I must admit I did find some of these things, shall we say, unusual?
I agree with your comments for sure. But, I would have to say that all of these variables must add up to the tubes acting differently as circuit elements. That is, because they are made differently, they must exhibit different plate curves or different capacitances to sound different. If they don't, then they must behave electrically the same and sound the same.
I would say even further, that it is possible to make two tube with different internal structure to have the same plate curves and capacitances. They will look different, but it seems to me that they must, nevertheless, sound the same. No?
This is a good point though, which leads me to ask: how much different can a tube be from its "nominal" characteristics before it is no longer that tube.
In other words, if I'm building a 12AX7 don't I have to try to hit the nominal plate curves and capacitances (within some acceptable delta)? If I don't and I miss by a lot, then can I claim to be manufacturing a 12AX7? (Although I might anyway).
Now, if part of the tuberolling business is selecting out tubes that are grossly improperly manufactured, then I can see why folks would do it. But then, they're not rolling tubes of the same type anymore, they're selecting from several different types of tubes with the same label to find the one for which the circuit was originally designed. This I believe.
Now we talk about sound
Tubes have no sound, only systems employing circuits using tubes can be said to have a sound. And if the curves are the same, the noise is the same, and the microphonics are the same, exchanging tubes will not result in the sound of the system changing. If those factors are different, there may or may not be a change in the sound of a system, depending on overall design.
Aha! A Platonist!
Tubes have no sound, only systems employing circuits using tubes can be said to have a sound. And if the curves are the same, the noise is the same, and the microphonics are the same, exchanging tubes will not result in the sound of the system changing. If those factors are different, there may or may not be a change in the sound of a system, depending on overall design.
Aha! A Platonist!
Folks, my experience of tube rolling is very limited but also very recent (within the last month).
I had the oppourtunity to visit a friend running a Sony SACD player into a custom built preamp using 6CG7's, then into a custom built single ended power amp using 13ei's as the output tubes. Speakers were Vienna Acoustics (Mozarts I believe).
The preamp used an offboard power supply and all equipment was on racks.
When I arrived the preamp was using a pair of vintage Mullard 6CG7's, later they were changed to vintage telefunken 6CG7's and then to a apir of new Electroharmonix 6CG7's.
These were the only changes made to the system and the tubes were subbed back and forth a number of times on a range of music (Kind of Blue being the main reference disc but female vocal used also)
The differences were not subtle.
To broadly categorise:
The Mullards had a more smooth and pleasant midrange, slightly rolled off treble and a bass region that was both slightly recessed and loosely controlled. a rea "classic tube" sounding tube.
The Telefunkens were tight, controlled and precise. Certainly "brighter" in sound with a far more precise bass. On some recordings they could be accused of sounding harsh. a more intellectual and less emotional listening experience than the Mullards.
The Electroharmonix were a real suprise, especially as some of their more recent concoctions have recieved a fair amount of praise. Bass was extremely overblown and substantially uncontrolled, midrange and treble detail was muddied. Very first initial impressions following on from the Telefunkens were that we had returned to a musical sounding tube but this initial impression quickly changed to a feeling that something sounded very wrong.
Remember that plate curves only show steady state conditions.
How a pair of plates spaced apart with electrons wafting between them can exhibit an altered frequency response i cannot comprehend, all I can say is that they categorically do.
Apologies for the lack of scientific data to add to the discourse apart form my subjective impressions.
drew
I had the oppourtunity to visit a friend running a Sony SACD player into a custom built preamp using 6CG7's, then into a custom built single ended power amp using 13ei's as the output tubes. Speakers were Vienna Acoustics (Mozarts I believe).
The preamp used an offboard power supply and all equipment was on racks.
When I arrived the preamp was using a pair of vintage Mullard 6CG7's, later they were changed to vintage telefunken 6CG7's and then to a apir of new Electroharmonix 6CG7's.
These were the only changes made to the system and the tubes were subbed back and forth a number of times on a range of music (Kind of Blue being the main reference disc but female vocal used also)
The differences were not subtle.
To broadly categorise:
The Mullards had a more smooth and pleasant midrange, slightly rolled off treble and a bass region that was both slightly recessed and loosely controlled. a rea "classic tube" sounding tube.
The Telefunkens were tight, controlled and precise. Certainly "brighter" in sound with a far more precise bass. On some recordings they could be accused of sounding harsh. a more intellectual and less emotional listening experience than the Mullards.
