I spent some time with a prototype single ended tube amp that had 3 nested feedback loops. Having been enamored by tube designs using 300B tubes and NO feedback I was sure it was going to sound terrible. But the demos I heard were impressive and I really admired the engineer and his unusual background. Needless to say, that amp is gorgeous. I purchased it, and a later prototype and a matching preamp. But 6wpc ended up not really doing the job I wanted so I have gone hybrid using both tube and SS. But the feedback design had a great sound and I keep them for a "baseline" system when my perception needs to be restored after too much experimentation. Kind of like the guy who wants a great orchestral system so has to keep going to live performances to stay"tuned in".
Feedback is to the audio engineer what seasoning is to the chef. The right amount will bring out the flavor of something that is already good, but too much is like trying to cover the taste of bad meat.
I'm not sure what you mean by "amount of feedback"? Less feedback means more gain. In an op-amp architecture, the gain of 1 can be achieved with a voltage follower (100% feedback) that usually will have much less noise and distortion than if you run little feedback e.g. 100 times gain. Generally for op-amps, more feedback will improve the signal within the frequency band supported by the amp. But of course, feedback can also cause issues especially when the slew rate is poor against the frequencies used.
Feedback has a massive impact on the sound of an amp. Nelson pass probably explains it better than anyone on his website. High feedback is what makes Op Amps and Class D amps sound relatively inferior IME.
I started out working for Analog Devices after getting out of the army. Back then if you wanted a good opamp you bought one in a plastic case with discrete components that was filled with epoxy. The ic opamps back then were pretty much at the low end of the scale. There was a bit of a spec war going on between Burr Brown and us so we both used tricks to "extend the response of our products and while these did extend the response you were better off not using a product in that range if you wanted the best performance.
You will HAVE to get Darren to explain properly... you only gave a peripheral summary ... with no real explanation as to how feedback really works ... It won't matter if Darren gets technical because only technical minded people would be watching ...
If the feedback signal is too delayed then the difference between input and feedback will yield audible transient intermodulation distortion or ringing which you can see with a square wave signal on an oscilloscope. It was a big topic in the early days of transistor amps where slew rate became a spec thing for amps.
Yes, slew rate became popular with wideband/ highspeed amps in the mid/ late '70's. I was a Kenwood rep when they introduced their "Highspeed" line with bandwidths out to 200KHz. I toured with a "Highspeed Clinic" doing in store tests for slew rates and rise times for any amp that walked in. A 10Khz square wave at rated power into 8ohms. The Kenwood KA-907 I had for reference always blew anything else away. I had to stop testing Sony's because their recently introduced switching PS would literally always go up in smoke!
This is largely not a thing now, of course, do to much improved designs and the universal knowledge of how to tame the feedback loop. No self-respecting audio designer would eschew feedback in he audio range. The only time it can be an issue is if you do something stupid like use 100’ of zip cord for a remote speaker installation. The wire will look like a resonant circuit due to the large distributed capacitance and inductance. It would not be subtle, either. The oscillations would probably cause lots of distortion and blocking in the audio.
@@mysock351C Yes exactly right. A good audio engineer can design a circuit where the use of feedback does great things to the audio quality. Personally I also like servo subwoofers where the feedback helps manage the cone motions to track more precisely the signal reducing the decay. Unfortunately, not many servo subwoofers are being produced in recent years and that's a bummer.
@@ThinkingBetter A point often overlooked by audiophiles is that analog electronics owes its existence to feedback. Back in the early 20th century, it was what made tube amps with reasonable distortion and power outputs possible. And it would be nice to see servo controlled subwoofers more. Most competent DSP based designs without any control have fairly modest distortion of a few percent at moderate SPL levels, but with lots of excursion the woofer starts to get nonlinear.
@@ThinkingBetter Although with the advances in motor and suspension design one day servo control may not even be needed. Its amazing how much better modern drivers are compared to what we had back in the 80's and 90's.
