View Full Version : Valve Distortion
I have been tube rolling for a while now in the Croft 25R+ and have noticed that some tubes distort quite significantly. I now only use NOS Tesla ECC83 32's in my preamp and they are crystal clear with no distortion at all. Some of my previous tubes, Mullard, JJ Tesla and GE have all distorted the sound to a degree.
Anyone else found this happened with their valves / equipment?:scratch:
PS. The distortion is most significant on vocals.
Stratmangler
25-05-2015, 10:00
Could be a biasing issue https://www.watfordvalves.com/pdfs/guitaristv02-01_biasing.pdf
It could have been due to some valves being thinner glass, or touching the amp casing, and more susceptible to vibrations - damping rings could possibly fix.
But I suspect it's probably a biasing issue.
Damping rings helped my valves, Svetlanas and Mullards. Somewhat subtle but nevertheless worth doing.
Something's wrong there, Jim, as that's not normal. There should be no audible distortion with any valves. Therefore, there is either a set-up/connection issue, or the valves you mentioned were faulty.
Marco.
Firebottle
25-05-2015, 10:23
James if you recall I've touched on this when I've been round at yours talking valve kit.
The different characteristics of the valves will make them operate at a different point on the transfer curve. The published transfer curve will be for an 'average' valve, so it is almost inevitable that different valves will operate at perhaps sub-optimal conditions, unless load and biasing resistances/impedances are altered for optimum loading.
The good news is that my newly developed circuit (for phono or line stages) operates the valve in a 'super linear' fashion, as I've termed it. This results in absolutely no change to RIAA eq with valve substitution or valve ageing :)
For a little feedback on how the line stage performs see this post: http://theartofsound.net/forum/showthread.php?38340-New-project-appearance&p=650327#post650327
:cool: Alan
Marco, just seen your post, I think James is referring to subtleties in the sound, but distortion non the less.
Yup. It's interesting, though, Alan that in all the years I've done tube-rolling, I've NEVER encountered that problem, other than if the valves themselves were duffers (and very obviously so)...
I wonder why that is - could I have just been extremely lucky with all the 100s of tubes I've rolled, and the many different bits of equipment I've used them with? :scratch:
Or maybe all the valve kit I've had has been self-biasing? If so, that includes my current Croft preamp.
Marco.
Yes, but if the circuit is designed for a ecc83 (for example) any ecc83 (type) should work fine in it, talking about "optimising" is just smoke and mirrors (IMHO). Just a few little items
1. All valves distort, as do all active devices, no active device has a linear transfer function, just some closer than others.
2. We can measure this distrortion, and individual valves will distort more or less than others, especially of its a two or more stage device as the distortion in each stage will either add or subtract
3. Once you get beyond broken or incorrect valves its hard to measure a difference between two (same type) valves of differing makes, they will all have measureable differences, but so will two valves of the same make.
4. Given 1, what we hear and like about one valve over the other is generally how the distortion matches our ears and that of other parts of the system (and the air remember).
5. BUT as well as 4, the physical construction of a valve will matter. The insides of a valve are bits of metal held apart in a vacume. External vibrations will affect that geometry, so they will be microphonic to a greator or lessor extent. The signals going through a valve will also exert forces on the parts of the valve, so those forces will mean the valve is changed by its own signal will also have an effect on how it sounds. The latter is harder to measure and is (I believe) responsible for a lot of what we regard as the "sound" of a mullard compaired to a jj ecc83 (for example).
And yes, you can wrap a valve with other parts that will mean that the part to part differences between sample has no (or close to no) effect on the operation of the valve, thats what good design does, look at the designs from the likes of HP and Tek in the 1960's for gerat examples how to do that. A good design will assume the expected vaiarion and work with it rather than hoping it doesnt exist.
Yes, but if the circuit is designed for a ecc83 (for example) any ecc83 (type) should work fine in it...
