Well, I used the same value caps, just better quality than stock. Since Manley himself made triode, and triode switchable amps I doubt it diverged much from his aims. I happen to have owned two of the 75/75 amps, and have seen the innards of a couple more of his larger amps. In the early days, they weren't using consistent component values anyway. He enlarged the caps for the main power supply and switched to better caps once he got established. The earlier amps used a bunch of photoflash caps. He also seemed to use some circuit resistor values in a range of values. They were all in places it likely didn't matter terribly. Some of his amps had 220k input impedance, some 100k, some 174k due to the resistors at the input. I experimented with that, and you get essentially no difference in sound or amp frequency response even up to a 1 mega ohm impedance. He even used parallel resistors of a slightly different value vs a single resistor in the overall feedback. The values were close, but not exact.
I think Manley was an excellent designer. VTL's were good I think because the circuits were simple, well executed and not critically designed to the ragged edge, they had rather beefier power supplies than most commercial competitors, and they had quite good output transformers. Mr. Manley seemed quite adept at mixing those simple variations for a pleasing result.
I also think some have misunderstood my post, probably my fault due to the title. VTL's didn't sound overtly tube, murky or soft. They in fact sound quite transparent, resolving, strong, and musical. However, the component changes seemed to make all those strengths even more apparent. Switching to triode simply sounded even more inviting. Though I used them both in Ultralinear, and triode at times. My point was what sounded every step of the way like obtaining a clearer, cleaner, transparent, higher resolution result more true to the source was in fact a happy accident. The improvements, which subjectively most everyone who heard them considered improvements, were the result of subtle enhancing colorations. The Spectrals in comparison, that sounded good, but second best to the VTL's, were in fact much higher fidelity amplifiers. So good they were fully capable of passing unecumbered the signal pre-conditioned by passing through the VTL amps.
So I think lots of effort is wasted in the wrong direction with improving home sound reproduction. If such subjective enhancements were better understood, they could be accomplished without building big, hot, expensive tube amplifiers, especially where you need lots of power for your speakers. It is my opinion the place to do this is instead at the pre-amp. I have in mind building a mini-push-pull tube pre-amp with output transformers. Have a 12AX7 or similar driving each channel. You could replicate in small scale a triode push-pull amp that only drives the inputs of a quality solid state amp. You could get exactly the sound of a big triode doing this. You would only need three 12AX7 tubes per channel.
Alternatively single ended fans could do the same thing. Build very small single ended amps with these driving solid state amps of quality. I do realize most tube pre-amps are single ended in nature. But they don't have transformer coupling at the output, and aren't loaded down to produce the same results as single ended power amps connected to speakers. Anyone with a good single ended amp could try this for themselves. Load the output with power resistors, and tap it with a voltage divider to drop the output voltage down to pre-amp levels. Play the result through some SS amps and see what you think.
Thanks for the graphs. Interesting that H3 & H5 seem to be the same between the Push-Pull and the SET. It the H2 that's so much lower on P-P (as one would expect). So that low H2 offers little or no masking of H3.
Funny that H4 is MIA on both amps. That is also what I see on my little 6V6 SE amp.
Funny that H4 is MIA on both amps. That is also what I see on my little 6V6 SE amp.
You should've seen what P-P transistor amps typically show.
I'm only saying it because this is tube amp section. SS amps with heaps of feedback may measure better into an electronic load such as a resistor, but how about the electromechanical interface between the loudspeaker, air and ultimately our ears?
I remember someone here measured the harmonic content of amplifiers, through speakers, at the listening position, and SETs actually performed far better than super linear SS amps!
I remember someone here measured the harmonic content of amplifiers, through speakers, at the listening position, and SETs actually performed far better than super linear SS amps!
To borrow an analogy, digital, and to a lesser extent, sand will give you the musical image as seen through a pane of slightly to somewhat etched glass, but most tube gear will provide a much clearer vision of the original image, if possibly a bit wavy near the edges and maybe with a hint of tinting.
I know which I prefer for enjoyable listening!
I know which I prefer for enjoyable listening!