Most people just don't go out and purchase individual op-amps. However for those that may...You can examine the mfgr''s data sheet a determine just "how bad" the "feedback" will be in the circuit even with 100% feedback (aka a text-book buffer circuit). OP Amp buffer circuits (circuits that use a straight wire feedback connection and therefore utilize 100% negative feedback) are ubiquitous in most pro. recording consoles A-to-D, Streamers and DACS. So we are all constantly listening to music that has already been "processed" using tons of negative feedback regardless of what our home systems contain. The distortion contributors in the vast majority of op-amp based circuits has more to do with the "purity" of capacitors, resistors and parasitic inductance surrounding it. NOT the amount of feedback. Negative Feedback circuitry "inside" a power amp is trickier subject since you have to deal with both voltage and current gain stages. Many companies have received patents on power amplifier feedback circuit designs.
@@glenncurry3041 You might be surprised how important an analog education is or was!😉 Anyway as a working engineer through the latter part of the 20th century you had to keep up with a lot of that digital stuff anyway..
@@shipsahoy1793 Just kidding! They were still teaching tubes when I got mine in the early '70's. Digital was monostable multivibrators! Memory was magnetic core.
I think Paul really needs to make a video on the evils of NOT using feedback. Some of the in-house designs by PS Audio have the worst objective (and even subjective) performance ever measured in solid state equipment. Yet their modular amp designs blithely ignore the fact that their COTS class-D modules used HEAVILY rely on feedback to keep their distortion low. 🤔
There are some issues with this. Op amps by definition (all of them) require feedback to function correctly. (Op amps that do not require feedback--because it's already applied internally to the part--are called instrumentation amplifiers.) Open loop bandwidth, and (usually) open loop gain, are not generally important. The relevant parameter here is Gain Bandwidth Product (GBP). But this whole discussion, IMO, is far beyond the scope of most audio gear enthusiasts. Some viewers surely already know all this, but electronic design is far too technical for most non-engineers. Those who want to learn would be better served by a more academically appropriate process. An introductory treatise cannot be taught starting at this high level and not just verbally with a few minutes here and there.
Might not always agree with Paul but generally explains things in a way that can be understood but on this occasion I found the explanation of feedback to be all over the place and a little messy. Sorry.
You can prove it is not good if you are using too much. You can see it on a spectrum analyzer and a FFT plot. A little is fine but a lot changes the harmonic distortion profile.
There is a thing which two parts of audio world (means producers of amplifier for driving the speaker sets and producers of speakers to be driven) do not want to comment. It is the common output of both treated as acoustic result . Amplifier prodecers present their developement with pure resistor's load and producers of speaker drivers measure them as mostly linear in single mode. Then listener gets in salon two and makes with connection of two of them one. Speaker sets which have no any standard behavoiur appear now as new component of amplifier. Producer was't even suspecting it there to be by any time😄. There comes the lost battlefield due to feedback which connects complicated behaviour of crosovers and even more unsettled speakers behaviour back to input of amplifier - all of them with messed by resonances phases. Feedback strong regule is not to have any frequency depended phase shifts within operation and the same is for amplifier. . But the feedback voltage is equal to speaker set impedance multiplied by speaker set current . That current depends on resonances and undergoes shifts. resulting in shifting the output voltage So it at least seems to me in theory That is why my idea goes to make feedback as separate from speakers as possible - good exapmle were in even prewar radios tube amplifiers where feedback was connected to secondary winding of output transformer - the fb voltage was excited by transformer current directly on it's constant impedance . Good enough for equalizing main speaker resonance . But with middle resonances created by crossovers I see no solution except that made by Philips known as (troublesome) movement feedback which separates amplifier fb from speaker set crossover.
@Douglas Blake You did not read what I said and stright rectify me?. Check - I had written it isolates from "speaker set crossover" not speaker. Anyway it isolates also from speakers which are not in motion loop To high degree. Anyway thanks for confirmation that I am right.