Yup, that was my point, and is exactly what I've experienced. I've 'rolled' hundreds of different valves over the years, of all different types and in all different circuits, and as long as the circuit of the equipment I'm using has been designed for use with a particular valve, it matters not a jot which make it is, as long as it's the same type of valve, in an electrical sense.
Therefore, I've not had to 'adjust' anything in order to remove audible distortion (that is, blatant distortion when something is obviously wrong, rather than the euphonic type that exists, as you've outlined, in all valves, by their very nature) :)
Marco.
Yes, but if the circuit is designed for a ecc83 (for example) any ecc83 (type) should work fine in it, talking about "optimising" is just smoke and mirrors (IMHO). Just a few little items
1. All valves distort, as do all active devices, no active device has a linear transfer function, just some closer than others.
2. We can measure this distrortion, and individual valves will distort more or less than others, especially of its a two or more stage device as the distortion in each stage will either add or subtract
3. Once you get beyond broken or incorrect valves its hard to measure a difference between two (same type) valves of differing makes, they will all have measureable differences, but so will two valves of the same make.
4. Given 1, what we hear and like about one valve over the other is generally how the distortion matches our ears and that of other parts of the system (and the air remember).
5. BUT as well as 4, the physical construction of a valve will matter. The insides of a valve are bits of metal held apart in a vacume. External vibrations will affect that geometry, so they will be microphonic to a greator or lessor extent. The signals going through a valve will also exert forces on the parts of the valve, so those forces will mean the valve is changed by its own signal will also have an effect on how it sounds. The latter is harder to measure and is (I believe) responsible for a lot of what we regard as the "sound" of a mullard compaired to a jj ecc83 (for example).
And yes, you can wrap a valve with other parts that will mean that the part to part differences between sample has no (or close to no) effect on the operation of the valve, thats what good design does, look at the designs from the likes of HP and Tek in the 1960's for gerat examples how to do that. A good design will assume the expected vaiarion and work with it rather than hoping it doesnt exist.
Excellent post Nick.
I have found that using a test mic and doing simple system pink noise FR checks, it is easy to pick up the effects of valve rolling on FR.
By running controls of "no change" just to analyse test results variance, I find virtually none. Then roll some valves and hey presto, 2-3DB shows up. Recent Lampizator triode tests indicate the VT-52 as having the best frequency response (i.e. the least variance over the whole FR range).
The Tesla E88CC also fares very well over some cheap Russian rocket logo jobbies.
I just use my lugs! ;)
I can always hear differences between different makes of tubes of the same type... Therefore, simply select the sound that I like best. Job done - no measuring needed :)
Marco.
I just use my lugs! ;)
I can always hear differences between different makes of tubes of the same type... Therefore, simply select the sound that I like best. Job done - no measuring needed :)
Marco.
So do I. What is really interesting is that the 6A3, for instance, is a bit of a bad boy frequency response wise. It sounds quite a bit more lively and spatial than the VT-52.
I'm just trying to align what I hear with some test results. If you go back and read what I said about all the triodes I tried in the Lampi Big 7 from subjective tests (done months before I got the mic out), and read what I say about the VT-52, it certainly aligns well with the test results.
The point is Marco, the test results aren't dreamland cuckoo bullshit. You can blatantly hear what they show.
So to a very large extent no mic is needed.;)
Yup, but then put the same valve in a different amplifier or circuit (one that still uses the same valves, but has a different musical presentation) and subsequently how you thought you'd 'nailed' the sound of that valve changes yet again... ;)
Therefore, I'm not a big believer in that certain makes of valves have an intrinsic sound, rather their sound 'aligns' with that of the circuit they're used in - and so you hear what you hear in any given application/set of circumstances :)
Marco.
And so in different circuits the measurebations will almost certainly be different. I'm not interested in circuits I don't own, however.
I'd still expect some consistency in results across circuits, though. A shit valve will never measure as well as a good one. Also, pink noise FR tests are just one type of measurement. Impulse tests at various frequencies might be interesting. That said pink noise covers the entire FR.