Transistors are so nonlinear compared to tubes, that they need quite a bit of negative feedback to linearize them, which causes harmonic distortion products to extend way out above the fundamental, which gives the sound a fake sense of clarity in the short term, and gets tedious to listen to over time. Tubes can be run with little or no feedback and avoid that problem pretty much entirely. A low or no feedback tube amp more typically has a fast drop off of harmonic distortion products (unless poorly designed or over driven of coarse). Symmetrical topologies will only generate odd harmonic distortion products, especially when negative feedback causes their symmetry to be effectively exact. The 2nd harmonic (which is the same note as the fundamental but one octave higher) enhances the sound of the music, unless so overdone that it causes significant I.M. distortions. Both tubes and transistors can sound incredible if the circuit is designed well. Bottom line: Tubes tend to distort much more like the human ear-brain mechanism distorts; fast drop-off of harmonic products, and all harmonics are included. Transistors more often have less measurable distortion on the bench with test signals, but it's the nasty kind. Only odd harmonics, and they extend way out from the fundamental. In a push-pull output circuit, I consider crossover distortion one of the biggest culprits of bad sound. Transistor circuits are usually pretty bad on this, regardless of feedback, which can make it worse. Lots of higher order harmonics. Especially for low level signals. Tubes usually do much better until clipping, and then it depends on the topology of the phase splitter and bias levels.
If you compare a transistor power amp with a tube power amp, don't forget to take into account the source impedance of the amp as viewed by the speaker. If the speaker has a passive crossover, and it's hooked to a low or no feedback tube amp (where the amps output Z is often 1 to 8 ohms), that can throw off the calibration of the passive crossover pretty badly. The high feedback amps will have output Z that is well under an ohm, so that problem is nilch. There's more, but maybe this is enough for now.
True.
Not true, except possibly in a fully symmetrical circuit. The exponential BJT curve has just as much even as odd order.
I suspect this is because some designers, and most tweakers, do not realise how dependent on output topology is the correct setting for quiescent current. For example, CFP typically needs 10's of mA, Darlington needs 100's of mA.
You're right. I mis spoke.
I should have proof read what I wrote more carefully. When the topology was symmetrical, and substantial negative feedback was applied, we measured very little even harmonic distortion content. The 3rd was very dominant, and the product extended way out from the fundamental. When the feedback was reduced way down and the drive to the push-pull output stage was unbalanced by one dB, the 2nd harmonic became dominant, and the products above that tapered off quickly. If a symmetrical topology includes plenty of even harmonics, I would guess that it's because of loose tolerances causing a slight gain differential driving the output section, as is probably typical in tube circuits. I may be wrong, but I believe that high feedback will cause a symmetrical topology to be so accurately symmetrical as to largely get rid of the even harmonics.
I should have proof read what I wrote more carefully. When the topology was symmetrical, and substantial negative feedback was applied, we measured very little even harmonic distortion content. The 3rd was very dominant, and the product extended way out from the fundamental. When the feedback was reduced way down and the drive to the push-pull output stage was unbalanced by one dB, the 2nd harmonic became dominant, and the products above that tapered off quickly. If a symmetrical topology includes plenty of even harmonics, I would guess that it's because of loose tolerances causing a slight gain differential driving the output section, as is probably typical in tube circuits. I may be wrong, but I believe that high feedback will cause a symmetrical topology to be so accurately symmetrical as to largely get rid of the even harmonics.
You will never, never, never, ever have a perfect amplification device. The perfect lens for telescopes, microscopes, or cameras has yet to be ground. The "wire with gain" does not exist. Ain't. Gonna. Happen.
You can try to do your best, but you'll never get there. It wouldn't do you any good since whatever you play through your hypothetical wire with gain, perfect amp has already been through just essssss-loads of imperfect devices between the time the sound was picked up in the studio and delivered to you on vinyl, tape, or CD. All amps are, more or less, "FX boxes".
Hollow state does sound better, but there is no reason for solid state to sound as hideous as it does. I've done both, and have solid state designs that come quite close, and definitely outperform any Big Box SS amp I've ever heard.