The Electroharmonix were a real suprise, especially as some of their more recent concoctions have recieved a fair amount of praise. Bass was extremely overblown and substantially uncontrolled, midrange and treble detail was muddied. Very first initial impressions following on from the Telefunkens were that we had returned to a musical sounding tube but this initial impression quickly changed to a feeling that something sounded very wrong.
Remember that plate curves only show steady state conditions.
How a pair of plates spaced apart with electrons wafting between them can exhibit an altered frequency response i cannot comprehend, all I can say is that they categorically do.
Apologies for the lack of scientific data to add to the discourse apart form my subjective impressions.
drew
Appearances can be deceptive
Compared with the norm, I've seen some valves of alternative construction, that were electrically identical, while others that were visually identical were way off.
I think you can only go so far on this though. As frequencies rise, C & L both become more significant. There's not much "cosmetic" variation in valves designed for UHF and up. There are also fewer "equivalents".
runeight said:
Compared with the norm, I've seen some valves of alternative construction, that were electrically identical, while others that were visually identical were way off.
I think you can only go so far on this though. As frequencies rise, C & L both become more significant. There's not much "cosmetic" variation in valves designed for UHF and up. There are also fewer "equivalents".
Not accepted. This thread is not about sound at all. It's about discussing why there cannot be, isn't and never will be any audible difference between tube brands. The main topic movers happily confirm and prove to each other the abscence of audible differences and now you dare spoil the scientific purity with your anecdotal observations?! If you're not interested in discussing the fine points of watching paint dry you have nothing really to contribute here.
Does that mean I can save heaps of money by having them prove to me that greencaps sound the same as teflon/tinfoil multicaps? Cool, think of all the cheap electros and bell wire I'll be able to buy to finish off my system.
In fact, if I stopped listening and just looked at published specs, audio would be so much cheaper.
Yay! I'm enlightened now!
Drew
On the contrary. I found snippets like the one bellow better fun than any magic.
I am still to meet an objectivist who is willing to listen 'hard'. This one at least honestly admits to listening laziness.
Most Objectivists are too busy reading Ayn Rand to spend any time messing with audio, so it's kind of a moot point.
🙂 🙂 🙂 🙂 So richly deserved!! What exactly was I trying to say? Who am I? And why am I here?
Since I started this thread, I must say that I don't feel this way. As I mentioned in another thread my background and, perhaps, temperment don't tend in this direction. In the late 19th and early 20th centuries the physicists believed that they had everything understood. All they had to do was to apply the theories that they had developed to the world that they knew.
Within the first decade of the 20th century Max Planck and Albert Einstein blew their self-satisfied universe apart with quantum theory and relativity. Both of these phenomena are totaly contrary the intuition and evidence of our senses. But they are very real.
In approaching the thread the way I did, I was trying to eliminate the things that I understand to try to get to the things that I don't understand about tubes.
I am not saying that this isn't true, only that I don't understand how it can be true. And if you all are hearing these differences, please help me to understand what is happening.
Yes, I mostly agree with this. But I agree because I think that what are normally published as specs don't really contain enough data about how the equipment really works. For example, THD is a reasonable measure, but maybe not as relevant as IMD or Phase Distortion or even something else that we don't yet understand. But, what is it?
And it may be that an auditory experience is very subjective. It may be that for each individual any particular listening experience is different from other listening experiences and that this changes with our state of mind at any given time. I am willing to believe this, not as an "it's all in your mind" phenomena that is easily dismissed, but as a genuine part of the human cognitive system that makes listening to music a richer experience because of it.
But I do admit to trying to think about here whether we are talking about measurable differences among different brands of tubes or something else. And if something else, I really want to know what that is. Can you folks offer me some ideas? And I am not setting you up for ridicule (at least from my perspective).
Runeight
At some level this undoubtfully is true, but there is little point going there. Most listeners seem to agree upon the effects of 'tweaks' so the experience is not so individual most of the time.
May i suggest a personal, first hand listening experience as an essential starting point for your journey. You only require a very simple audio chain to conduct meaningful tests and of course, certain listening abilities. If listening to the same track 50 times in a row proves too boring than maybe you're not too well suited for this position 🙂 I have no idea what is your current audio setup but once you develop the ability to listen any setup is really good enough provided the signal path is short and simple.