One more point - I don't listen to VT-52 all the time. I get bored with it and change to other triodes. That's a major reason for buying the DAC. That, and the fact it is, to me, better sounding than anything else I have ever tried by a margin.
Self-justification in front of others. I dunno. I must be getting weak...:lol:
I would have thought it very unlikely that any valve (itself) is likely to have anything but a flat frequency response in the audio to low RF range DC - 100k, what will vary is how the valve's anode resistance, grid capacitance and transconductance interacts with the particular circuit around it. I would especially expect varying results if the valves used are not all the same heater voltage where some will be more or less underheated.
Serge thought pretty much the same Nick.
Get the test mic out and try it for yourself. It is definitely there. I spent ages on the Wam proving it to Serge a few years back.
I would have thought it very unlikely that any valve (itself) is likely to have anything but a flat frequency response in the audio to low RF range DC - 100k, what will vary is how the valve's anode resistance, grid capacitance and transconductance interacts with the particular circuit around it. I would especially expect varying results if the valves used are not all the same heater voltage where some will be more or less underheated.
Er hold on what are you saying here - that you may measure FR differences due to the various electrical parameters changing?
I would have thought it very unlikely that any valve (itself) is likely to have anything but a flat frequency response in the audio to low RF range DC - 100k, what will vary is how the valve's anode resistance, grid capacitance and transconductance interacts with the particular circuit around it.
The sums up, technically, what my own experience and ears tell me :)
Justin, I agree with what you're saying. The only point I was making was that you have to avoid the tendency of telling others how you think certain valves 'sound' (say, when asked on forums), as what you've heard/measured in one circuit, will not necessarily apply in another.
In that respect, someone could have an opposing view in relation to the Tesla and Russian valves you cited. Therefore, at the end of the day, they would be as 'right' about the matter as you are. In essence, of course, there is no 'right'. Simply choose your favourite 'flavour' of coloration, and live with it, as everything about valves is circuit and system dependent ;)
Marco.
Firebottle
25-05-2015, 14:13
The signals going through a valve will also exert forces on the parts of the valve, so those forces will mean the valve is changed by its own signal will also have an effect on how it sounds. The latter is harder to measure and is (I believe) responsible for a lot of what we regard as the "sound" of a mullard compaired to a jj ecc83 (for example).
Interesting thoughts Nick, you could well be right. There has to be something in the mix that affects the sound, particularly comparing valves that are virtually identical internally.
:) Alan
Justin, I agree with what you're saying. The only point I was making was that you have to avoid the tendency of telling others how you think certain valves 'sound' (say, when asked on forums), as what you've heard/measured in one circuit, will not necessarily apply in another.
You might be confusing me with someone else there as I usually always say YMMV depending on your circuit/system or words to that effect.
I'm pleading really quite not guilty on that one:) No offense taken BTW.
Lol - no worries. However, I didn't mean it like that. I know you always say 'YMMV'.
Clearly, you have formed some fixed (although undoubtedly valid) opinions on the valves you've used/measured in the context of your system. However, those results won't necessarily apply elsewhere.
Essentially, this is the point I'm debating with you:
I'd still expect some consistency in results across circuits, though. A shit valve will never measure as well as a good one.
Excluding outright faulty, define what you consider as "shit'?
Is that "shit", as is the case in your system, or "shit" as might not be the case in someone else's, both for valid reasons (due to the same valves being used in different circuits)? Do you see the point I'm making? :)
Outside of something that's blatantly faulty or incompatible, I don't believe that any specific valve is universally "shit".
Marco.
Er hold on what are you saying here - that you may measure FR differences due to the various electrical parameters changing?