This gets into using feedback to create an extremely high Damping Factor spec instead of good basic low internal resistance designs. High feedback loops in the output to create a high DF just adds gain to compensate for voltage drop across a higher internal resistance circuit. It does not change the internal resistance. Counter EMF from the speaker is dropped across it instead of being able to control the driver directly. So high DF amps often sound worse than modest DFs. Anything over 100 is artificial.
@@glenncurry3041 It is not clear for me why you want to explain me things which I was digging some 40 years ago, got own detailed knowledge and after I made my own motion speaker fb resigned from because found it "troublesome" Did you check all you told by youurself? . Anyway thanks for effort
@Douglas Blake I was watching a youtube the other day where they were trying to prove an extremely high DF using heavy feedback was OK even with an internal output resistance of 10ohms because when the load changed the voltage stayed the same. Ignoring the increased drop across the internal 10ohms only being compensated by increasing the gain. As if the 10 ohms did not exist because they could compensate with gain? But that 10ohms was still there in the output circuit messing up the counter EMF! But they had some +2,000 DF! And yes I think language is a problem in this discussion.
Unwanted phase shift is the actual evil of negative feedback.🤣 I can’t believe people are actually losing sleep over this kind of stuff🤣 Ps) hope it’s a joke..😉
Very high open loop bandwidth opamps that are current feedback based sound very good.
I spent some time with a prototype single ended tube amp that had 3 nested feedback loops. Having been enamored by tube designs using 300B tubes and NO feedback I was sure it was going to sound terrible. But the demos I heard were impressive and I really admired the engineer and his unusual background. Needless to say, that amp is gorgeous. I purchased it, and a later prototype and a matching preamp. But 6wpc ended up not really doing the job I wanted so I have gone hybrid using both tube and SS. But the feedback design had a great sound and I keep them for a "baseline" system when my perception needs to be restored after too much experimentation. Kind of like the guy who wants a great orchestral system so has to keep going to live performances to stay"tuned in".
Feedback is to the audio engineer what seasoning is to the chef. The right amount will bring out the flavor of something that is already good, but too much is like trying to cover the taste of bad meat.
Excellent metaphor.
I'm not sure what you mean by "amount of feedback"? Less feedback means more gain. In an op-amp architecture, the gain of 1 can be achieved with a voltage follower (100% feedback) that usually will have much less noise and distortion than if you run little feedback e.g. 100 times gain. Generally for op-amps, more feedback will improve the signal within the frequency band supported by the amp. But of course, feedback can also cause issues especially when the slew rate is poor against the frequencies used.
Feedback has a massive impact on the sound of an amp. Nelson pass probably explains it better than anyone on his website. High feedback is what makes Op Amps and Class D amps sound relatively inferior IME.
I started out working for Analog Devices after getting out of the army. Back then if you wanted a good opamp you bought one in a plastic case with discrete components that was filled with epoxy. The ic opamps back then were pretty much at the low end of the scale. There was a bit of a spec war going on between Burr Brown and us so we both used tricks to "extend the response of our products and while these did extend the response you were better off not using a product in that range if you wanted the best performance.
You will HAVE to get Darren to explain properly... you only gave a peripheral summary ... with no real explanation as to how feedback really works ...
It won't matter if Darren gets technical because only technical minded people would be watching ...
If the feedback signal is too delayed then the difference between input and feedback will yield audible transient intermodulation distortion or ringing which you can see with a square wave signal on an oscilloscope. It was a big topic in the early days of transistor amps where slew rate became a spec thing for amps.
Yes, slew rate became popular with wideband/ highspeed amps in the mid/ late '70's. I was a Kenwood rep when they introduced their "Highspeed" line with bandwidths out to 200KHz. I toured with a "Highspeed Clinic" doing in store tests for slew rates and rise times for any amp that walked in. A 10Khz square wave at rated power into 8ohms. The Kenwood KA-907 I had for reference always blew anything else away. I had to stop testing Sony's because their recently introduced switching PS would literally always go up in smoke!
