Brian,
His point was another one. He mentioned "..the (incomplete) estimates from psychoacoustcs of audibility...".
Psychoacoustics is the operative word here. The fact that you think/know your/our audio perception is extremely fine is the whole point.
There are whole sections of industry making good money based on people's inability to tell fake from real but thinking they can.
Jan Didden
His point was another one. He mentioned "..the (incomplete) estimates from psychoacoustcs of audibility...".
Psychoacoustics is the operative word here. The fact that you think/know your/our audio perception is extremely fine is the whole point.
There are whole sections of industry making good money based on people's inability to tell fake from real but thinking they can.
Jan Didden
I think a lot of the problem is also to differentiate between "realism of reproduction," and "pleasant to listen to." How do you define the two in terms of something that can be measured, and more importantly, how do you measure in a credible manner?
fizzard said:
This is also an important issue. In my opinion, you can enjoy music and experience it as 'pleasant to listen to' with very little or no correlation to the objective (however defined) sound quality. Songs heard over the kitchen radio also can give me goosebumps once in a while. For me pleasant to listen to has something relaxed, informal, not having to listen too much to the equipment.
But when I listen to a new amp or speaker, I listen differently. I try to unravel the sound, try to hear the individual instruments, the dynamics, the soundstage, etc. A different way to listen.
Jan Didden
jcx wrote:
This thread reminds me of the first CD players where the popular engineering theory at the time, heavily promoted by Philips marketeers, led me to expect "the perfect sound". But the sound was pants...not even as good as a Dual C505 with a medium priced MM cartridge. That is a fact.
Similar camps arose. I'll paint the caricateur. Those who believed that the improved measurements necessarily implied improved sound tried to fit the listener to the theory; when the listener objected with complaints of discomfort and headaches the believers discredited the listeners, explaining that their hearing was not used to the realistic dynamics of CD or their perception was spoiled by becoming too acustomed to vinyl. I am not joking...these conclusions were in the mags at the time. Very much akin to the earlier tube/transistor debate. It was10 years before I bought a CD player. That's how long it took specialist audio companies to get the engineering good enough and to bring the theory up to scratch.
In either case one can hear when something is not realistic.
This thread reminds me of the first CD players where the popular engineering theory at the time, heavily promoted by Philips marketeers, led me to expect "the perfect sound". But the sound was pants...not even as good as a Dual C505 with a medium priced MM cartridge. That is a fact.
Similar camps arose. I'll paint the caricateur. Those who believed that the improved measurements necessarily implied improved sound tried to fit the listener to the theory; when the listener objected with complaints of discomfort and headaches the believers discredited the listeners, explaining that their hearing was not used to the realistic dynamics of CD or their perception was spoiled by becoming too acustomed to vinyl. I am not joking...these conclusions were in the mags at the time. Very much akin to the earlier tube/transistor debate. It was10 years before I bought a CD player. That's how long it took specialist audio companies to get the engineering good enough and to bring the theory up to scratch.
The tube -vs- transistor and vinyl -vs- digital debate still rage on. The height of both of those debates was before my time though.
To a certain extent I liken tube amps and full range high sensitivity speakers to an old woman with lipstick. Some people like a little "tone" to their music. And like the little old woman needs to put a little more lipstick on before she sees it in the mirror, as people age they need a little more "tone" before they hear it.
To a certain extent I liken tube amps and full range high sensitivity speakers to an old woman with lipstick. Some people like a little "tone" to their music. And like the little old woman needs to put a little more lipstick on before she sees it in the mirror, as people age they need a little more "tone" before they hear it.
Fizzard, it is much more difficult to make a solid state amp sound as good or better than a good vacuum tube amp, than you appear to realize now. I have worked on the problem for 40 years, and still have a lot to learn. It is NOT the addition of distortion that makes a tube sound better. In fact, many tube amps are MORE linear (open loop) than most solid state amps. 'Maybe' this is a clue to better sounding amps.
Many here sound exactly the same as I did and thought, 40 years ago: Feedback and sophisticated circuits will get a real jump on tubes. I have tried for 40 years, and still I know that I have not been completely successful. You will find this too, IF you get an independent comparison made of your efforts against the best tube designs.