No what I am saying that just about any valve will have a flat frequency response from DC to 100kHz, probably a lot higher. However the circuit around the valve will affect gain and frequency response. If you add heaters not being driven to spec then all sorts of differences will occur. If (as I think you are) you replace a 2a3 with a 45, then depending on the heater circuit alone all sorts of differences will be apparent. But if you drive, heat and load a 2a3 and a 45 correctly and stay within both valves linear range then you will start to find it harder to tell them apart.
BTW, why are you using a test mic, why not measure the output of the DAC directly? (unless its a intentional experiment to see the effect of microphonics).
Lol - no worries. However, I didn't mean it like that. I know you always say 'YMMV'.
Clearly, you have formed some fixed (although undoubtedly valid) opinions on the valves you've used/measured in the context of your system. However, those results won't necessarily apply elsewhere.
Essentially, this is the point I'm debating with you:
Excluding outright faulty, define what you consider as "shit'?
Is that "shit", as is the case in your system, or "shit" as might not be the case in someone else's, both for valid reasons (due to the same valves being used in different circuits)? Do you see the point I'm making? :)
Outside of something that's blatantly faulty or incompatible, I don't believe that any specific valve is universally "shit".
Marco.
Well absolutely within the context of my own system - in the Lampizator threads where I have spoken about valve rolling more than anywhere there are multiple extended clauses to that effect.
An example of a "shit" valve? Well I believe I documented one in a Lampizator thread - it was a Valvo "pinched waist" type that was incredibly microphonic. Now that can sound quite entertaining in the context that it behaves a bit like a cartridge pickup in a poorly isolated TT. No way can it be regarded as technically competent as the construction mandates otherwise. It probably measures fine on a tube tester, however, for obvious reasons, not when subject to resonance.
Note I am not knocking Valvo generally as I have had some excellent Valvo valves. But those? One has to question why the pinched waist was made, as it appears NOT to support the internal structure adequately.
Nick - the Lampi has a two position switch - high and low heater. One dedicated for 2A3 and one for 45 (labelled) but the high switch is also intended to accommodate 6A3, 300B, VT-52 etc and the low RCA 245, Cunningham CX345 etc. In fact, a lot of different triodes will work in this "genius" LOL circuit that I believe adjusts itself somehow to some extent depending on what it sees.
I use a mic to determine the overall system effect at the listening position as I am only interested in correlating what I hear with what the mic "hears".
But in the thread where I endeavoured to prove simple tube rolling of the same type effects FR, I used various 211 type triodes in the same fixed bias circuit. I can probably find it with a lot of Googling. I could also prove it to you in the flesh but it is a bit of a drive:)
Note small triodes in the same preamp circuit also effect FR according to my measurements. Hence the E88CC comment above.
An example of a "shit" valve? Well I believe I documented one in a Lampizator thread - it was a Valvo "pinched waist" type that was incredibly microphonic.
Note I am not knocking Valvo generally as I have had some excellent Valvo valves. But those? One has to question why the pinched waist was made, as it appears NOT to support the internal structure adequately.
Okies, but are they all like that - perhaps you just got a bad batch? What if someone else used a "pinched waist" Valvo and claimed it as the best tube he'd ever heard (in the context of his system) - would it still be categorised as "shit"? ;)
Btw, have you tried those rubber 'cock rings', you can slip over the glass structure of valves? They can reduce microphony quite considerably! :)
Marco.
Nick - the Lampi has a two position switch - high and low heater. One dedicated for 2A3 and one for 45 (labelled) but the high switch is also intended to accommodate 6A3, 300B, VT-52 etc and the low RCA 245, Cunningham CX345 etc. In fact, a lot of different triodes will work in this "genius" LOL circuit that I believe adjusts itself somehow to some extent depending on what it sees.
Well, without seeing what was in the "genius" circuit, I am not sure how it could provide 5v @ 1.2A, 2.5v @ 2.5A and 7v 1.18A, I guess it could try and regulate to a fixed power but at 6W, 6.25W and 8.26W it will not be able to heat all of them to spec, so I would expect different results with different valves. The 45 at 1.5A. I can see that maybe 300b, vt52 and 45 could be heated with a current source set at 1.2A, but again, it would be a crap shoot what actual voltage each heater saw and the heater supply can make a huge difference to a DHT.