Design for good open loop performance, add just that amount of local and/or gNFB to improve it. What you don't want is to be pouring on the NFB to hide your open loop design faux pas. That's the surest path to sonic mediocrity.
You can try to do your best, but you'll never get there. It wouldn't do you any good since whatever you play through your hypothetical wire with gain, perfect amp has already been through just essssss-loads of imperfect devices between the time the sound was picked up in the studio and delivered to you on vinyl, tape, or CD. All amps are, more or less, "FX boxes".
Hollow state does sound better, but there is no reason for solid state to sound as hideous as it does. I've done both, and have solid state designs that come quite close, and definitely outperform any Big Box SS amp I've ever heard.
Design for good open loop performance, add just that amount of local and/or gNFB to improve it. What you don't want is to be pouring on the NFB to hide your open loop design faux pas. That's the surest path to sonic mediocrity.
Last edited:
Tubes vs Trasistors, and other love songs
I loved all the comments and criticisms that have come to the table here. There's nothing like some juicy audio engineering collaboration... Before I forget, 12AX7's are relatively non-linear tubes. 6SL7 is a step in the right direction (what I plan to use), as long as you consider their increased microphonic sensitivity (if it's an issue). To be fair to transistors, I was stunned by the sound quality of a Carver transistor poweramp (Maybe an AV705? not sure - 100watts/ch., five channels in one chassis) driving the Linkwitz Orions, with all those op-amps in the signal path (active crossover)). The source material was as good quality as it gets (Jennifer Crock recordings of the Portland chamber orchestra) - that had something to do with it. So transistors CAN sound spectacular, although over time it might have gotten a bit tedious - only got to listen for an hour.
My fairly extensive research over the last decade got me to believe that phase margin in high feedback amps is trickier than most EE's realize, and the feedback correction stage (diff. amp or whatever) is significantly non-linear itself and gets much worse at the higher frequencies (not just because feedback is rolled off at high F).
I think it was Crowhurst who was saying back in the 1950's that energies would travel around the feedback loop endlessly, generating new I.M. products and other distortions each time around the loop, which makes you wonder if a couple of steady sinewaves on a test bench is really a good representation of actual music, with all of its complexity (dozens if not hundreds of different frequencies beating against each other at any given moment, and what about temporal psycho-acoustic effects, etc.). It could be that any particular distortion may be of very low amplitude measured on a bench, but hundreds of tiny distortions generated by the complexity of actual music might add up to an audible coloration(?). So I've become pretty critical of high feedback circuits. The best reason for feedback IMO is to reduce the output Z on a low feedback tube power amp, so the highly reactive speaker cone is more tightly held, especially at bass frequencies.
Tube amps are IMO definitely an art object. Maybe it's part about nostalgia for me too. That said, since enjoyment is more important to me than what test equipment says is higher fidelity on a bench, I'll admit that I welcome and even somewhat encourage the existence of the 2nd harmonic presence. It's the same note an octave higher. The one harmonic that never really gets in the way musically. It somehow makes the music feel more real to me. It is a coloration, but more helpful than damaging in my mind (Too much 2nd can imply high I.M. though). Higher order harmonics are almost invariably bad. It's like the really cute girl shows up at your party, but she's got her boyfriend with her... And the boyfriend is the son of the lead singer in your band..... Not that that happened to me recently. (sorry)
What I look for as much as anything, is the distortion spectrum shape, regardless of tubes vs transistors, one tube vs another, one topology vs another. I recently put a Gary Pimm cascode FET current source in place of the plate resistor on a 6SN7, and saw what I consider the perfect distortion spectrum shape. The 2nd was 65dB down, the 3rd about 80 dB down, and no visability of any other harmonics, with about a 5 volt sinewave at its output. We varied a bunch of things and it was still superb.
