Okay, soliciting opinions.
A few years back, I had Quad ESL-63 speakers, driven by VTL 75/75 amps. I had converted the amps to triode, upgraded coupling caps to some exotic brands, put all Vishay resistors in the circuit. Bypassed the power supply caps with some exotics. Used very expensive tubes in the KT-88 configuration. I thought it the most musical amp ever, and several experienced audiophiles concurred.
Best SS amps I had ever had were Spectrals. Had a DMA-50 on hand. Not the equal of the VTL, but quite nice, enjoyable SS amp.
I began wondering what more could be done to improve things. All options left were complex and expensive. I wondered.....am I close to perfection or far away. Is the VTL with mods etc. 95% of possible or only something like 50%? I considered the Spectral to be subjectively about two thirds as good as the VTL.
Decided I could figure it out with inserting a device in series and seeing how much degradation occurred. The VTL was much more musical, natural, spacious, 3D, more resolving etc. etc. So would it be close to perfection (straight wire with gain vs. the source) or far from it. Decided I would series devices to find out.
Loaded the VTL down with power resistors and then tapped the result with a voltage divider network so the overall result was unity gain (or as close as possible with the Fluke multimeter to 3 decimal places). Had this unity gain VTL feed the Spectral DMA-50 input driving ESL-63's. I expected to find out how much the VTL colored the signal vs. a direct connection to the Spectral. My expectation was the VTL was doing little wrong as it was so spacious, detailed, and enveloping when driving the speakers directly.
So imagine my surprise when feeding a CD out to the VTL, and then to Spectral amps that I heard the VTL sound or at least 99% of it. Spacious, detailed, musical, enjoyable, and dynamic. What was going on here? How could the musically inferior, lower rez Spectral manage this? Well reversing the amps, having the Spectral loaded with power resistors, then feeding the VTL at unity gain, I heard no difference. I mean, Spectral in vs Spectral out, I simply could hear nothing. The Spectral should have been true to source. It had exemplary signal to noise, more than 1 megahertz bandwidth, high slew rate, low distortion etc. I simply couldn't hear it in or out of circuit.
Left me with one conclusion: Yes, tubes were subjectively much more enjoyable, more resolving, more musical.....and it was all due to a coloration. Delicious, seductive, but not at all transparent. The Spectral was very transparent. Tubes were a delicious coloration.
I tried the same experiment with a few other SS amps. Not many could equal the Spectral. Some large Muse amps, sounded 85% like the VTL, but overlaid it with a bit of MOSFET mist, and noise. A Electron Kinetics overlaid it with a bit of edgy quality, though only in the treble and most sounded 75% subjectively like the VTL's.
So what other conclusion could one make other than tubes were a coloration, delightful and nice, but a coloration all the same. I think the belief that tubes are truer to source have mislead much of the audio world. I don't have a problem with those who say if it sounds better it is, but do realize it is a coloration, and not true to source. Don't then base your design theories on a mistaken notion.
Oh well, just wondering what comments folks would have on this.
FYI, I currently use a digital amp, which seems even more true to source, and fully transparent than the Spectral of yore. I think we have an unacknowledged golden age of audio now. But people are pursuing an illusion instead.
A few years back, I had Quad ESL-63 speakers, driven by VTL 75/75 amps. I had converted the amps to triode, upgraded coupling caps to some exotic brands, put all Vishay resistors in the circuit. Bypassed the power supply caps with some exotics. Used very expensive tubes in the KT-88 configuration. I thought it the most musical amp ever, and several experienced audiophiles concurred.
Best SS amps I had ever had were Spectrals. Had a DMA-50 on hand. Not the equal of the VTL, but quite nice, enjoyable SS amp.
I began wondering what more could be done to improve things. All options left were complex and expensive. I wondered.....am I close to perfection or far away. Is the VTL with mods etc. 95% of possible or only something like 50%? I considered the Spectral to be subjectively about two thirds as good as the VTL.
Decided I could figure it out with inserting a device in series and seeing how much degradation occurred. The VTL was much more musical, natural, spacious, 3D, more resolving etc. etc. So would it be close to perfection (straight wire with gain vs. the source) or far from it. Decided I would series devices to find out.