But in the thread where I endeavoured to prove simple tube rolling of the same type effects FR, I used various 211 type triodes in the same fixed bias circuit. I can probably find it with a lot of Googling. I could also prove it to you in the flesh but it is a bit of a drive
Again, I can imagine that, all it takes is the ra of the valves to differ, combined with the leakage inductance of the output transformer and the HF response could vary within the audio range. But load the same valve with a resistive or tuned load and you would find the HF is flat for some while. Remember that the 211 was rated as a RF power amplifier up to 80Mhz, so the valve itself will be not even working hard at 100kHz
Marco, we could rant on forever... the Valvo in question was noted to be microphonic elsewhere. Even the ebay seller said they were microphonic. I was just curious and wanted to try them:) They were "tap the chassis and hear a big noise through the speakers" microphonic. Very much like resting the stylus on a groove with no platter speed and tapping the chassis, maybe a bit worse TBH (depends on yer deck, though). That effect really can make a DAC sound much more like a TT. You might like them. Really.
I've got some "cock rings" LOL round the small signal valves in my preamp i.e. the Tesla E88CC crossed swords.
Well, without seeing what was in the "genius" circuit, I am not sure how it could provide 5v @ 1.2A, 2.5v @ 2.5A and 7v 1.18A, I guess it could try and regulate to a fixed power but at 6W, 6.25W and 8.26W it will not be able to heat all of them to spec, so I would expect different results with different valves. The 45 at 1.5A. I can see that maybe 300b, vt52 and 45 could be heated with a current source set at 1.2A, but again, it would be a crap shoot what actual voltage each heater saw and the heater supply can make a huge difference to a DHT.
On the Lampi thread some guy was basically saying it gets closer with some valves than others, which I suspected anyway. He was in there, with test kit, curious, checking it out. Maximum respect.
It doesn't matter if the valves aren't run to spec in some cases. What does matter is you can "f&*^ around" with the sound of the DAC to your liking - be that technically poor or absolutely bang on spec wise.
Marco, we could rant on forever... the Valvo in question was noted to be microphonic elsewhere. Even the ebay seller said they were microphonic. I was just curious and wanted to try them:) They were "tap the chassis and hear a big noise through the speakers" microphonic. Very much like resting the stylus on a groove with no platter speed and tapping the chassis, maybe a bit worse TBH (depends on yer deck, though). That effect really can make a DAC sound much more like a TT. You might like them. Really.
I've got some "cock rings" LOL round the small signal valves in my preamp i.e. the Tesla E88CC crossed swords.
Indeed - I trust you can see my point, though. Making generalisations about anything can be dangerous! ;)
Yeah, those CRs can be handy sometimes, and not just on small signal valves.....
Marco.
Thanks guys for all the info but as I am not a tecky some of this goes over my head.
So trying to work out what is going on in my particular case I presume i must have been using a duff valve which could have gone seriously microphonic or was just not suited to the particular circuit it was being used in?:scratch:
I can switch the valves in question and the degree of distortion can be heard so it was not a minor issue with SQ.
Next time your over Alan (Firebottle) I will demonstrate in the Croft and you can hear for yourself.:)
Firebottle
25-05-2015, 20:30
Look forward to it James. I'll bring my meter so that I can get some measurements of the operating conditions to see if that will give any clues to the differences.