So now I've got to build a Line Preamp. How can you not use a circuit that works that good(?). You've got no negative feedback at all, and those distortion numbers and spectrum shape... I can't resist... I've been wanting to do a 4 section James style (passive Baxandall) tone circuit for years, to show all those "high-enders" that tone controls can be a good thing, when there's no neg feedback and questionable phase margin or slewing related distortion issues. You add an input selector and a volume control and you've got a whole preamp. Getting good enough pots isn't trivial though. I just ordered 5 of the TKD CP600 series pots at $35 each. I'm trying to get some discontinued Noble AP25's too, to compare. They're getting hard to find. It's just that she was SO HOT!!!!! Dam.
I loved all the comments and criticisms that have come to the table here. There's nothing like some juicy audio engineering collaboration... Before I forget, 12AX7's are relatively non-linear tubes. 6SL7 is a step in the right direction (what I plan to use), as long as you consider their increased microphonic sensitivity (if it's an issue). To be fair to transistors, I was stunned by the sound quality of a Carver transistor poweramp (Maybe an AV705? not sure - 100watts/ch., five channels in one chassis) driving the Linkwitz Orions, with all those op-amps in the signal path (active crossover)). The source material was as good quality as it gets (Jennifer Crock recordings of the Portland chamber orchestra) - that had something to do with it. So transistors CAN sound spectacular, although over time it might have gotten a bit tedious - only got to listen for an hour.
My fairly extensive research over the last decade got me to believe that phase margin in high feedback amps is trickier than most EE's realize, and the feedback correction stage (diff. amp or whatever) is significantly non-linear itself and gets much worse at the higher frequencies (not just because feedback is rolled off at high F).
I think it was Crowhurst who was saying back in the 1950's that energies would travel around the feedback loop endlessly, generating new I.M. products and other distortions each time around the loop, which makes you wonder if a couple of steady sinewaves on a test bench is really a good representation of actual music, with all of its complexity (dozens if not hundreds of different frequencies beating against each other at any given moment, and what about temporal psycho-acoustic effects, etc.). It could be that any particular distortion may be of very low amplitude measured on a bench, but hundreds of tiny distortions generated by the complexity of actual music might add up to an audible coloration(?). So I've become pretty critical of high feedback circuits. The best reason for feedback IMO is to reduce the output Z on a low feedback tube power amp, so the highly reactive speaker cone is more tightly held, especially at bass frequencies.
Tube amps are IMO definitely an art object. Maybe it's part about nostalgia for me too. That said, since enjoyment is more important to me than what test equipment says is higher fidelity on a bench, I'll admit that I welcome and even somewhat encourage the existence of the 2nd harmonic presence. It's the same note an octave higher. The one harmonic that never really gets in the way musically. It somehow makes the music feel more real to me. It is a coloration, but more helpful than damaging in my mind (Too much 2nd can imply high I.M. though). Higher order harmonics are almost invariably bad. It's like the really cute girl shows up at your party, but she's got her boyfriend with her... And the boyfriend is the son of the lead singer in your band..... Not that that happened to me recently. (sorry)
What I look for as much as anything, is the distortion spectrum shape, regardless of tubes vs transistors, one tube vs another, one topology vs another. I recently put a Gary Pimm cascode FET current source in place of the plate resistor on a 6SN7, and saw what I consider the perfect distortion spectrum shape. The 2nd was 65dB down, the 3rd about 80 dB down, and no visability of any other harmonics, with about a 5 volt sinewave at its output. We varied a bunch of things and it was still superb.
So now I've got to build a Line Preamp. How can you not use a circuit that works that good(?). You've got no negative feedback at all, and those distortion numbers and spectrum shape... I can't resist... I've been wanting to do a 4 section James style (passive Baxandall) tone circuit for years, to show all those "high-enders" that tone controls can be a good thing, when there's no neg feedback and questionable phase margin or slewing related distortion issues. You add an input selector and a volume control and you've got a whole preamp. Getting good enough pots isn't trivial though. I just ordered 5 of the TKD CP600 series pots at $35 each. I'm trying to get some discontinued Noble AP25's too, to compare. They're getting hard to find. It's just that she was SO HOT!!!!! Dam.
