao

James
User avatar
Nick
Site Admin
Posts: 15708
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#117 Re: Fun with Feedback

Post by Nick »

normally when GFB is derived in a system it does work best when taken from the loudspeaker terminal but not always
Yes, always YMMV, but there is maybe a bit more noise in this than there has to be. I think Steve was maybe referring back to when he found with his NVA amp that moving the NFB take off point made a difference. This is generally true that in a PP system you want the feedback to be a combinational point and not on one leg from the output devices. This is one of Self's sources of distortion. In that context taking it from the point the two output devices meet or the speaker terminal is not that much different other than you are including some more internal wiring into the loop. This is entirely different from the cases I think James is talking about where it may not be a push pull system, or there may be balanced feedback.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Nick
Site Admin
Posts: 15708
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#118 Re: Fun with Feedback

Post by Nick »

there is no universal panacea
Just to reiterate that, no of course not. However feedback in all its forms is just another tool in the box, and like every tool, its important to understand how and when (or not) to use it. And to do that you need to understand how it works.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#119 Re: Fun with Feedback

Post by JamesD »

Hi Nick,

Thanks for the clarification and indeed the case I was referring to ended up with balanced feedback off each collector circuit and was a Class A PP solid state amplifier but I've had similar experiences with SE SS and valve amps too. My experience is that its easier to optimise the way a stage sounds when using local feedback rather than GFB. I also don't like the way GBW works under GFB loops - its much easier to define and control GBW under local stages than under GFB - at least to me it is :-)

Agree completely on the toolbox statements!
User avatar
Mike H
Amstrad Tower of Power
Posts: 20157
Joined: Sat Oct 04, 2008 5:38 pm
Location: The Fens
Contact:

#120 Re: Fun with Feedback

Post by Mike H »

Nick wrote: Tue Jul 12, 2022 11:45 am
Your mention of positive feedback .. that must be caused by over correction
Not sure what you mean, but negative feedback happens if you take your input signal, amplify it, take the output signal, multiply it by -1 then add a fraction of that to the input signal. The -1 means its negative feedback. However at a given frequency, say 200kHz (to avoid you getting worried about delays that span cycles of audio) a half cycle takes 2.5us. If the delay through the amp is getting to that duration, then the -1 becomes +1 and so the negative feedback becomes positive.
If feedback is positive and gain is >1, oscillation will take place. (It's a quote, if badly remembered. :lol: )

How does NFB become positive with gain >1 ? What Nick said. :D
 
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
Post Reply