With all this talk of the imperfections of what appears to be a simple device - an amplifier - and the apparent problems of feedback, I'm surprised that no one mentions the idea of real feedforward error correction i.e. learning exactly what a low-ish feedback amplifier is going to do, and pre-distorting the input to compensate. I presume that this would only be applicable to a specific load, and ambient temperature etc. but would get you closer to perfection without the evil feedback. If you were prepared to countenance the devil's own digital, you might do it with, effectively, a neural network that is trained to massage the signal non-linearly in many dimensions rather than just a simple FIR filter. I rather like the idea of connecting a neural network to wheezing steam-powered vacuum tubes.
In some, yes, others, no. IMO (no proof), this is likely a major factor in the "sound" of DHT amps.
It might be due to microphonic tubes or because of difference of an assumed behavior of a dummy load and real speakers.
For example I am seriously skeptical about electrostatic panels or Maggies powered by a Graaf that would roll off at 20KHz. As well I am seriously skeptical about audiophile that will use Graaf gear with inferior speakers limited up to 20KHz.
However no doubt that perfect 300B SET sound could be easily killed by a wrong loweff solid state oriented speakers especially in bass octaves. In contra hieff Fostex with oscillated ssamp would be an insane headache and in most cases with hi-fi consumer ssamps Fostex or Klipsch is a headache.
For example I am seriously skeptical about electrostatic panels or Maggies powered by a Graaf that would roll off at 20KHz. As well I am seriously skeptical about audiophile that will use Graaf gear with inferior speakers limited up to 20KHz.
However no doubt that perfect 300B SET sound could be easily killed by a wrong loweff solid state oriented speakers especially in bass octaves. In contra hieff Fostex with oscillated ssamp would be an insane headache and in most cases with hi-fi consumer ssamps Fostex or Klipsch is a headache.
@SY
What sort of levels of microphony are we talking about? People round here can readily discern distortions and noise at 100dB below the signal (so they say) and from my own experience of shouting at a 'valve' (as we call them in the UK) and recording the result in the past, microphony could be much higher than that. Presumably it's not always harmonic, and not linear..?
What sort of levels of microphony are we talking about? People round here can readily discern distortions and noise at 100dB below the signal (so they say) and from my own experience of shouting at a 'valve' (as we call them in the UK) and recording the result in the past, microphony could be much higher than that. Presumably it's not always harmonic, and not linear..?
Your parenthetical is spot on.
If tapping on various spots on my amplifier produces no discernible sound from the speakers, I cease to worry about this as an error source. The microphonics from DHTs that I've tapped peak right in the brightness range, thus my speculation.
Isn't it a strange coincidence that the point beyond which amplifiers start to behave unacceptably, apparently, is at the upper limit of human hearing? The same electronic components that are used for processing signals in the many megaherz or gigaherz ranges (i.e. nothing to do with audio) can only just be configured to reproduce a complex waveform whose components are all within a bandwidth of 20kHz, or so it seems. It's a good job that the physics of air, and the laws of biology mean that our ears evolved the way they did.
Or is it the case that even if our ears ran out at 10Hz, people would still manage to make a living telling us that our amplifiers are rubbish, and that they can hear huge differences between valves and transistors, and that circuits without feedback are so much better than those with?
Or is it the case that even if our ears ran out at 10Hz, people would still manage to make a living telling us that our amplifiers are rubbish, and that they can hear huge differences between valves and transistors, and that circuits without feedback are so much better than those with?
I would say some as the various methods effect where on the curve the tube runs and all the interrelated effects that argument or cancel each other. Hint: "curve" I have only built a couple of tube amps on a junk Chinese 6P1 chassis and my only textbook is the one from Jones, so I may be missing something.
Going back to find hints as why one may prefer one technology or another, what do you think about compression fooling one to think there is more detail? Just like how many JBL speakers jumped off the shelf in the old L100 days because of their pronounced mids. A/B test, WOW factor et al.
To earlier comment on price. Granted, at any price point, good design is still good design. Unfortunately, pretty quick the designer has to decide to spend real money. Bigger transformers, regulators, replacing resistors with ccs'a and so on. Good mechanical build gets expensive too. I am sure those out there who design commercial products can attest how hard it is to bring a concept in and still reach a viable market. That's the other 90% of engineering.
Once these things come together they exceed my range of excuses. All of this is luxury. None of us need any of this stuff. I am perusing these thoughts as a hobby, to learn, to understand. Besides, I sold my TVR so I have to have some trouble to get into.
Thanks John on reminding me how old some of those amps are. It would be a far better shake to get hold of a much newer one. One from this century.
Coppertop, the REAL answer is to listen for yourself, and to trust your own responses to different amplifier designs. It is true that we audio designers do NOT completely understand what is most important in amplifier design, and I personally have striven to understand it for more than 40 years.
My personal philosophy is to try to cover EVERYTHING that I know is potentially an 'error' in an amplifier's design. So, I will start with the most linear circuit topology that I can think of. Then, I will run it as close to Class A as the heatsinks available will allow. Then, I will make sure that the circuit is fast, at least 100V/us, often faster, and stable with virtually any capacitive load.
Then, (in short form) I will make sure that the amp has lots of peak drive current capability, 30A min, 60A typical.