I see very little point going through hundreds of simulations of simple (and well researched) PS topologies if you don't really listen to the effects each of these topologies have on the sound of an audio stage.
In fact the basis for the 'sound' of passive componets, wire, tubes, support structures, etc is well understood. The real issue with the 'objectivists' is not that a capacitor with a particular type of dielectric does not generate distortion or that group delay/skin effect in wire does not exist; their only real claim is that these effects are so low down that we can't hear them. And that is where the waters get really murky and no amount of simulation will tell you whether something is audible or not.
What makes things even worse for the scientist is that none of the classically recognised parameters like thd/imd have any meaningful correlation (outside of gross effects) to the listening experience. Years ago i noticed that the type of capacitor used in my phono pre had much stronger effect on the perceived tonality than the accuracy of the RIAA curve. And there were no Multicaps/Hovlands at the time so please don't tell me i was brainwashed by evil charlatans.
So, good luck in getting to the secret of audio nirvana, but with only theorising and simulating and no first hand experience the conclusions are very predictable.
At some level this undoubtfully is true, but there is little point going there. Most listeners seem to agree upon the effects of 'tweaks' so the experience is not so individual most of the time.
May i suggest a personal, first hand listening experience as an essential starting point for your journey. You only require a very simple audio chain to conduct meaningful tests and of course, certain listening abilities. If listening to the same track 50 times in a row proves too boring than maybe you're not too well suited for this position 🙂 I have no idea what is your current audio setup but once you develop the ability to listen any setup is really good enough provided the signal path is short and simple.
I see very little point going through hundreds of simulations of simple (and well researched) PS topologies if you don't really listen to the effects each of these topologies have on the sound of an audio stage.
In fact the basis for the 'sound' of passive componets, wire, tubes, support structures, etc is well understood. The real issue with the 'objectivists' is not that a capacitor with a particular type of dielectric does not generate distortion or that group delay/skin effect in wire does not exist; their only real claim is that these effects are so low down that we can't hear them. And that is where the waters get really murky and no amount of simulation will tell you whether something is audible or not.
What makes things even worse for the scientist is that none of the classically recognised parameters like thd/imd have any meaningful correlation (outside of gross effects) to the listening experience. Years ago i noticed that the type of capacitor used in my phono pre had much stronger effect on the perceived tonality than the accuracy of the RIAA curve. And there were no Multicaps/Hovlands at the time so please don't tell me i was brainwashed by evil charlatans.
So, good luck in getting to the secret of audio nirvana, but with only theorising and simulating and no first hand experience the conclusions are very predictable.
> .02% THD is .02%THD, isn't it? Even if it's created by two different tubes?
Disagree on several levels.
0.02%THD in tubes is almost unknown. At high level, that needs a lot of feedback. At low levels, noise will mask any 0.02%THD. Yes, an analyzer may detect 0.02%THD on a small signal under 0.2% noise, and the ear may too. But I'd rather yack about more realistic levels.
Say you have one amp with "0.5%THD" that is mostly 2nd harmonic, and another that is "0.5%THD" but mostly 5th, 7th, 9th harmonics. The 0.5% 2nd is essentially inaudible. The 0.5% 5,7,9th subtly grates.
I feel that we should multiply each harmonic by its order. 0.5% times 2 is 1%. 0.5% times 9 for the 9th harmonic is 4.5%. The latter is 4 times more annoying than the former. That's not quite how the ear hears it. But it is a lot closer to how we hear than any THD number that sums all the partials the same. This was not an issue when THD was invented, because most amps had a declining distortion spectrum and the high partials "hid" behind the large 2nd and 3rd. When we got heavy feedback amps, we could make that 2nd and 3rd vanish, reduce the 5th, but in many cases the 7th and 9th are higher than in the simple amps of yesteryear.
THD is meaningless unless it is VERY small (20dB below noise at all levels), or the distortion spectrum is specified.
Why could "same" tubes sound different? Consider the grid wires. If they are equally spaced and pitched, the curvature is mostly "large" 3/2-power law, giving significant low-order distortion and insignificant high-order distortion. But now bend one of the grid wires a little closer. It will cause a steeper slope of the characteristic but will cutoff sooner. While passing through this range, the tube's smooth curve will kink. Much less than 1% compared to the total range of the tube. It looks the same on a curve tracer. You can't measure it on a needle-meter, and if you do you call it measurement error. It does not significantly affect the usual tube-type distortion tests done at 5%THD. But it puts high-order kinks and partials in the sound. This alone could explain some differences between mature tube factories with some high-spec products (US/EUR 1950s) and jury-rig tube factories with hungry unfussy customers (most current producers).