:) Alan
audioglow
26-05-2015, 07:56
Hi James
Yes I have experienced this issue in the past - having tube rolled for some 50 years now - in both power and pre amps. Although pre amps are more prone to "bad" tubes of course - and female vocals are the hardest to reproduce. Many issues in the past have variable "faults" including poorly seated tubes and dirty pins (its amazing how little oxidation can cause issues) - but the most likely issue in your case is the tube themselves - many Mullards out there are fakes (even though the box and printing look good) and have even seen 1000s in Hong Kong recently being made! and ebay is a haven for tube pulls and nearly dead tubes - many from guitar amps where they are driven hard...As an owner of 10K tubes from various sources I always test before using or selling (part of my retirement fund)! and I fine approx 50% are not "new" even though they look new. Designers use certain tubes in there design criteria and changing that part will many times change the sound (hence tube rolling) as most tubes will read differently by nature of their design and structure and manufacture so one tube may work differently in different amps.GE tubes particulary the green label types were produced in there millions and had very variable quality control and these particularly are prone to distortion. Finally (and this is a guess as I dont have a schematic for the croft) many designers "pair" both sides of a dual triode and as such a "matched" tube on each side will work better, and teslas tend to be very consistant on both sides of the dual triode and maybe the others are not.
Hi James
Yes I have experienced this issue in the past - having tube rolled for some 50 years now - in both power and pre amps. Although pre amps are more prone to "bad" tubes of course - and female vocals are the hardest to reproduce. Many issues in the past have variable "faults" including poorly seated tubes and dirty pins (its amazing how little oxidation can cause issues) - but the most likely issue in your case is the tube themselves - many Mullards out there are fakes (even though the box and printing look good) and have even seen 1000s in Hong Kong recently being made! and ebay is a haven for tube pulls and nearly dead tubes - many from guitar amps where they are driven hard...As an owner of 10K tubes from various sources I always test before using or selling (part of my retirement fund)! and I fine approx 50% are not "new" even though they look new. Designers use certain tubes in there design criteria and changing that part will many times change the sound (hence tube rolling) as most tubes will read differently by nature of their design and structure and manufacture so one tube may work differently in different amps.GE tubes particulary the green label types were produced in there millions and had very variable quality control and these particularly are prone to distortion. Finally (and this is a guess as I dont have a schematic for the croft) many designers "pair" both sides of a dual triode and as such a "matched" tube on each side will work better, and teslas tend to be very consistant on both sides of the dual triode and maybe the others are not.
Hi Ian,
thanks for your insights, you have confirmed what I have suspected regarding NOS valves and tubes in general. Much of the stuff circulating is of suspect quality,age and use! I only buy from genuine suppliers who know the use and origin of what they are selling.
Indeed one guy I buy from lets me know the tubes are used NOS but have been tested and evaluated before I get them.
How much stuff out there is not only not genuine but of a suspect age and use! I think there is quite a lot especially when there is so much money to be made from NOS tubes.
I am also learning that not all valves even the same type ie ECC83 perform exactly the same in a given circuit, with some performing/ operating better than others which is not a surprise considering the difference in manufacturing materials and processes. Reminds me when my other half and myself both bought a Ford Focus each - both manufactured in the same year but one built in Germany and the other I think Spain. They were so different could have been different models of cars!!
I have at last found valves which work extremely well in my Croft pre and some of the information in previous posts including your own experience would suggest that the distortion I experienced was certainly caused by the valves and not the equipment they were used in otherwise I would still be experiencing distortion.
However I would be interested to know if the distortion was caused by a valve that was knackered/microphonic or wether it was an incompatability with the circuit is was used in?
When I mention distortion I don't mean the crackly fuzzy distortion that can be experienced with electronics but the way in which the valve distorted the vocals or instruments which sightly blurred their SQ. Switching to the NOS Tesla's bought about a crystal clear clarity to everything almost like a camera focusing on a subject.
Firebottle
26-05-2015, 12:23
James is there one valve position that seems worst affected or is it all valve positions?
Hi Alan,
The biggest difference was noticeable in the phono stage but significant improvement were also made when the valves were changed in the regulation stage. The Linestage did not really make much difference.
Powered by vBulletin® Version 4.2.3 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.