Being more in diy audio and reading up on different class a amplifier designs of Nelson pass you really begin to appreciate his genius for making terrific sounding simple circuit designs with the lowest number of parts. And usually throwing in some tricks and some of his matched rare out of production parts.
Agreed. Follow Papa Pass on DIYaudio and it becomes clear. Or build your own triode based amp and listen to it. I do the latter. Then with the simple circuit choose parts carefully. Even in a low feedback system the choice of the single resistor used in the feedback loop can make the difference. Every part degrades the signal just a little. It all adds up. Use a lot of them and you simply miss some of the low level information that was in the source. And you end up with a hollow facsimile of the original that measures well, but sounds like a dead machine vs live voice and instruments.
My feedback is that feedback can be great also in audio. As for all things in audio engineering, great sound is a product of great engineering and passion. Chances are that all your favorite music has gone through a ton of amps (incl. opamps) with feedback. Often feedback is what makes the output much more immune to the load impedance. Some circuits are impossible without feedback. E.g. how would you make a class D amplifier without it?
I had a play with this on my favourite circuits in LTSpice, I took 20dBs out and saw reduced bandwidth and much lower slew rate but better stability, distortion was 10x higher but still
There goes that whooshing sound again...as much of the conversation flies over my head. BUT, I'm learning something! Thanks so much for the glimpse into the design theory and rules that you must play by. Fascinating!
The year is 2020, if you want to design audio circuits without negative feedback, vacuum tubes are the only way to do it. Whatever hapenned to progress? 🤔
Read up on siltechcables.com Their Saga amp used DC to power vacuum tubes. I’ll paste it from their site: Elimination Of AC Circuitry The biggest source of self-noise in a conventional amplifier is the AC supply, both the incoming signal itself and the massive transformers and power supply components necessary to handle it. Siltech eliminated this problem by making the line-stage and separate V1 voltage gain stage battery powered, using sophisticated rechargeable cells to drive specially selected low-noise tubes. The results combine the remarkably linear voltage gain of vacuum tubes with the ghostly quiet background of a pure DC supply.
Sometimes simple is best. A few years ago I built a La Pacifica phono stage, a few jFets and some resistors and caps for the passive frequency network and that was it. Not perfect but a lot better than anything at it's cost point that I have seen.
Burson (Composer 3x) prides themselves on simplicticy. It sounds fantastic paired w/ Lsim 707, A-S2100 & AQ cables. The opamps are directly next to XLR outputs. I returned Topping D90. I'm trying a Gustard X26 just to round out the $1400dac shootout. Probably stepping up to a 300w Mcintosh solution, but the Yamaha is fantastic!!! :D :D :D
Hegel SoundEngine doesn't use any feedback but adaptive feedforward instead and should have zero crossover distortion. But does is sound like a feedback amp or not? Would it still be better with distortion?
I think he's saying to design linear, low noise, responsive circuits first, then maybe add some negative feedback later. I don't know much about analog audio but this is an interesting concept.
Simple linear circuit, low or no feedback. Triodes operating in the flat line of the curve. Now you are talking my preferred system. Still love to hear what Darren believes is removed from the signal by higher feedback designs that kills the sound that makes the toe tap. My guess is there are low level signals in the pico volt range that simply get flattened in a complex high feedback circuit. Then missing is the information that our brain recognizes as emotion cues.
Tubes are 3/2 power transconductance linearity. Bipolar is logarithmic transconductance linearity. JFET is square law linearity. MOSFET approaches pure linear transconductance.
Bruno Putzey's likes feedback in his Mola Mola Kaluga class D amps...They sound amazing and I bought them. He is the Master. they do not get to warm...The capacitors cost more than the Hypex module he also invented!
@Benoît I think when the best and perfect amplifiers face with real world (best) speakers they lose 90% of efforts to make the amplifier perfect. Speakers have non-linear impedance input with capacitors (crossover) that can introduce lot of IMD distortion. How the perfect amp can help with speakers non-linear input and IMD?
High-end guitar amp manufacturers have been dealing with this for years. That's why Fender and Marshall guitar amps have remained relatively unchanged since the 60's, and are still the best sounding guitar amps.