Note that tube specs are quite loose. 12AX7 is speced for just two conditions and no promise what it will do at any other condition. The spec-sheet condition is often in the top 50% of the tube's power ability, while we often run tubes below 25% of rated power. 6DJ8 is essentially speced for one condition, at much higher bias than usually used in audio (indeed so high that life is shortened). This means that a 6ES8 is "same as" 6DJ8... AT that one bias condition. The 6ES8 has several more spec points, but if you are selling "same as 6DJ8" you don't care about that. You can "meet spec" at the one or two spec points and be way off at other points.
I do have a hard time with conductors of different metals having different sounds. I'm more willing to accept that junctions of dis-similar metals have an effect, but there are so MANY junctions in any practical system that I don't know where to start. I will note that the press-in pins on vacuum tubes are very prone to poor oxide contact and small amounts of distortion. If you are going to tube-roll, at least use a good contact enhancer so you hear tube-guts, not fixable problems on pins and sockets.
Disagree on several levels.
0.02%THD in tubes is almost unknown. At high level, that needs a lot of feedback. At low levels, noise will mask any 0.02%THD. Yes, an analyzer may detect 0.02%THD on a small signal under 0.2% noise, and the ear may too. But I'd rather yack about more realistic levels.
Say you have one amp with "0.5%THD" that is mostly 2nd harmonic, and another that is "0.5%THD" but mostly 5th, 7th, 9th harmonics. The 0.5% 2nd is essentially inaudible. The 0.5% 5,7,9th subtly grates.
I feel that we should multiply each harmonic by its order. 0.5% times 2 is 1%. 0.5% times 9 for the 9th harmonic is 4.5%. The latter is 4 times more annoying than the former. That's not quite how the ear hears it. But it is a lot closer to how we hear than any THD number that sums all the partials the same. This was not an issue when THD was invented, because most amps had a declining distortion spectrum and the high partials "hid" behind the large 2nd and 3rd. When we got heavy feedback amps, we could make that 2nd and 3rd vanish, reduce the 5th, but in many cases the 7th and 9th are higher than in the simple amps of yesteryear.
THD is meaningless unless it is VERY small (20dB below noise at all levels), or the distortion spectrum is specified.
Why could "same" tubes sound different? Consider the grid wires. If they are equally spaced and pitched, the curvature is mostly "large" 3/2-power law, giving significant low-order distortion and insignificant high-order distortion. But now bend one of the grid wires a little closer. It will cause a steeper slope of the characteristic but will cutoff sooner. While passing through this range, the tube's smooth curve will kink. Much less than 1% compared to the total range of the tube. It looks the same on a curve tracer. You can't measure it on a needle-meter, and if you do you call it measurement error. It does not significantly affect the usual tube-type distortion tests done at 5%THD. But it puts high-order kinks and partials in the sound. This alone could explain some differences between mature tube factories with some high-spec products (US/EUR 1950s) and jury-rig tube factories with hungry unfussy customers (most current producers).
Note that tube specs are quite loose. 12AX7 is speced for just two conditions and no promise what it will do at any other condition. The spec-sheet condition is often in the top 50% of the tube's power ability, while we often run tubes below 25% of rated power. 6DJ8 is essentially speced for one condition, at much higher bias than usually used in audio (indeed so high that life is shortened). This means that a 6ES8 is "same as" 6DJ8... AT that one bias condition. The 6ES8 has several more spec points, but if you are selling "same as 6DJ8" you don't care about that. You can "meet spec" at the one or two spec points and be way off at other points.
I do have a hard time with conductors of different metals having different sounds. I'm more willing to accept that junctions of dis-similar metals have an effect, but there are so MANY junctions in any practical system that I don't know where to start. I will note that the press-in pins on vacuum tubes are very prone to poor oxide contact and small amounts of distortion. If you are going to tube-roll, at least use a good contact enhancer so you hear tube-guts, not fixable problems on pins and sockets.