Many here sound exactly the same as I did and thought, 40 years ago: Feedback and sophisticated circuits will get a real jump on tubes. I have tried for 40 years, and still I know that I have not been completely successful. You will find this too, IF you get an independent comparison made of your efforts against the best tube designs.
I would still be doing tubes if they hadn't gotten so expensive. My main tube amps only use something like 10dB of feedback, yet have well over 100kHz bandwidth and THD around .1%. Most solid state stuff can't perform that well so close to open loop.
(True confessions: I do sometimes get tempted to do an OTL amp...)
There's a neat thing that you can do with tubes that many solid state designs can't handle, which is that you can vary the feedback on the fly with a pot, the same way you would with a tone control. (Yes, you have to adjust the volume, unless you go to the trouble to gang pots so that the volume varies along with the feedback--I never bothered, I just turned the volume up and down to compensate.) That is the quickest, easiest, and most iron-clad way I know of to demonstrate what feedback does to sound. It's one thing to listen to a solid state amp for a while, turn it off, take it to the bench, pull out the resistors, put in the next set, carry the amp back in, plug it in, turn it on...and...oh, fooey! just grab the knob and crank and you can hear it in real time.
And before anyone asks, yes, the characteristics of feedback are pretty much the same regardless of whether you're diddling a tube circuit or a solid state one.
Solid state is unquestionably more reliable; if nothing else, you don't have to replace the confounded things every couple thousand hours. The problem I have is that solid state has a hard time doing things that tubes do effortlessly. It's frustrating. I've got a rule of thumb that states that well designed class A solid state is more-or-less equivalent to class AB tubes. Unfortunately, I've never heard a solid state amp that can compete with tubes run class A...but my gawd, the heat...(and expense and weight and size and...)
It was hilarious to go through the '70s as a musician. Every frickin' month the manufacturers would come out with another amp and the hype would start all over again: This time we've really done it. No, really. No, seriously, we really mean it. No kidding. We've made a solid state amp that sounds like tubes. No...stop laughing guys...really...we mean it this time...
And always it would be pathetic.
Okay, so it's easy to laugh because people who play electric instruments frequently play distorted. But tubes for instruments are pretty much the same as tubes for stereo in the sense that they sound better and no one really knows why. It's easy to prattle on about second harmonics and such, but adding second harmonic doesn't make solid state sound like tubes in either stage gear or stereo, so obviously the truth is more complicated than just distortion spectra. Solid state's gotten better, no question, but the comparison is still "it sounds as good as tubes" when speaking of solid state, and not the other way around.
The more linear thing is often laughed at by solid state people, but there's this one nifty little thing...look at the transfer curve for a tube and you get this sorta S-shaped thing, just like solid state. Most solid state people never get beyond the shape--no better than solid state, they chortle. Worse, in fact. Blah, blah, blah. But the detail they fail to notice is the scale for the voltage--that S-shaped curve is stretched out over hundreds of volts. Now, take a couple of volt swing and superimpose it on that curve.
Intellectually, you know that the Earth is round, but it sure doesn't look that way, particularly if you're someplace like Kansas. The curve of the Earth's surface, because it's so stretched out, begins to feel more like a straight line. And so it is with tubes. The trick works better in preamps than amps, because you're using less of the available curve.
So what to do?
The obvious thing to try is to use tubes in the front end and solid solid state in back, but it doesn't work well because, while tubes can swing voltage like nobody's business, they don't provide enough current to drive the solid state. Bummer.
The second thing to try is...well...stretching out the curve on solid state devices. This goes flat-slap against The Gospel of High Current, because you can't have both high voltage and high current at the same time or the device goes poof. Bummer. And besides, all the members of the Church of High Current laugh at you.
But it does produce some interesting results.
Grey
(True confessions: I do sometimes get tempted to do an OTL amp...)
There's a neat thing that you can do with tubes that many solid state designs can't handle, which is that you can vary the feedback on the fly with a pot, the same way you would with a tone control. (Yes, you have to adjust the volume, unless you go to the trouble to gang pots so that the volume varies along with the feedback--I never bothered, I just turned the volume up and down to compensate.) That is the quickest, easiest, and most iron-clad way I know of to demonstrate what feedback does to sound. It's one thing to listen to a solid state amp for a while, turn it off, take it to the bench, pull out the resistors, put in the next set, carry the amp back in, plug it in, turn it on...and...oh, fooey! just grab the knob and crank and you can hear it in real time.