I never understood why the audiophile community has a beef with feedback. Using well-designed and properly biased amplifier stages with substantial feedback is where your ultra-low distortion figures come from as it only makes them that much more linear. Additionally it also allows amplifiers to have very low output impedances, sometimes less than a milliohm at the output of the final EF stages, making for essentially perfect voltage amplifiers. On a small 20W/channel amp I have, even with cheap spring-loaded speaker terminals, it has less than 50 milliohms of effective impedance. In fact, _less_ feedback can be worse than none at all, as it will be enough to accentuate additional harmonics due to the process of feeding them back to the input stages, but not enough to suppress them fully as would happen in a proper audio amplifier that has roughly 80-90dB of excess gain.
@G Guest I would also agree that there are times to eschew feedback, but these should be done solely on circuit performance. Interesting to mention the fuses being inside the feedback loop. I've heard of that, but have yet to see any examples. I wonder how nonlinear a fuse could actually be? If they are running the fuse close to its limits it certainly could be, but that also shortens its life. On the other hand, it might allow it to blow fast enough to save the output devices.
Feedback is great until it isn’t. I think the beef is almost always related to running gear into clipping or into tough loads. The simple solution is to go “pro” with higher voltage rails and keep audio amplifier signals far far away from clipping. The biggest challenge to me is still the noise floor - this is where “pro” audio signal levels beat cheap RCA hands down. Great gear never loses details even at low volumes because it operates intrinsically at higher signal levels.
Brilliant! No offense, Paul, (or maybe some) but Darren can straighten out a circuit concept that you can, unaided, turn to a pretzel. ;) Chris? Same thing. Keep bringing these fellows back. Do you have a DAC engineer who might untangle the mess the PR guys and many pundits have made about DAC's? Or, a class A, A/B, D power stage guy, who can explain what it might take to make one sound very much like the others and why. There's way too much witch doctor sticks and bones rather than wisdom going on in audio; way too many trademarks and memes without grounding in reality.
I was once working for a big car company testing electronic circuits that they put into cars. The circuits were so complex for no apparent reason because they were designed by electronics college students that were fresh engineers and not by real electronics engineers with many years of experience. I always said that reduction of complexity is the real art and not designing a text-book circuits because you are hired to do that.
@@ricochetinmyskull Ya but we went through this before on here. The difference between junk and good parts comes out to be pennies in the grand scheme of high end audio. They also said that a simpler design is better which means even less parts and less money. Like anything that is just a sum of existing parts you are paying for the design.
You can not expect to create an amplifier involving non-linear semiconductor junctions without applying feedback. it seems like a delusion to me to expect otherwise.
great tec talk and it means??????. now please use a explantion on what all that means.so do i use a mic next to my amp get feed back and watch it chatch fire like i did at 12 with my stereo yes mic into stereo up all the way burn baby burn.i did not know ot would go up in flames dam allide stereos.junk....
What rubbish. None of these Charlestons can tell us what actually is the problem with global feedback apart from parroting everyone else's line that apparently it is bad ! Or is it a case that if you keep repeating the same message then eventually everyone will agree that it is bad without having to backup their claims with anything other than hear-say ? All semiconductor manufacturers particularly DAC designers do not intentionally introduce distortions because they think the majority will like the sound. No, they always try to minimize distortions because they don't believe their product should interfere with the recording chain and rightfully so ! Only PS Audio seems to intentionally break this rule :D LOL
I didn't get that out of it at all. Global feedback is just a tool which when applied in moderation can have great results. Just like medications, too much can have negative consequences to the sound as an unintended side effect even though the THD numbers look good.
Ok Im done with this video. High slew rate and bandwidth causes them to sound bad? Are you kidding me? Like an IC LM3886 amplifier with 18V/us slewrate and 8MHz of bandwidth that produces distortion figures down around the 0.003-0.005% THD, and they dont have a near cult-like following for nothing: They sound very good! The high slew rate and bandwidths are there for a specific reason: To allow the feedback to work its magic. Sounds like the tail wagging the dog as usual in the audiowanker community. People with more money than sense get this notion that something is bad, and companies cater to them to get that money, regardless of facts.
