I've been slowly modding my amp one step at a time, now I'm playing with the negative feedback circuit. Since I changed my output tubes from pentode UL to triode I've read feedback is unnecessary. So far the results have been very good, lowering the feedback has resulted in more detail and a cleaner sound, not to mention more gain on the input. Now I'm contemplating removing it altogether, the circuit is a simple one, a wire from the 8 ohm tap to a resistor and a bypass capacitor on the cathode of the input tube. What I'd like to know is what is the safest way to remove the feedback completely? I've read where I can just move the wire from the 8 ohm tap to the ground, or just unsolder the wire and tape it. Couldn't I also just solder the resistor directly to ground and not have the wire running all the way through the chassis? This maybe a temporary mod as I'm trying to find out what's the best sounding setup. TIA.
Negative Feedback Questions
- Thread starter DanTana
- Start date
Rockmonton
Infected with Altecitis
personally, i would just unsolder it from the 8 ohm tap, and tape it to the chassis, but make sure no contact is being made.
ilimzn
Well-Known Member
Actually, this may increase the local feedback in the input stage but also shift the bias point in the input tube to a lower current - depending on how much feedback there is, it could be negligible, but it could also be quite signifficant and possible not good.
If you have to do that (gain and distortion will go up, possibly quite drastically - more data on the amp would help to determine this), you need to unsolder the feedback point from whichever output tap it is connected to and solder it to the ground point on the transformer. This way NFB is deactivated but bias stays the same.
If you have to do that (gain and distortion will go up, possibly quite drastically - more data on the amp would help to determine this), you need to unsolder the feedback point from whichever output tap it is connected to and solder it to the ground point on the transformer. This way NFB is deactivated but bias stays the same.
Thanks for the replies, I had thought about the implications of just leaving the wire off, and soldering it to ground did seem to make the most sense. Like I said, this might just be a temporary mod, as each step I like to be able to reverse it back if needed.
I think feedback is sometimes used for the wrong reasons, covering up flaws in other parts of the circuit or reducing distortion to what is a more marketable figure. I doubt if the SET users would agree feedback is necessary.
I think feedback is sometimes used for the wrong reasons, covering up flaws in other parts of the circuit or reducing distortion to what is a more marketable figure. I doubt if the SET users would agree feedback is necessary.
Who says Neg feedback is bad. Hell I don't even know nor care what it is but isn't The feedback what make tubes sound the way they do. Are you measuring this Feedback with test gear.
For those kids watching at home leave the mad mods to Dan before trying these at home yourselves.
Keep that fire extinguisher handy
For those kids watching at home leave the mad mods to Dan before trying these at home yourselves.
Keep that fire extinguisher handy
tubino
A guy with too many tubes
Dan,
I think it can be instructive to see what happens without neg FB, but there's not a lot of info in your post. Does this loop go to the input tube? You can put in a little pot to increase the resistance from whatever it is now to a whole lot more, like 500K, which is like no loop at all, and see what happens.
Yes, you'll have a LOT more gain. If you're just fooling around, you can plug in a 12AT7 for 12AX7, or 12AU7 for 12AT7, to drastically lower gain.
One of the bad things that is likely to happen is that output impedance will go way up. Mismatches with speakers are much more likely. Bass may get wooly and boomy. Using 16 ohm speakers is recommended to hear whatever good and bad things happen (typically a clearer and more forward midrange, overemphasized).
Most tube amps depend on feedback for linearity, to put it mildly. Some very very good tube amps have excellent linearity without NFB. Unfortunately removing NFB doesn't make a console amp a very very good amp! But it's fun to see what happens, and unless you manage to short something, it should be harmless fun.
I think it can be instructive to see what happens without neg FB, but there's not a lot of info in your post. Does this loop go to the input tube? You can put in a little pot to increase the resistance from whatever it is now to a whole lot more, like 500K, which is like no loop at all, and see what happens.
Yes, you'll have a LOT more gain. If you're just fooling around, you can plug in a 12AT7 for 12AX7, or 12AU7 for 12AT7, to drastically lower gain.
One of the bad things that is likely to happen is that output impedance will go way up. Mismatches with speakers are much more likely. Bass may get wooly and boomy. Using 16 ohm speakers is recommended to hear whatever good and bad things happen (typically a clearer and more forward midrange, overemphasized).