And before anyone asks, yes, the characteristics of feedback are pretty much the same regardless of whether you're diddling a tube circuit or a solid state one.
Solid state is unquestionably more reliable; if nothing else, you don't have to replace the confounded things every couple thousand hours. The problem I have is that solid state has a hard time doing things that tubes do effortlessly. It's frustrating. I've got a rule of thumb that states that well designed class A solid state is more-or-less equivalent to class AB tubes. Unfortunately, I've never heard a solid state amp that can compete with tubes run class A...but my gawd, the heat...(and expense and weight and size and...)
It was hilarious to go through the '70s as a musician. Every frickin' month the manufacturers would come out with another amp and the hype would start all over again: This time we've really done it. No, really. No, seriously, we really mean it. No kidding. We've made a solid state amp that sounds like tubes. No...stop laughing guys...really...we mean it this time...
And always it would be pathetic.
Okay, so it's easy to laugh because people who play electric instruments frequently play distorted. But tubes for instruments are pretty much the same as tubes for stereo in the sense that they sound better and no one really knows why. It's easy to prattle on about second harmonics and such, but adding second harmonic doesn't make solid state sound like tubes in either stage gear or stereo, so obviously the truth is more complicated than just distortion spectra. Solid state's gotten better, no question, but the comparison is still "it sounds as good as tubes" when speaking of solid state, and not the other way around.
The more linear thing is often laughed at by solid state people, but there's this one nifty little thing...look at the transfer curve for a tube and you get this sorta S-shaped thing, just like solid state. Most solid state people never get beyond the shape--no better than solid state, they chortle. Worse, in fact. Blah, blah, blah. But the detail they fail to notice is the scale for the voltage--that S-shaped curve is stretched out over hundreds of volts. Now, take a couple of volt swing and superimpose it on that curve.
Intellectually, you know that the Earth is round, but it sure doesn't look that way, particularly if you're someplace like Kansas. The curve of the Earth's surface, because it's so stretched out, begins to feel more like a straight line. And so it is with tubes. The trick works better in preamps than amps, because you're using less of the available curve.
So what to do?
The obvious thing to try is to use tubes in the front end and solid solid state in back, but it doesn't work well because, while tubes can swing voltage like nobody's business, they don't provide enough current to drive the solid state. Bummer.
The second thing to try is...well...stretching out the curve on solid state devices. This goes flat-slap against The Gospel of High Current, because you can't have both high voltage and high current at the same time or the device goes poof. Bummer. And besides, all the members of the Church of High Current laugh at you.
But it does produce some interesting results.
Grey
Grey,
The interface between tube and SS is the single greatest challenge of hybrid design......
Hugh
The interface between tube and SS is the single greatest challenge of hybrid design......
Hugh
No kidding. Nelson kinda lowered the boom on me a while back--asked me to throw together a front end for "The Monster." The problem being that it's something like a thousand watt amp with a MOSFET rear end. Thatsa lotsa pF to load and unload! I ended up using 6550s in triode mode for the driver stage. No, I haven't heard it. To my knowledge, no one has ever built one. I built the front end to Nelson's specs for voltage swing and current, but it was strictly a bench job. I don't have an output section on that scale, and the guys who were the Official Builders haven't finished their portion as far as I know.
I've still got a few tubes lying around that need to be used up, but I haven't decided what I'm going to do with them. I've got some 6AS7s--might be fun to use them in a high current driver stage. (No, I don't have enough on hand to do a decent OTL.) Honestly, though, the odds are against me. I've never heard a hybrid circuit that justified its existence no matter who designed it, and I don't claim that I will be the one to slay that particular dragon.
Grey
I've still got a few tubes lying around that need to be used up, but I haven't decided what I'm going to do with them. I've got some 6AS7s--might be fun to use them in a high current driver stage. (No, I don't have enough on hand to do a decent OTL.) Honestly, though, the odds are against me. I've never heard a hybrid circuit that justified its existence no matter who designed it, and I don't claim that I will be the one to slay that particular dragon.
Grey
I decided not to try, myself. It is just too much of a tradeoff, whichever end you choose to be solid state.