That's the rumor. The truth is quite different- run correctly, they are outstandingly linear.
http://www.diyaudio.com/forums/tubes-valves/164920-guess-tube.html
I don't think Crowhurst said that, since he actually understood how feedback worked.
Sorry, that read much harsher in tone than what I meant. The notion of some sort of signal "circulation" is a common one. It just isn't the case. With respect to audio frequencies (and RF, for that matter), the voltages in the circuit all rise and fall together. Delay times from one part of the circuit to another are negligible. What is important is phase leads and lags, and Crowhurst wrote extensively about that. As well, there is a notion about feedback and the order of distortion- which is partially true. But it's not hard, with feedback, to get all of the higher order harmonics down to a totally insignificant level, assuming competent circuit design.
What feedback does, however, is make stability and overload considerations MUCH more critical. A good designer can use feedback effectively- in the hands of a less-than-good designer, feedback can make a circuit perform very poorly.
What feedback does, however, is make stability and overload considerations MUCH more critical. A good designer can use feedback effectively- in the hands of a less-than-good designer, feedback can make a circuit perform very poorly.
If a circuit had perfect parts and was designed perfectly, then perhaps negative feedback wouldn't cause harmonic distortion products to extend way out from the fundamental. In the real world, the correction stage that combines the input signal with the feedback signal, sending a correction signal to the output stage, is only so good, and often not real good. With tubes, if well designed, you don't need more than a few dBs of feedback, if any at all, to have some of the best sound I've heard. With transistors I'm not sure anyone has been able to avoid neg feedback, so they may have very low levels of distortion with one or two tones on a bench, but the harmonic and I.M. products do extend way out, and could cause slewing related issues.
You may be right. If you have pure odd-order distortion, then feedback will not re-introduce even-order. If you have pure even-order, then feedback will create odd-order too.
No, unless 'perfect' means no distortion to start with. If you have some distortion, then feedback will always add higher orders, while reducing the overall level. This is a simple matter of mathematics, not a sign of poor engineering.
I'm not sure things are quite as bad as that. However, unless the feedback signal is fed back to the actual input point (it usually isn't) there is the possibility of problems. An SS amp typically uses the other base of the input LTP, while valves often use the input cathode as the feedback point. In both cases the designer is hoping that the common-mode gain of the stage far exceeds its differential-mode gain (I suspect that most designers don't even think about this). LTP with current-source tail may be fine, except that SS amps have high feedback factors so need high ratio between common-mode and diff-mode. Valves may run closer to the edge, but are less distorting anyway.
Unless distortion is so low that no feedback is needed, a few dBs of feedback are likely to sound worse than no feedback. Use none, or enough!
Well, I won't/can't argue with what you like to listen to- that's what makes a horse race. But the notion that zero feedback means low orders of distortion and high feedback means high orders of distortion is just not correct ASSUMING competent design. Here's three examples (from Stereophile magazine, www.stereophile.com) from expensive and well-regarded (in the high end niche market) amplifiers, one a tube no-feedback design, one a solid state no-feedback design, and the third a transistorized very high feedback design. Which one shows lower orders of distortion?
Attachments
Exceptions?
Sy, you may be right, there are exceptions to my global sounding statements. But if there's anything I've learned over the years it's that there are always variables, usually lots, and without doing measurements myself, or at least seeing all the tech info on how test results are created, I have to take them with a grain of salt. But thanks for the feedback.
Sy, you may be right, there are exceptions to my global sounding statements. But if there's anything I've learned over the years it's that there are always variables, usually lots, and without doing measurements myself, or at least seeing all the tech info on how test results are created, I have to take them with a grain of salt. But thanks for the feedback.
I also wanted to comment that a good friend of mine who is VERY adept with tube anything, has personally measured the linearity of virtually every triode ever made, dozens of samples of each of the more common types, and he tells me that the 12AX7 is actually pretty non-linear, and the 6SN7 and 6SL7 are amongst the most linear. Furthermore, I've got the Morgan Jones Valve Amps book sittin here, and he did linearity tests on these tubes himself and shows the results in his book, which agree with what my friend found.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.