DrewP said:
No, they don't. Chances are, differences in Gm, mu and Rp cause different levels of Miller C and other stage impedances, especially important if driving any sort of passive filter (as I mentioned above using 12AX7 vs. AT vs. AU in my preamp, which uses a passive Baxandall network).
Tim
analog_sa said:
Ah, so it's just like religion.. you can't prove that God doesn't exist so he must.
Tim (goes to church regularly in case you were wondering)
Not sure if i understand the above logic. I'd also be going to church regularly if God was speaking to me on a daily basis but what my ears can't hear does not necessarily exist 🙂
helpful
Just a thought run8, if you can purchase some insulated steel wire, you could make ICs and sonically compare them to copper/silver ICs. Although not a tube, it might be an interesting experiment.
------------------------------
.02% isn't that hard to do in tube design, although I think you probably meant in an amp as a whole.
------------------------------
Might want to read "Darwin's Black Box" by Dr. Behe. He is a microbiologist dealing with cells and systems in the body.
I believe his contention is that in the systems of the eye/sight, immune system, clotting system etc.; that the odds of putting together just one system is some -20 powers less than what scientists consider impossible, which I believe he said was something like 10 to the -25th or -45th powers. Since there are thousands of systems in the body, the odds of the body evolving is so close to impossibly even scientists would have to stretch their own definition immensely.
And each system has to be created whole, or else the system doesn't work (so why evolve it?) or the body dies (again how does it evolve if it dies?). Some systems would cause death before the body had a chance to replicate itself.
-----------------------------------
Another point, archeologists have dated some bible manuscripts well before the prophisied events, and in some cases even giving names of the future rulers and about the future kingdoms themselves.
Just something to ponder.
Just a thought run8, if you can purchase some insulated steel wire, you could make ICs and sonically compare them to copper/silver ICs. Although not a tube, it might be an interesting experiment.
------------------------------
.02% isn't that hard to do in tube design, although I think you probably meant in an amp as a whole.
------------------------------
Might want to read "Darwin's Black Box" by Dr. Behe. He is a microbiologist dealing with cells and systems in the body.
I believe his contention is that in the systems of the eye/sight, immune system, clotting system etc.; that the odds of putting together just one system is some -20 powers less than what scientists consider impossible, which I believe he said was something like 10 to the -25th or -45th powers. Since there are thousands of systems in the body, the odds of the body evolving is so close to impossibly even scientists would have to stretch their own definition immensely.
And each system has to be created whole, or else the system doesn't work (so why evolve it?) or the body dies (again how does it evolve if it dies?). Some systems would cause death before the body had a chance to replicate itself.
-----------------------------------
Another point, archeologists have dated some bible manuscripts well before the prophisied events, and in some cases even giving names of the future rulers and about the future kingdoms themselves.
Just something to ponder.
contact potential
One effect I have noticed from testing a tube circuit idea (on the bench) was the large variation of contact potential across tubes of even the same brand. (I was using diodes, so was easy to observe this on a curve tracer as a voltage offset.) This same affect in triodes etc. would lead to effective variations in the operating bias level and would no doubt cause differences in sound, since different portions of the tube characteristics would be used.
Another factor could be the variation of oxide coatings between tubes causing different levels of emissivity. And to top it off, emissivity changes noticeably during the first operating days of most tubes until it settles down to a stable value.
Tube rolling tests really need to include a bias adjustment for the tube being swapped. Also, I rarely hear of significant quantities of the same tube type being tested to see if the effects noticed are consistant. This is not to say that there couldn't be differences, just that I have my doubts of their repeatability in another amplifier which may be biased differently.
Don
One effect I have noticed from testing a tube circuit idea (on the bench) was the large variation of contact potential across tubes of even the same brand. (I was using diodes, so was easy to observe this on a curve tracer as a voltage offset.) This same affect in triodes etc. would lead to effective variations in the operating bias level and would no doubt cause differences in sound, since different portions of the tube characteristics would be used.
Another factor could be the variation of oxide coatings between tubes causing different levels of emissivity. And to top it off, emissivity changes noticeably during the first operating days of most tubes until it settles down to a stable value.
Tube rolling tests really need to include a bias adjustment for the tube being swapped. Also, I rarely hear of significant quantities of the same tube type being tested to see if the effects noticed are consistant. This is not to say that there couldn't be differences, just that I have my doubts of their repeatability in another amplifier which may be biased differently.
Don
- Status
- Not open for further replies.