Most tube amps depend on feedback for linearity, to put it mildly. Some very very good tube amps have excellent linearity without NFB. Unfortunately removing NFB doesn't make a console amp a very very good amp!
But it's fun to see what happens, and unless you manage to short something, it should be harmless fun.Yes, the feedback goes from the 8 ohm OPT tap to the input tube cathode with a resistor and capacitor in-between. I realize that feedback can make a amp more linear, flatten the response, but aren't these just ways to correct deficiencies that shouldn't of been there in the first place? In a triode connection the feedback increases third order harmonics, in a pentode in increases 2nd order harmonics, I think those are pretty significant, for their various uses, if I had my tubes in pentode I would leave it, but since I'm now using triode is why I want to try to remove it. Also, the injection of the signal also causes slight, but nevertheless, some signal delays, which masks details and "smears" the sound? I have a few lower gain input tubes I can use, right now I'm using a 12AV7 (5965) about 40 MU, as input and the ECC99 about 22 MU, as the phase splitter.
tubino
A guy with too many tubes
Dan, yes, the NFB is correcting deficiencies, but those deficiences can be so major (limited bandwidth of the OPT) that without it, the amp's deficiencies are emphasized, and not always easy to correct! To put it another way, an amp that doesn't need NFB is likely a whole 'nother amp! Nothing wrong with finding out for yourself, though.
I'm not sure you're right about the 2nd and 3rd order harmonics. It's certainly true that with PP the odd order harmonics dominate, and second order dominate with SE, but I'm not aware of the correlation with triode and pentode.
The deHavilland amps that DingusBoy and I had at AK Fest are SET's with zero NFB. The early pairs from deHav had a pot to adjust NFB, from 0 to something like 9db. I never found a combination in which feedback helped the sound. It was interesting to hear how the NFB veils the sound a bit though...
I've done the same thing you're doing now, with everything from console amps to Eico HF-86s and Dyna 70s. On the Eico I thought it sounded better with no feedback, and a 12AU7 replacing the 7247, even without optimizing the operating points on the newly low-gain triode. But you can't really tell what NFB is doing wrong until you have a circuit that doesn't need it, IMO.
I'm not sure you're right about the 2nd and 3rd order harmonics. It's certainly true that with PP the odd order harmonics dominate, and second order dominate with SE, but I'm not aware of the correlation with triode and pentode.
The deHavilland amps that DingusBoy and I had at AK Fest are SET's with zero NFB. The early pairs from deHav had a pot to adjust NFB, from 0 to something like 9db. I never found a combination in which feedback helped the sound. It was interesting to hear how the NFB veils the sound a bit though...
I've done the same thing you're doing now, with everything from console amps to Eico HF-86s and Dyna 70s. On the Eico I thought it sounded better with no feedback, and a 12AU7 replacing the 7247, even without optimizing the operating points on the newly low-gain triode. But you can't really tell what NFB is doing wrong until you have a circuit that doesn't need it, IMO.
Mike Stehr
Poverty Audio
"One of the bad things that is likely to happen is that output impedance will go way up. Mismatches with speakers are much more likely. Bass may get wooly and boomy. Using 16 ohm speakers is recommended to hear whatever good and bad things happen (typically a clearer and more forward midrange, overemphasized)."
I made a SE triode strapped 6BQ5 amp with no feedback, the bass damping wasn't too bad, but it's a screamer in the midrange. And it seems rolled-off in the highs because of that....
A single-ended triode amp essentially has feedback in the triode tube itself.
I'm probably wrong, but I think it's the capacitance between the grid to filament/cathode?
"Most tube amps depend on feedback for linearity, to put it mildly."
Yup. And it's easily done with a scope.
It never hurts to try things, however, ya gotta find out..... I had to try the triode strapped 6BQ5. I liked it for a couple days and then I didn't.....
I made a SE triode strapped 6BQ5 amp with no feedback, the bass damping wasn't too bad, but it's a screamer in the midrange. And it seems rolled-off in the highs because of that....
A single-ended triode amp essentially has feedback in the triode tube itself.
I'm probably wrong, but I think it's the capacitance between the grid to filament/cathode?
"Most tube amps depend on feedback for linearity, to put it mildly."
Yup. And it's easily done with a scope.