I don't think he was implying that higher bandwidth and slew rates are inherently bad in and of themselves. He mentioned dependency on the quality of the circuit layout, so it sounds like his concern is regarding the fact that faster devices are more prone to break into undesired oscillation. Even when an oscillation isn't directly audible it can rob the audio of tone quality, detail and dynamics. DC stage biases can offset when this happens. Also, devices can be so busy clipping from a high-frequency oscillation that they can't keep up to reproduce the audio detail correctly. None of this is a problem with extra design effort, layout, and testing, but that's a lot more work than simply going with a lower bandwidth device that's fast enough but not too fast. Anyway, I'm 98% sure that's what his thought process was when he said that. The LM3886 isn't the best example for this. I've worked with a couple engineers who would design their audio amps with RF and high-speed video components, swearing the extra effort was worth it because they "sounded better", which I think is just nuts.
@@0x07AF I know what you're referring to, and it _can_ impact the audio. I've seen oscillation with ICs if you try to do something weird with the feedback network that takes the gain below the minimum specified at some point: It will oscillate. I would agree with you that using very fast RF components is unnecessary, but this is mainly down to difficulty in achieving stability, and wasted bandwidth on something mundane like audio as you alluded to. If done correctly, though, they will work just as well as any other linear amplifier (and sound the same), just that you've squandered the excess performance of the active devices. Its also worth mentioning that properly designed linear amplifiers will show absolutely NO difference in sound from one unit to the next as they almost universally have low THD, very low output impedance, and sufficient bandwidth to cover the audio range. In other words, they are essentially perfect voltage amplifiers. The only exception could be low frequency bass where amplifiers are usually DC compensated to ensure they a) dont have substantial offsets, and b) dont pass a large DC offset from another device onto the speakers and damage them, as you may know. This obviously will roll the response off at very low frequencies. However, virtually all of the difference in sound will be due to which loudspeakers you use as they obviously use many different designs to achieve various goals, but with audio amplifiers, the design objective is very well defined as it just has to amplify. Easier said than done, but still something that should pose no great difficulty. This is something the audiophile community has misunderstood, as there are many solid state amplifiers with mediocre, if not abysmal performance characteristics. The lack of feedback (or reduced feedback) means that usually many more active devices are needed in the final stage to reduce beta variation and output impedance. I think there's one company that uses something like *FOURTY* output transistors in their amplifier that does not have traditional feedback. The specs are very good, but it seems like a lot of work to avoid installing two resistors and a capacitor as would usually be done, and they (as usual) tell some bunk story of why feedback is bad. Even a quick flip through PS Audio's catalog of amps shows this. They have only modest THD and output impedances on many of their amplifiers priced at many thousands of dollars, which seems like a raw deal for the consumer. Will it impact your audio? Doubt it. But for that kind of money Id expect at least an order of magnitude better performance from amplifiers of that caliber. With that said I dont hate PS, and I rather like the Sprout compact amp, which looks like a solid product. But for the rest Im rather unimpressed.
Great chat, guys. When Darren mentioned simple circuits that sound great, I immediately thought of Nelson Pass.
Thanks, Darren and Paul!
Me too!
Nice! It’s totally true.
Definitely Nelson PASS
@@stimpy1226
No doubt about it!
Being more in diy audio and reading up on different class a amplifier designs of Nelson pass you really begin to appreciate his genius for making terrific sounding simple circuit designs with the lowest number of parts. And usually throwing in some tricks and some of his matched rare out of production parts.
Agreed. Follow Papa Pass on DIYaudio and it becomes clear. Or build your own triode based amp and listen to it. I do the latter. Then with the simple circuit choose parts carefully. Even in a low feedback system the choice of the single resistor used in the feedback loop can make the difference. Every part degrades the signal just a little. It all adds up. Use a lot of them and you simply miss some of the low level information that was in the source. And you end up with a hollow facsimile of the original that measures well, but sounds like a dead machine vs live voice and instruments.
Loving these more technical luncheon clips, Darren's an incredible speaker 👌
My feedback is that feedback can be great also in audio. As for all things in audio engineering, great sound is a product of great engineering and passion. Chances are that all your favorite music has gone through a ton of amps (incl. opamps) with feedback. Often feedback is what makes the output much more immune to the load impedance. Some circuits are impossible without feedback. E.g. how would you make a class D amplifier without it?