Loaded the VTL down with power resistors and then tapped the result with a voltage divider network so the overall result was unity gain (or as close as possible with the Fluke multimeter to 3 decimal places). Had this unity gain VTL feed the Spectral DMA-50 input driving ESL-63's. I expected to find out how much the VTL colored the signal vs. a direct connection to the Spectral. My expectation was the VTL was doing little wrong as it was so spacious, detailed, and enveloping when driving the speakers directly.
So imagine my surprise when feeding a CD out to the VTL, and then to Spectral amps that I heard the VTL sound or at least 99% of it. Spacious, detailed, musical, enjoyable, and dynamic. What was going on here? How could the musically inferior, lower rez Spectral manage this? Well reversing the amps, having the Spectral loaded with power resistors, then feeding the VTL at unity gain, I heard no difference. I mean, Spectral in vs Spectral out, I simply could hear nothing. The Spectral should have been true to source. It had exemplary signal to noise, more than 1 megahertz bandwidth, high slew rate, low distortion etc. I simply couldn't hear it in or out of circuit.
Left me with one conclusion: Yes, tubes were subjectively much more enjoyable, more resolving, more musical.....and it was all due to a coloration. Delicious, seductive, but not at all transparent. The Spectral was very transparent. Tubes were a delicious coloration.
I tried the same experiment with a few other SS amps. Not many could equal the Spectral. Some large Muse amps, sounded 85% like the VTL, but overlaid it with a bit of MOSFET mist, and noise. A Electron Kinetics overlaid it with a bit of edgy quality, though only in the treble and most sounded 75% subjectively like the VTL's.
So what other conclusion could one make other than tubes were a coloration, delightful and nice, but a coloration all the same. I think the belief that tubes are truer to source have mislead much of the audio world. I don't have a problem with those who say if it sounds better it is, but do realize it is a coloration, and not true to source. Don't then base your design theories on a mistaken notion.
Oh well, just wondering what comments folks would have on this.
FYI, I currently use a digital amp, which seems even more true to source, and fully transparent than the Spectral of yore. I think we have an unacknowledged golden age of audio now. But people are pursuing an illusion instead.
I'd say your setup is about 99.5%, but some folks will argue that the 0-99% range is either void or irrelevant and therefore your current setup is only 50%.
There's two three things to keep in mind in (DIY) audio:
1. Performance is subjective.
2. The law of diminishing returns applies here like in no other field.
3. As a synthesis of 2 & 3: the larger the investment in time & money you make in achieving a marginal improvement, the larger your own, subjective gain in quality is likely to be.
From the above follows a simple rule I live by as a DIY-er: "tinker, and be happy".
Valve amps don't have to colour the sound. It just happens to be the case that people who prefer coloured sound also prefer valve amps. This is probably because for a given level of electronic expertise it is easier to build a reasonably good valve amp than a reasonably good solid-state amp. Both will be coloured, but the valve colour will be more pleasant to listen to than the SS colour. Most valve lovers deny this vehemently. I don't.
Having built their reasonably good valve amp, which sounds better than a reasonably good SS amp, they then believe that the colour (whose presence they usually deny) is 'better' than accuracy. They get used to this coloured sound, so when they hear uncoloured sound (perhaps from a much better SS amp) they don't like it. This has been known for decades, yet is still routinely denied. You can build a valve amp which does not noticeably colour the sound, but they would reject this as sounding too 'solid-state' or 'clinical'.
The solution is simple. People should say "I prefer the sound of X amp", rather than "X amp is better".
most of my tube-o-holic friends like a certain amount of coloration in the sound. Not "Dynaco PAS" levels, but a touch of "bloom and life" as they call it. I'm more in the transparent/detailed camp which leads to never-ending arguments of what is more 'realistic'.
I've heard some extremely colored systems - like a pair of 300B PP amps running Altec VOTTs - and there was this big golden hue to everything we played, no matter what the recording quality was like. To me that is coloration. To my friend, it was pure heaven. He enjoyed it and more power to him.
I've heard some extremely colored systems - like a pair of 300B PP amps running Altec VOTTs - and there was this big golden hue to everything we played, no matter what the recording quality was like. To me that is coloration. To my friend, it was pure heaven. He enjoyed it and more power to him.
I have to agree with this statement. Although valves-mainly triodes, produce 2nd harmonic,semiconductors produce 3nd and this could be a fundamental diference. But again a triode could be far more linear than a pentode and someone doesn't have to use FB at all.This perhaps gives a more naturall sound.