Finally, I test for the worst case crossover point in the transfer function and monitor the 7th harmonic distortion generated. This is at the transition between Class A and Class AB (about 25W in the JC-1).
Am I completely successful? No, but I do pretty good. There are people of my acquaintance who have owned and sold their JC-1's for another amp. One of the other amps is the AYRE. It is ALWAYS a wake-up call when this happens.
Now what advantages does Charles Hansen of Ayre, have over my Parasound designs?
First, he has COMPLETE control over his design. I am limited to circuit design, getting a proper circuit layout and most of the obvious active and passive parts.
I have little control over internal connectors, housekeeping circuits, protection, RFI proofing the power supply, being stuck with a huge toroid transformer, rather than a super-quality EI or R core transformer. I'm not complaining exactly, I get to control the bulk of the design, but IF I personally made the amp, myself, it would cost at least 3 times as much and sound only a little bit better, on some occasions. In this case, the associated equipment and audio sources would dominate any decision as to an audio difference. But heck, I could pick the exact brand of solder, even, just like I did with the CTC Blowtorch, that just about everybody likes. '-)
My personal philosophy is to try to cover EVERYTHING that I know is potentially an 'error' in an amplifier's design. So, I will start with the most linear circuit topology that I can think of. Then, I will run it as close to Class A as the heatsinks available will allow. Then, I will make sure that the circuit is fast, at least 100V/us, often faster, and stable with virtually any capacitive load.
Then, (in short form) I will make sure that the amp has lots of peak drive current capability, 30A min, 60A typical.
Finally, I test for the worst case crossover point in the transfer function and monitor the 7th harmonic distortion generated. This is at the transition between Class A and Class AB (about 25W in the JC-1).
Am I completely successful? No, but I do pretty good. There are people of my acquaintance who have owned and sold their JC-1's for another amp. One of the other amps is the AYRE. It is ALWAYS a wake-up call when this happens.
Now what advantages does Charles Hansen of Ayre, have over my Parasound designs?
First, he has COMPLETE control over his design. I am limited to circuit design, getting a proper circuit layout and most of the obvious active and passive parts.
I have little control over internal connectors, housekeeping circuits, protection, RFI proofing the power supply, being stuck with a huge toroid transformer, rather than a super-quality EI or R core transformer. I'm not complaining exactly, I get to control the bulk of the design, but IF I personally made the amp, myself, it would cost at least 3 times as much and sound only a little bit better, on some occasions. In this case, the associated equipment and audio sources would dominate any decision as to an audio difference. But heck, I could pick the exact brand of solder, even, just like I did with the CTC Blowtorch, that just about everybody likes. '-)
I've always wondered about this. Could the heat be part of the charm of Class-A transistor amps? Especially those with the driver transistors on the same heatsink as the output devices. Just wondering.
That has nothing to do with compression. I'm sorry, there's absolutely nothing inherent in tubes that results in compression (or high distortion or low damping or...).
You can stop wondering. If it were thermal distortion, you'd see an increase with decreasing frequency. You don't.
Heat may indeed be part of the charm, but it's not due to electrical signals.
If "typical" means those amps that are missing proper cathodes and half their output stage, then compensate by throwing iron at it, then yes, you might be right.
Hi,
It may help, or not depending on application, however only for static conditions.
There is so much thermal resistance between the silicone chip and any heatspreader or just the envoironment, by the time a common heatsink can react, the signal is long past. Even having everything on the same die is no guarantee...
Ciao T
It may help, or not depending on application, however only for static conditions.
There is so much thermal resistance between the silicone chip and any heatspreader or just the envoironment, by the time a common heatsink can react, the signal is long past. Even having everything on the same die is no guarantee...
Ciao T
Pano, you only have 1/2 the story. Class A is not necessarily the most thermally stable operating condition. It can be shown that heavy Class AB-1 is as good or better, if you go back to the root equations. However, thermal stability, both transient and long term is really important, and real designers make sure that the THERMAL CAPACITANCE is very high by close mounting the output devices on good insulators (if necessary) properly torqued down and used with thermal grease on a substantial heatsink . This removes most of the short term gain changes due to temperature.
High input stage operating current, using only a part of the class A portion can also be helpful.
High input stage operating current, using only a part of the class A portion can also be helpful.
I'm sure the difference would be load dependent, well based on the Stereophiles test results, JC1 vs the Ayre.
Yes thermal tracking IMO does cause a change in sound with SS amps.
Class-a sound is more consistent due to the high bias and the heat associated with it. All SS amps sound different and require some warming period for there sweet spot, due to thermal tracking they very rarely stay within this sweet spot during use, unless sweet spot temps are very well controlled.
Class-d suffers very little from this, so once in there operating sweet spot there sonic character changes very little , you either like it or not. Class-a gets similar response , just at the other end of the temperature spectrum.
Ideally the amp should get up to it's sweet spot temp and stay there, this is difficult unless designed for some specific load. Matching and oversized heatsinks pays dividends here, but doesn't erradicate it.
of course your mileage may vary .....
Well not according to the test performed by Atkinson , heat did make a difference to distortion, power and bandwidth.
- Status
- Not open for further replies.