I really like the more in depth content of these videos that is not covered in the ask paul videos !
Nerd talk, love it.
Thank you for all these PS Audio info offerings.
Thanks guys, yes to sound over specs, Ps audio, Nelson pass and Burson audio I think are on the same page, oh also many VTL AMPS.👍
I had a play with this on my favourite circuits in LTSpice, I took 20dBs out and saw reduced bandwidth and much lower slew rate but better stability, distortion was 10x higher but still
There goes that whooshing sound again...as much of the conversation flies over my head.
BUT, I'm learning something!
Thanks so much for the glimpse into the design theory and rules that you must play by.
Fascinating!
Same here but isn't it great to listen to someone passionate about what they do?
@Fat Rat 🤮🤮😱 😄🤪
The year is 2020, if you want to design audio circuits without negative feedback, vacuum tubes are the only way to do it. Whatever hapenned to progress? 🤔
Read up on siltechcables.com
Their Saga amp used DC to power vacuum tubes.
I’ll paste it from their site:
Elimination Of AC Circuitry
The biggest source of self-noise in a conventional amplifier is the AC supply, both the incoming signal itself and the massive transformers and power supply components necessary to handle it. Siltech eliminated this problem by making the line-stage and separate V1 voltage gain stage battery powered, using sophisticated rechargeable cells to drive specially selected low-noise tubes. The results combine the remarkably linear voltage gain of vacuum tubes with the ghostly quiet background of a pure DC supply.
Great format.
Sometimes simple is best.
A few years ago I built a La Pacifica phono stage, a few jFets and some resistors and caps for the passive frequency network and that was it. Not perfect but a lot better than anything at it's cost point that I have seen.
Sounds kinda like, "Everything is a compromise" in genius-amp-speak.
I love love love these videos!!!! 🥰🥰🥰
Burson (Composer 3x) prides themselves on simplicticy. It sounds fantastic paired w/ Lsim 707, A-S2100 & AQ cables. The opamps are directly next to XLR outputs. I returned Topping D90. I'm trying a Gustard X26 just to round out the $1400dac shootout. Probably stepping up to a 300w Mcintosh solution, but the Yamaha is fantastic!!! :D :D :D
Hegel SoundEngine doesn't use any feedback but adaptive feedforward instead and should have zero crossover distortion. But does is sound like a feedback amp or not? Would it still be better with distortion?
this guys a genius. not certain i should be eavesdropping on his conversations...
He knows his stuff ... so do I, and I find him very interesting albeit he could pass for me when talking about feedback and op amp performance
I think he's saying to design linear, low noise, responsive circuits first, then maybe add some negative feedback later. I don't know much about analog audio but this is an interesting concept.
Simple linear circuit, low or no feedback. Triodes operating in the flat line of the curve. Now you are talking my preferred system.
Still love to hear what Darren believes is removed from the signal by higher feedback designs that kills the sound that makes the toe tap. My guess is there are low level signals in the pico volt range that simply get flattened in a complex high feedback circuit. Then missing is the information that our brain recognizes as emotion cues.
Sadly, I have no idea what all that means
Tubes are 3/2 power transconductance linearity. Bipolar is logarithmic transconductance linearity. JFET is square law linearity. MOSFET approaches pure linear transconductance.
You got it. This is why I like a tube preamp with a MOSFET power amp.
Bruno Putzey's likes feedback in his Mola Mola Kaluga class D amps...They sound amazing and I bought them. He is the Master. they do not get to warm...The capacitors cost more than the Hypex module he also invented!
@Benoît
I think when the best and perfect amplifiers face with real world (best) speakers they lose 90% of efforts to make the amplifier perfect. Speakers have non-linear impedance input with capacitors (crossover) that can introduce lot of IMD distortion. How the perfect amp can help with speakers non-linear input and IMD?
but all digital amps have a tremendous amount of delay. That is why I do prefer tube in class A amps.
do an A B comparison,play tracks of each what does it sound like?