It's been very surprising to hear to my 845 amplifiers and find to be neutral,except a more liquid,tube-like bass,which is very pleasant.
The same occurs with the amplifiers of a friend of mine,using 845s too.
In generall, I believe that this matter has to do a lot with the approach of the designers and someone could be lead to different results in any case.
And of course,tubes have become a fashion in today's market,so many companies are into that, producing devices with profit in mind.
Triodes roughly follow a 3/2 power law. This contains all harmonics, although the 2nd dominates. With a sufficiently high anode load, local feedback (also 3/2 power) cancels the grid distortion so you get reasonable linearity. Most circuits do not have a sufficiently high anode load. Pentodes also have a 3/2 law, but no anode feedback to correct it. (Adding external feedback using resistors does not solve this problem - you can't turn a pentode into a triode that way).
BJT follow an exponential law quite closely. This contains all harmonics, although the 2nd dominates. Alternatively, if used as current amplifiers they can be reasonably linear over a limited range but beta is so poorly controlled that you can't easily design this way.
FET are roughly square-law, so mainly 2nd.
The net result is that if you want to build an amp without feedback then you are more or less forced to use triodes, rely on their internal feedback, and get used to hearing colouration. Any other technology requires the careful application of the right amount of feedback. Using the wrong amount of feedback, or applying it badly, creates problems. Getting it wrong is surprisingly common!
BJT follow an exponential law quite closely. This contains all harmonics, although the 2nd dominates. Alternatively, if used as current amplifiers they can be reasonably linear over a limited range but beta is so poorly controlled that you can't easily design this way.
FET are roughly square-law, so mainly 2nd.
The net result is that if you want to build an amp without feedback then you are more or less forced to use triodes, rely on their internal feedback, and get used to hearing colouration. Any other technology requires the careful application of the right amount of feedback. Using the wrong amount of feedback, or applying it badly, creates problems. Getting it wrong is surprisingly common!
Maybe this is the case of overloading.
Why Tube Amplifiers : Ayon Audio
Regarding the FETs, I agree since they are voltage controlled just like the tubes.
Why Tube Amplifiers : Ayon Audio
Regarding the FETs, I agree since they are voltage controlled just like the tubes.
That Ayon Audio page contains a lot of misleading woffle. It is stuff like that which gets valves a bad name among real engineers. They assume we are all that ignorant.
For example:
For example:
When anything does symmetric clipping, the dominate distortion product is the third. That is a matter of mathematics, not technology. It includes valves.
Not true. Both can be characterised as either - Norton and Thevenin say so. Almost all precision use of transistors (including the input LTP in a SS amp) relies on their transconductance characteristics, not current amplification.
I agree. The theoretical ideal amp is a straight wire with gain. Everything else just has some degree of distortion.
So, it boils down to what type of distortion you prefer or if you prefer no distortion.
However, those that enjoy distortion in their amps rather call it colorization or musical or some other subjective unmeasurable term.
In the end, build/buy what you like and enjoy it .
A word of caution. Harmonics do not necessarily = coloration. The ear does not work that way.
Harmonics can mask other harmonics to the extent that you simply don't hear them, even at elevated levels - it sounds like a pure sine wave. This was discovered over 90 years ago. Some types of distortion are simply inaudible.
Sure, I've heard very lush, warm tube amps. A lot of fun. But not all sound like that, in fact many don't.
Harmonics can mask other harmonics to the extent that you simply don't hear them, even at elevated levels - it sounds like a pure sine wave. This was discovered over 90 years ago. Some types of distortion are simply inaudible.
Sure, I've heard very lush, warm tube amps. A lot of fun. But not all sound like that, in fact many don't.
"Using the wrong amount of feedback, or applying it badly, creates problems. Getting it wrong is surprisingly common! "
"I think that's also the reason why triode amps with no feedback are so popular in this forum"
The bandwidth of the typical low cost tube output transformers is too low to support any decent global feedback these days. One is forced to use local feedback for pentodes, with maybe 6 dB of global feedback. The more expensive transformers still don't give you any design data, just expensive, useless, hype driven perks like special core material or silver wire. Since when has anyone seen a long E lamination output xfmr. for bandwidth, they don't exist anymore. Price drives demand, demand drives the parts, quality has descended to WalMart or Made in China levels. So tube amps will remain triode and "colored" unless you custom design everything.