It never hurts to try things, however, ya gotta find out..... I had to try the triode strapped 6BQ5. I liked it for a couple days and then I didn't.....
I'm really trying to see how much detail I can reveal from my sources, I understand feedback masks this? Linearity is important also, one thing if nothing else, I'll find out how good my OPT's are I've got several flavors of tubes to play with, 12AU7, ECC99, 12AV7, 12BH7. I can also learn a little bit more about my speakers while I'm experimenting. So far, I'm really surprised how well they can do microdynamics, that's why I'm trying to extract more detail from my amp. Personally, I believe the perfect amp has no feedback, not saying mine is or ever will be, but I think that "should" be the goal of any good designer.
Thanks for all the feedback and idea's and theory, if I don't blow up my amp...lol...I'll at least learn something.
I've got several flavors of tubes to play with, 12AU7, ECC99, 12AV7, 12BH7. I can also learn a little bit more about my speakers while I'm experimenting. So far, I'm really surprised how well they can do microdynamics, that's why I'm trying to extract more detail from my amp. Personally, I believe the perfect amp has no feedback, not saying mine is or ever will be, but I think that "should" be the goal of any good designer. Thanks for all the feedback and idea's and theory, if I don't blow up my amp...lol...I'll at least learn something.
ilimzn
Well-Known Member
How do you understand feedback masks detail from your sources? Can you explain or even describe this mechanism?
Please do not take this as a rant but just as a sanity check:
I fail to understand how complex sentences that start with 'feedback is' and possibly continue with 'good/bad', AND then have one or more of 'when/because/in case of's and similar, with each of those having multiple 'if/but/or/not clauses' to them, end up conveniently shortened to 'feedback is good' or 'feedback is bad' in audiophile circles, while the mid sized book that follows is interestingly 'forgotten'. Why is that? Is it because it's so much easyer to memorize the currently fashionable mantra, whereas it takes REAL effort figuring out how and why it all works? Lets face it, this is not simple stuff when one really digs into it. And as Einstein said, things should be as simple as possible - but he also added, BUT NO SIMPLER.
In a perfect world, where amplifying elements of circuits are perfectly linear and, come to it, otherwise perfect, feedback would just be a needlessly complicated way of doing something that is already inherent in the circuit. But we do not live in a perfect world. Feedback is a tool - and like any tool, it's not inherently good or bad. It's all about what you do with it. Kind of like a hammer. It's good if you hit a nail on the head, it's bad if you hit yourself on the head. To use it, you need to first learn which end is the business end, and which is the holding end. Granted, there are several ways of learning - you certainly do learn if you survive the hitting yourself in the head part
I do admire that you wish to experiment for yourself. There is far too much dogma in audio, and trying things out is one way to avoid this - though you still need to know how to interpret the results. One thing which is typically done in experiments, is insuring within reason that only one variable is changed. In your experiment, this is not the case. As a reality check: you could just as well enhance the details in your sorces using a carefully designed tone control.
Please do not take this as a rant but just as a sanity check:
I fail to understand how complex sentences that start with 'feedback is' and possibly continue with 'good/bad', AND then have one or more of 'when/because/in case of's and similar, with each of those having multiple 'if/but/or/not clauses' to them, end up conveniently shortened to 'feedback is good' or 'feedback is bad' in audiophile circles, while the mid sized book that follows is interestingly 'forgotten'. Why is that? Is it because it's so much easyer to memorize the currently fashionable mantra, whereas it takes REAL effort figuring out how and why it all works? Lets face it, this is not simple stuff when one really digs into it. And as Einstein said, things should be as simple as possible - but he also added, BUT NO SIMPLER.
In a perfect world, where amplifying elements of circuits are perfectly linear and, come to it, otherwise perfect, feedback would just be a needlessly complicated way of doing something that is already inherent in the circuit. But we do not live in a perfect world. Feedback is a tool - and like any tool, it's not inherently good or bad. It's all about what you do with it. Kind of like a hammer. It's good if you hit a nail on the head, it's bad if you hit yourself on the head. To use it, you need to first learn which end is the business end, and which is the holding end. Granted, there are several ways of learning - you certainly do learn if you survive the hitting yourself in the head part
I do admire that you wish to experiment for yourself. There is far too much dogma in audio, and trying things out is one way to avoid this - though you still need to know how to interpret the results. One thing which is typically done in experiments, is insuring within reason that only one variable is changed. In your experiment, this is not the case. As a reality check: you could just as well enhance the details in your sorces using a carefully designed tone control.