High-end guitar amp manufacturers have been dealing with this for years. That's why Fender and Marshall guitar amps have remained relatively unchanged since the 60's, and are still the best sounding guitar amps.
I never understood why the audiophile community has a beef with feedback. Using well-designed and properly biased amplifier stages with substantial feedback is where your ultra-low distortion figures come from as it only makes them that much more linear. Additionally it also allows amplifiers to have very low output impedances, sometimes less than a milliohm at the output of the final EF stages, making for essentially perfect voltage amplifiers. On a small 20W/channel amp I have, even with cheap spring-loaded speaker terminals, it has less than 50 milliohms of effective impedance. In fact, _less_ feedback can be worse than none at all, as it will be enough to accentuate additional harmonics due to the process of feeding them back to the input stages, but not enough to suppress them fully as would happen in a proper audio amplifier that has roughly 80-90dB of excess gain.
@G Guest I would also agree that there are times to eschew feedback, but these should be done solely on circuit performance. Interesting to mention the fuses being inside the feedback loop. I've heard of that, but have yet to see any examples. I wonder how nonlinear a fuse could actually be? If they are running the fuse close to its limits it certainly could be, but that also shortens its life. On the other hand, it might allow it to blow fast enough to save the output devices.
Feedback is great until it isn’t. I think the beef is almost always related to running gear into clipping or into tough loads. The simple solution is to go “pro” with higher voltage rails and keep audio amplifier signals far far away from clipping. The biggest challenge to me is still the noise floor - this is where “pro” audio signal levels beat cheap RCA hands down. Great gear never loses details even at low volumes because it operates intrinsically at higher signal levels.
I didn´t understand a bit! I wish I could
Just like almost everything in life: perfect simplicity is way more challenging than complexity
Some will say: "Real men don't use feedback"
Brilliant! No offense, Paul, (or maybe some) but Darren can straighten out a circuit concept that you can, unaided, turn to a pretzel. ;) Chris? Same thing. Keep bringing these fellows back. Do you have a DAC engineer who might untangle the mess the PR guys and many pundits have made about DAC's? Or, a class A, A/B, D power stage guy, who can explain what it might take to make one sound very much like the others and why. There's way too much witch doctor sticks and bones rather than wisdom going on in audio; way too many trademarks and memes without grounding in reality.
Rails = Ground & B+
The guy who designed a 700+-component phono preamp is preaching circuit simplicity. Interesting....
I was once working for a big car company testing electronic circuits that they put into cars. The circuits were so complex for no apparent reason because they were designed by electronics college students that were fresh engineers and not by real electronics engineers with many years of experience. I always said that reduction of complexity is the real art and not designing a text-book circuits because you are hired to do that.
@@VladoT Reminds me of programmers too.
I guess what he meant is simplicity in the signal path.
Tubes are under appreciated in modern Hifi. Thanks for another great discussion!!
Darren should be teaching upper level courses in audio design. Seriously!
No he shouldn't. He should be sacked for incompetence and misleading statements.
@@rentabomb Why not spell out what you mean. Leaving me in suspense.
So basically it's the design not the product's parts.
Both are necessary for a good sounding product. They are only talking about circuit design in this video.
@@ricochetinmyskull Ya but we went through this before on here. The difference between junk and good parts comes out to be pennies in the grand scheme of high end audio. They also said that a simpler design is better which means even less parts and less money. Like anything that is just a sum of existing parts you are paying for the design.
You can not expect to create an amplifier involving non-linear semiconductor junctions without applying feedback. it seems like a delusion to me to expect otherwise.
great tec talk and it means??????. now please use a explantion on what all that means.so do i use a mic next to my amp get feed back and watch it chatch fire like i did at 12 with my stereo yes mic into stereo up all the way burn baby burn.i did not know ot would go up in flames dam allide stereos.junk....
What rubbish. None of these Charlestons can tell us what actually is the problem with global feedback apart from parroting everyone else's line that apparently it is bad ! Or is it a case that if you keep repeating the same message then eventually everyone will agree that it is bad without having to backup their claims with anything other than hear-say ? All semiconductor manufacturers particularly DAC designers do not intentionally introduce distortions because they think the majority will like the sound. No, they always try to minimize distortions because they don't believe their product should interfere with the recording chain and rightfully so ! Only PS Audio seems to intentionally break this rule :D LOL
Should rename his company BS AUDIO. He should sack that engineer.