A VERY interesting compromise is the static induction transistor. This are still made for military microwave use. An associate of mine made the equivalent of a 12AX7 that looked on a tube curver tracker and acted exactly like a 12AX7. The same company made 10A devices as well, BUT their package was very special and very expensive. They could have been put into a plastic power case IF we had enough money, up front for thousands of devices.
A really interesting design would have been a differential input with the '12AX7' equivalent with a folded cascode P channel high voltage mosfet pair and then the 10A devices directly connected to the output. No transformers, mostly tubes, direct coupled. Wow!
A VERY interesting compromise is the static induction transistor. This are still made for military microwave use. An associate of mine made the equivalent of a 12AX7 that looked on a tube curver tracker and acted exactly like a 12AX7. The same company made 10A devices as well, BUT their package was very special and very expensive. They could have been put into a plastic power case IF we had enough money, up front for thousands of devices.
A really interesting design would have been a differential input with the '12AX7' equivalent with a folded cascode P channel high voltage mosfet pair and then the 10A devices directly connected to the output. No transformers, mostly tubes, direct coupled. Wow!
Is that the one that uses little silicon peaks to serve as cathodes? If I recall correctly they spew electrons, but I don't remember what they ever did for a control element.
Grey
Grey
Some More Thoughts on THD/IMD Alternatives
I was experimenting with some time domain distortion analysis techniques and came up with a hybrid time/frequency domain approach that yields some interestng results. Furthermore, it preserves the large dynamic range afforded by frequency domain analysis.
Consider a source that generates a bursted sine wave signal, such as 5 ms of 1 KHz sine followed by 5 ms of zero signal. (Such a signal can be easily coded and converted to analog with a high resolution D/A.) This signal is then applied to the amp under test. The signal at the amp's output is then captured and digitized using a high resolution A/D.
The next step is to remove any DC offset from the captured output signal. This is easily done because there are 5 ms intervals where there is zero input signal being applied. It is then necessary to determine the delay through the amplifier for a 1 KHz tone and time shift the output data so it precisely overlays the input data. The input-output delay will be substantially less than the digitization time, so one will need to interpolate between data samples. The final post processing step is to remove all the 5 ms intervals where the data is zero, and reassemble the 5 ms intervals with nonzero data into a continuous data stream. The timing for this operation needs to be referenced to the input data, not the output data, and represents the burst gating signal. Rather than time shifting the output data, it is also acceptable to time shift the burst gating signal by an amount equal to the amp's in-out delay.
The final step is to plot the FFT of the resulting, continuous, sine wave signal. If the amplifier has ideal transient behavior, the spectrum for the reassembled burst data will be the same as for a continuous sine signal at the input. In other words, it will show the same THD components and nothing else. Otherwise, one will observe additional distortion components, some of which may not be harmonically related to the input signal.
I have tried this approach, via simulation, on both a preamp and a power amp circuit with dramatically different results. The preamp shows almost no difference between bursted vs. continuous signals, while the power amplifier shows a 30 dB increase in harmonic distortion products plus a substantial number of nonharmonic components.
Interestingly, the power amp's continuous and bursted distortion components converged for input swings less than 1.0 VPP. However, when simulating with a continuous sinewave I did not observe any THD change around the 1.0V level.
I'll post some plots in the next day or so. Comments are welcome.
I was experimenting with some time domain distortion analysis techniques and came up with a hybrid time/frequency domain approach that yields some interestng results. Furthermore, it preserves the large dynamic range afforded by frequency domain analysis.
Consider a source that generates a bursted sine wave signal, such as 5 ms of 1 KHz sine followed by 5 ms of zero signal. (Such a signal can be easily coded and converted to analog with a high resolution D/A.) This signal is then applied to the amp under test. The signal at the amp's output is then captured and digitized using a high resolution A/D.
The next step is to remove any DC offset from the captured output signal. This is easily done because there are 5 ms intervals where there is zero input signal being applied. It is then necessary to determine the delay through the amplifier for a 1 KHz tone and time shift the output data so it precisely overlays the input data. The input-output delay will be substantially less than the digitization time, so one will need to interpolate between data samples. The final post processing step is to remove all the 5 ms intervals where the data is zero, and reassemble the 5 ms intervals with nonzero data into a continuous data stream. The timing for this operation needs to be referenced to the input data, not the output data, and represents the burst gating signal. Rather than time shifting the output data, it is also acceptable to time shift the burst gating signal by an amount equal to the amp's in-out delay.