This is largely not a thing now, of course, do to much improved designs and the universal knowledge of how to tame the feedback loop. No self-respecting audio designer would eschew feedback in he audio range. The only time it can be an issue is if you do something stupid like use 100’ of zip cord for a remote speaker installation. The wire will look like a resonant circuit due to the large distributed capacitance and inductance. It would not be subtle, either. The oscillations would probably cause lots of distortion and blocking in the audio.
@@mysock351C Yes exactly right. A good audio engineer can design a circuit where the use of feedback does great things to the audio quality. Personally I also like servo subwoofers where the feedback helps manage the cone motions to track more precisely the signal reducing the decay. Unfortunately, not many servo subwoofers are being produced in recent years and that's a bummer.
@@ThinkingBetter A point often overlooked by audiophiles is that analog electronics owes its existence to feedback. Back in the early 20th century, it was what made tube amps with reasonable distortion and power outputs possible.
And it would be nice to see servo controlled subwoofers more. Most competent DSP based designs without any control have fairly modest distortion of a few percent at moderate SPL levels, but with lots of excursion the woofer starts to get nonlinear.
@@ThinkingBetter Although with the advances in motor and suspension design one day servo control may not even be needed. Its amazing how much better modern drivers are compared to what we had back in the 80's and 90's.
Most people just don't go out and purchase individual op-amps. However for those that may...You can examine the mfgr''s data sheet a determine just "how bad" the "feedback" will be in the circuit even with 100% feedback (aka a text-book buffer circuit). OP Amp buffer circuits (circuits that use a straight wire feedback connection and therefore utilize 100% negative feedback) are ubiquitous in most pro. recording consoles A-to-D, Streamers and DACS. So we are all constantly listening to music that has already been "processed" using tons of negative feedback regardless of what our home systems contain. The distortion contributors in the vast majority of op-amp based circuits has more to do with the "purity" of capacitors, resistors and parasitic inductance surrounding it. NOT the amount of feedback. Negative Feedback circuitry "inside" a power amp is trickier subject since you have to deal with both voltage and current gain stages. Many companies have received patents on power amplifier feedback circuit designs.
F/B is one of my pet topics ... I'd love Darren to cut loose over this topic over lunch time ... I'll be with him all the way ...
Thanks for simple explanations to complicated subjects.
Some things are not so simple to explain properly to a layman. I’m actually glad that I got an electronics engineering degree 40 years ago.
So recently! LOL!
@@glenncurry3041 You
might be surprised how important an analog education is or was!😉
Anyway as a working engineer through the latter part of the 20th century you had to keep up with a lot of that digital stuff anyway..
@@shipsahoy1793 Just kidding! They were still teaching tubes when I got mine in the early '70's. Digital was monostable multivibrators! Memory was magnetic core.
@@glenncurry3041 👍
Lol.. yep, I worked with so many younger engineers that had no concept of tube operation or design. Lucky Us!!🥳
I think Paul really needs to make a video on the evils of NOT using feedback. Some of the in-house designs by PS Audio have the worst objective (and even subjective) performance ever measured in solid state equipment. Yet their modular amp designs blithely ignore the fact that their COTS class-D modules used HEAVILY rely on feedback to keep their distortion low. 🤔
There are some issues with this. Op amps by definition (all of them) require feedback to function correctly. (Op amps that do not require feedback--because it's already applied internally to the part--are called instrumentation amplifiers.) Open loop bandwidth, and (usually) open loop gain, are not generally important. The relevant parameter here is Gain Bandwidth Product (GBP).
But this whole discussion, IMO, is far beyond the scope of most audio gear enthusiasts. Some viewers surely already know all this, but electronic design is far too technical for most non-engineers. Those who want to learn would be better served by a more academically appropriate process. An introductory treatise cannot be taught starting at this high level and not just verbally with a few minutes here and there.