Then there is the trend of offering glamorous (expensive selling price driven) transmitting tube amps which require such high impedance output transformers (and deadly dangerous B+) that they hardly can even pass the audio band itself. The extreme reliance on flashy bling without any substance is the typical harbinger of a bubble about to collapse.
"I think that's also the reason why triode amps with no feedback are so popular in this forum"
The bandwidth of the typical low cost tube output transformers is too low to support any decent global feedback these days. One is forced to use local feedback for pentodes, with maybe 6 dB of global feedback. The more expensive transformers still don't give you any design data, just expensive, useless, hype driven perks like special core material or silver wire. Since when has anyone seen a long E lamination output xfmr. for bandwidth, they don't exist anymore. Price drives demand, demand drives the parts, quality has descended to WalMart or Made in China levels. So tube amps will remain triode and "colored" unless you custom design everything.
Then there is the trend of offering glamorous (expensive selling price driven) transmitting tube amps which require such high impedance output transformers (and deadly dangerous B+) that they hardly can even pass the audio band itself. The extreme reliance on flashy bling without any substance is the typical harbinger of a bubble about to collapse.
Last edited:
In practice though, it's unavoidable. Lots of coupling caps, output transformers, microphony, noisy and poorly regulated power supplies for the low signal stages - you don't even need to add distortion to the mix to know it's coloured. When synergy happens tube amps can sound magical but they're still coloured.
I remember reading something similar to 931410's experience on 6moons. An F amp driven by a 300B SET amp was apparently the bees knees.
Indeed, the most "fast" and "resolving" amps I've heard have all been tube based. If you haven't heard a really good OTL or SET (45s in particular can be ludicrously good), then any "warm thoughts" of tube amps may need revision. Since there's so much variance within tubes/ss, cd/vinyl, etc, I tend to default to what the very best sounds I've heard used.
Glass and Wax still are the heavyweight/high budget champs in my book.
Now this is very interesting, especially since you recently pointed out something that hadn't occurred to me before, but made perfect sense: this cap is an excellent conduit for RF picked up at the output, especially for amps with relatively high source impedances. I've always been of the Crowhurst school of, "Don't worry about that last few hundredths of a dB at 20kHz or the last 0.05% of distortion at 15kHz, get the amp stable without a lead network by compensation." I'd be curious about how you made the decision that the advantages of a lead cap outweighed the potential disadvantage of RF injection to the input.
The cap is only across part of the feedback resistor, so RF appearing at the speaker terminals is still attenuated at the input stage cathode. (This was a modified Mullard 5-20, but with half an ECC81 as input stage instead of EF86.) I intended to use a step network in the 81 anode, but found that the amp was quite stable with no HF peaks without it. I guess the ECC83 phase splitter had enough HF cut to stabilise the amp.
The original Mullard design (and the Leak etc. clones) used a step network, but it had higher open-loop gain and more feedback. The original design was possibly more marginal at the LF end too. There is a subsonic bump in the published Mullard frequency response!
You mention source impedance. My guess is that at RF frequencies most valve amps have a highish impedance because even if the OPT was still functional at these freqs, which it probably isn't, the feedback loop certainly isn't. A triode output would still be clobbered by the OPT, so only amps with a Zobel network would have a low impedance at RF. Few valves amps use a Zobel network.
The original Mullard design (and the Leak etc. clones) used a step network, but it had higher open-loop gain and more feedback. The original design was possibly more marginal at the LF end too. There is a subsonic bump in the published Mullard frequency response!
You mention source impedance. My guess is that at RF frequencies most valve amps have a highish impedance because even if the OPT was still functional at these freqs, which it probably isn't, the feedback loop certainly isn't. A triode output would still be clobbered by the OPT, so only amps with a Zobel network would have a low impedance at RF. Few valves amps use a Zobel network.
That link suffers from lack of specificity. The quote below is an example.
"and it has been well established that in tube amplifiers the onset of clip/overload as maximum power is reached is gradual and rising distortion is of predominately low even-order harmonic nature."
Here's a measurement of 2 tube amps. The difference in harmonics is not because it's a comparison between tube and transistor amps but because the circuit typology.
http://img221.imageshack.us/img221/3214/setvspptubamp.jpg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.