The final step is to plot the FFT of the resulting, continuous, sine wave signal. If the amplifier has ideal transient behavior, the spectrum for the reassembled burst data will be the same as for a continuous sine signal at the input. In other words, it will show the same THD components and nothing else. Otherwise, one will observe additional distortion components, some of which may not be harmonically related to the input signal.
I have tried this approach, via simulation, on both a preamp and a power amp circuit with dramatically different results. The preamp shows almost no difference between bursted vs. continuous signals, while the power amplifier shows a 30 dB increase in harmonic distortion products plus a substantial number of nonharmonic components.
Interestingly, the power amp's continuous and bursted distortion components converged for input swings less than 1.0 VPP. However, when simulating with a continuous sinewave I did not observe any THD change around the 1.0V level.
I'll post some plots in the next day or so. Comments are welcome.
john curl said:
Sure. But then again, I recall Bob Cordell's comparative listening tests at the 2006 RMAF between a tube and ss amplifier. IIRC, it was pretty impossible to hear which was which, UNLESS one of the amps was clipping (which mostly was the tube amp as it was lower power). That is also a big giveaway IMHO.
Jan Didden
Re: Some More Thoughts on THD/IMD Alternatives
Most ingeneous, I'm looking forward to the results. I wonder how different this test is, in exercising the amp, from that earlier test using a square-wave with superimposed higher-frequency sinewave signal?
Jan Didden
analog_guy said:
Most ingeneous, I'm looking forward to the results. I wonder how different this test is, in exercising the amp, from that earlier test using a square-wave with superimposed higher-frequency sinewave signal?
Jan Didden
Hi, Analog_guy,
Very nice concept 😀 Do you have graphs for comparison between preamp and power amp behavior?
Very nice concept 😀 Do you have graphs for comparison between preamp and power amp behavior?
No, I haven't heard that one. Bear in mind that I live in a poor part of the country and high end is pretty much a rich man's game. As such, high end dealers are pretty sparse in the Southeastern US. These days I hear new equipment mainly if I happen to go on a trip. Then you're pretty much at the mercy of whatever equipment they happen to have hooked up at the time.
(The most extreme example of this was in a shop in Santa Fe, New Mexico, where they had a gargantuan Krell amp hooked up to some no-name thousand dollar speakers in a room that looked more like a stock room [boxes piled higgledy-piggledy] than a place where you could actually listen. Made me wonder what they were thinking. If I were Krell, I'd give them a stern talking-to about demonstrating my gear under such conditions.)
I have to say that I've heard so many disappointing hybrid pieces--even from people who know how to do tubes, like conrad johnson--that I don't exactly get sweaty palms when I hear about another. I know the Lamm stuff exists, but I'm not planning a road trip just to hear it.
Grey
(The most extreme example of this was in a shop in Santa Fe, New Mexico, where they had a gargantuan Krell amp hooked up to some no-name thousand dollar speakers in a room that looked more like a stock room [boxes piled higgledy-piggledy] than a place where you could actually listen. Made me wonder what they were thinking. If I were Krell, I'd give them a stern talking-to about demonstrating my gear under such conditions.)
I have to say that I've heard so many disappointing hybrid pieces--even from people who know how to do tubes, like conrad johnson--that I don't exactly get sweaty palms when I hear about another. I know the Lamm stuff exists, but I'm not planning a road trip just to hear it.
Grey
GRollins said:
This is baloney on so many levels it is almost impossible to figure out where to begin.
Do you just make this suff up as you go along?
Grey, can we see some of your designs and expertise in action? sounds like you've got t his feedback thing solid state/tube issue all sorted out.
GRollins said:
Of course it changes the sound! Doing so, you go through a whole spectrum of different amplifiers, and each one has it's own signature. None of them are transparent, unless you have enough loop gain to let the feedback do it's job, which you never have in a tube amp. Face it, whatever you do to that tube amp, it will impart a signature on the sound. You may like that, nothing wrong with that, but it aint transparent, son.
I never thought I'd see a post proving that feedback works, being used against it. As Glen noted, this is so misleading as to defy any meaningfull discussion.
Jan Didden
- Status
- Not open for further replies.