Ilimzn, I don't pretend to know everything, if I did, I'd be designing my own equipment, but I do know one thing for certain, I know what I hear, is it better or is it worse? Feedback to my understanding is injecting part of the signal back through the circuit. It is used to lower distortion, and can increase overall frequency response, it can also increase damping factor. Those are the positive aspects, the negative ones to me, are the fact that it is also used to hide flaws in a circuit. If your design had 1hz-150khz without feedback with a perfect response they why use it?? If injecting a signal that is basically looping from one other part of the circuit, back to the beginning, isn't this also altering the time delay? I know where talking microseconds but it's still delay of some kind. Maybe a local feedback is a much better circuit, this is a global one and eventually, the signal goes round and round and each time is altered, then reinjected into the circuit again. These are my interpretations, if I'm wrong, then please correct me. If without a feedback your signal is 20hz-20khz at 1% distortion, then with feedback it is 1hz-1ghz with .00001% distortion sure that looks good on paper and that's what sells amplifiers, but does it sound better?? That is my whole argument. Nobody can hear past say 25khz, what good does it do to try to go to these astronomical frequency responses when you can't hear it? If you can extract more detail, soundstage, presence, whatever you want to call it, out of your music without it, isn't that the point? I'm trying to find out first hand if this is the case to my ears. So far, my ears agree with me.
After an extended listening session, with the feedback grounded, I'd have to say it is very good sounding. I had it pretty well filtered before, but with none I can tell a slight difference. It was drastic change cutting it in half, that's what got me going in this direction in the first place. The biggest change I can hear is in micro details coming through much better. cymbals seem to have a much better decay, bass is still very good. During high transients, I don't get that "smearing" where sounds seem to just get all confused with each other, now instruments are much better defined, you can pick out an instrument quite easily now during loud passages, it doesn't get lost. Voices seem a tad more natural sounding, soundstaging seems more vibrant and open. Overall, I'm very happy with sound, I don't notice any rolloff of highs, and definitely not in the bass. I've been listening to both CD's and vinyl, and I love how the vinyl sounds with this. CD's sound clearer, with more detail coming through. The most important part of this experiment for me, was also trying to determine where the weak spot was in my system. So far everything else has seemed to come through, so my amp must be the weak link. Now it is much closer to be able to deliver what the rest of my components are feeding it. Maybe now on par with the rest?
tubino
A guy with too many tubes
Dan, how did you cope with the extra gain? Do you notice that the distortion level goes ker-blooey when you push what it can put out? I find a good clean dynamic female vocalist will show that up easily.
Now that I think back on it, DingusBoy's 845 amps have the variable feedback adjustment. I wish we had tried that with the Royd's at AK Fest...
Your no NFB observations square with mine on good amps (on high impedance speakers that don't need the extra damping).
When you say you cut NFB in half, do you mean you doubled R, or you do mean db?
Now that I think back on it, DingusBoy's 845 amps have the variable feedback adjustment. I wish we had tried that with the Royd's at AK Fest...
Your no NFB observations square with mine on good amps (on high impedance speakers that don't need the extra damping).
When you say you cut NFB in half, do you mean you doubled R, or you do mean db?
Rick, At first I did double the R-value, from 100K and added another 100K resistor. That was when I noticed the most difference, I took that out and put in a 220K resistor and noticed less difference but some nonetheless. None seems to be best right now with my setup at least. I am using a 12AV7 now as input 40 mu, and a ECC99 MU 22 as phase splitter. The 12AV7 is a very clean and detailed tube, with very low plate resistance, same goes with the ECC99 but the Mu 22 is a just a bit on the low side as an input.
tubino
A guy with too many tubes
Also meant to ask: output tube is a pentode wired as triode? Sometimes you can find the plate curves for that config... remember it's still not a true triode... and not even true triodes are always so linear... I like the weak link idea, but seems like in my case it's always a matter of finding the most cost-effective approach to any particular link, when almost any can be improved upon! Funny, but the amp is the only area where I don't think about weaknesses at all in my system...