Might not always agree with Paul but generally explains things in a way that can be understood but on this occasion I found the explanation of feedback to be all over the place and a little messy. Sorry.
The lunch videos were excellent iirc
I'm looking forward to more lunch with x videos.
You can prove it is not good if you are using too much. You can see it on a spectrum analyzer and a FFT plot. A little is fine but a lot changes the harmonic distortion profile.
is it better to have a discrete op amp with feedback
And then there is feed forward.
Thanks, good info.
There is a thing which two parts of audio world (means producers of amplifier for driving the speaker sets and producers of speakers to be driven) do not want to comment. It is the common output of both treated as acoustic result . Amplifier prodecers present their developement with pure resistor's load and producers of speaker drivers measure them as mostly linear in single mode.
Then listener gets in salon two and makes with connection of two of them one. Speaker sets which have no any standard behavoiur appear now as new component of amplifier. Producer was't even suspecting it there to be by any time😄.
There comes the lost battlefield due to feedback which connects complicated behaviour of crosovers and even more unsettled speakers behaviour back to input of amplifier - all of them with messed by resonances phases. Feedback strong regule is not to have any frequency depended phase shifts within operation and the same is for amplifier. . But the feedback voltage is equal to speaker set impedance multiplied by speaker set current . That current depends on resonances and undergoes shifts. resulting in shifting the output voltage So it at least seems to me in theory
That is why my idea goes to make feedback as separate from speakers as possible - good exapmle were in even prewar radios tube amplifiers where feedback was connected to secondary winding of output transformer - the fb voltage was excited by transformer current directly on it's constant impedance . Good enough for equalizing main speaker resonance . But with middle resonances created by crossovers I see no solution except that made by Philips known as (troublesome) movement feedback which separates amplifier fb from speaker set crossover.
@Douglas Blake You did not read what I said and stright rectify me?. Check - I had written it isolates from "speaker set crossover" not speaker. Anyway it isolates also from speakers which are not in motion loop To high degree.
Anyway thanks for confirmation that I am right.
This gets into using feedback to create an extremely high Damping Factor spec instead of good basic low internal resistance designs. High feedback loops in the output to create a high DF just adds gain to compensate for voltage drop across a higher internal resistance circuit. It does not change the internal resistance. Counter EMF from the speaker is dropped across it instead of being able to control the driver directly. So high DF amps often sound worse than modest DFs. Anything over 100 is artificial.
@@glenncurry3041 It is not clear for me why you want to explain me things which I was digging some 40 years ago, got own detailed knowledge and after I made my own motion speaker fb resigned from because found it "troublesome" Did you check all you told by youurself? . Anyway thanks for effort
@Douglas Blake I have imperssion that our talking starts a typical gossip about audio with no hope for tech. conclusion. Thanks anyway.
@Douglas Blake I was watching a youtube the other day where they were trying to prove an extremely high DF using heavy feedback was OK even with an internal output resistance of 10ohms because when the load changed the voltage stayed the same. Ignoring the increased drop across the internal 10ohms only being compensated by increasing the gain. As if the 10 ohms did not exist because they could compensate with gain? But that 10ohms was still there in the output circuit messing up the counter EMF! But they had some +2,000 DF!
And yes I think language is a problem in this discussion.
Unwanted phase shift is the actual evil of negative feedback.🤣 I can’t believe people are actually losing sleep over this kind of stuff🤣
Ps) hope it’s a joke..😉
Oh, I am the first! where's Roderick today?:)
.... You are FIRST congratulations on this amazing performance you get 🥇🏆🍾🥂👏🇺🇸 You were so fast to click
@@RoderikvanReekum lol not so fast, 18 mins after publish :))
Slow like a snail ! But thankyou ;)
I get more than enough evil from my $20.00 Bluetooth speaker 😀🙃
Yes, two evils '$20' & 'Bluetooth' never equal good (right)
I think you proved feedback is bad by talking so much 🤔