Perhaps you shoul start by reading some of the publications by Dr Matti Otala 😉
I didn't get that out of it at all. Global feedback is just a tool which when applied in moderation can have great results. Just like medications, too much can have negative consequences to the sound as an unintended side effect even though the THD numbers look good.
@@zulumax1 What are these negative consequences and why ??
@@janjacobsen4804 And all have been well and truly debunked since those days ;)
1😉🎸😳😱🤭✌️🤙🎉🗽🇺🇸
Ok Im done with this video. High slew rate and bandwidth causes them to sound bad? Are you kidding me? Like an IC LM3886 amplifier with 18V/us slewrate and 8MHz of bandwidth that produces distortion figures down around the 0.003-0.005% THD, and they dont have a near cult-like following for nothing: They sound very good! The high slew rate and bandwidths are there for a specific reason: To allow the feedback to work its magic. Sounds like the tail wagging the dog as usual in the audiowanker community. People with more money than sense get this notion that something is bad, and companies cater to them to get that money, regardless of facts.
I don't think he was implying that higher bandwidth and slew rates are inherently bad in and of themselves. He mentioned dependency on the quality of the circuit layout, so it sounds like his concern is regarding the fact that faster devices are more prone to break into undesired oscillation. Even when an oscillation isn't directly audible it can rob the audio of tone quality, detail and dynamics. DC stage biases can offset when this happens. Also, devices can be so busy clipping from a high-frequency oscillation that they can't keep up to reproduce the audio detail correctly. None of this is a problem with extra design effort, layout, and testing, but that's a lot more work than simply going with a lower bandwidth device that's fast enough but not too fast. Anyway, I'm 98% sure that's what his thought process was when he said that. The LM3886 isn't the best example for this. I've worked with a couple engineers who would design their audio amps with RF and high-speed video components, swearing the extra effort was worth it because they "sounded better", which I think is just nuts.
@@0x07AF I know what you're referring to, and it _can_ impact the audio. I've seen oscillation with ICs if you try to do something weird with the feedback network that takes the gain below the minimum specified at some point: It will oscillate. I would agree with you that using very fast RF components is unnecessary, but this is mainly down to difficulty in achieving stability, and wasted bandwidth on something mundane like audio as you alluded to. If done correctly, though, they will work just as well as any other linear amplifier (and sound the same), just that you've squandered the excess performance of the active devices. Its also worth mentioning that properly designed linear amplifiers will show absolutely NO difference in sound from one unit to the next as they almost universally have low THD, very low output impedance, and sufficient bandwidth to cover the audio range. In other words, they are essentially perfect voltage amplifiers. The only exception could be low frequency bass where amplifiers are usually DC compensated to ensure they a) dont have substantial offsets, and b) dont pass a large DC offset from another device onto the speakers and damage them, as you may know. This obviously will roll the response off at very low frequencies. However, virtually all of the difference in sound will be due to which loudspeakers you use as they obviously use many different designs to achieve various goals, but with audio amplifiers, the design objective is very well defined as it just has to amplify. Easier said than done, but still something that should pose no great difficulty. This is something the audiophile community has misunderstood, as there are many solid state amplifiers with mediocre, if not abysmal performance characteristics. The lack of feedback (or reduced feedback) means that usually many more active devices are needed in the final stage to reduce beta variation and output impedance. I think there's one company that uses something like *FOURTY* output transistors in their amplifier that does not have traditional feedback. The specs are very good, but it seems like a lot of work to avoid installing two resistors and a capacitor as would usually be done, and they (as usual) tell some bunk story of why feedback is bad. Even a quick flip through PS Audio's catalog of amps shows this. They have only modest THD and output impedances on many of their amplifiers priced at many thousands of dollars, which seems like a raw deal for the consumer. Will it impact your audio? Doubt it. But for that kind of money Id expect at least an order of magnitude better performance from amplifiers of that caliber. With that said I dont hate PS, and I rather like the Sprout compact amp, which looks like a solid product. But for the rest Im rather unimpressed.