Acceptable variances for "matched" power tubes?

[Steve B]

I received an order of tubes that were supposed to be matched, yet they were slightly different.

There is a 6 milliamp difference on a pair of JJ 6V6's & a 4 milliamp difference on a pair of Winged C 6L6's.

A previous pair of JJ 6V6's I bought from another dealer were within a tenth of a milliamp to each other.

Am I being too picky?

Thanks,
Steve

[Tom Phillips]

IMHO your experience is not unusual. I have received "matched" tube sets from some vendors that I can only imagine were matched with a dart board. Others were always withing a couple of ma. I would consider anything within a couple of milliamps to be a very close match. Many peolpe will say that within 5 ma or even more is acceptable. I reality we should be talking about % rather than an absolute current. A couple of things to consider are: 1) The readings will probably drift as the tubes age. They may get closer together or drift apart. 2) A set of tubes may be closely matched under one set of operating conditions (Plate voltage and grid voltage) and less well matched under another set of operating conditions.

If the tubes were mine I'd bias them up to the average setting and listen to how they sound. They will probably be fine. If they really bug you I'd contact the vendor. If they make it right...cool. If they blow you off, then I wouldn't use them again.

Regards,
Tom

[Regis]

No, you are not being too picky. Hell yes I've seen this before, with all kinds of different brands of power tubes. I've talked about this before; a matched set of tubes with 4-5 ma difference is not a matched set of tubes, I don't give a damn what you say. People talk all the time about microphonic 12ax7's, but I've never seen one. I experienced my first microphonic preamp tube recently when I refurbished my Kay 720 and went through 3 5879's before I got one that worked.

However, virtually every set of matched power tubes I've seen in the last 5 or 6 years have had that 4-5 ma difference, and as far as I'm concerned that's not matched. This isn't to say that they won't sound good, but don't tell me they're matched when they aren't.

[Ray Ivers]

Regis,

I realize you probably won't give a damn what I say :D but IMO it's quite possible to have a truly matched tube set in a musically-important way (not merely in idling-heat-output) that nonetheless shows significant variances in DC current flow at idle between tubes. Unless you're using a toroidal OT - in which case a bias-servo circuit is in order IMO - 5mA or less DC mismatch won't matter at all IME. In fact, I've seen sets with much greater idle mismatches that put out very similar full-power currents, and sounded great; I'd trade DC current-matching at full power for DC current-matching at idle any day of the week, but I don't know of any tube seller that does this (and what is "full power"? or "idle", for that matter?). Idle DC-matching is really quick and easy to do, and it gives a number that can be readily checked with a meter - but usually it doesn't last, and unless you've got a serious mismatch (which IMO is important mainly for indicating large differences in baseline tube transconductances, which can make a real difference) it really isn't a matter of life and death tone-wise; my $.02, YMMV.

AC matching - now that's a different story. Getting a set of tubes that makes music in an identical fashion is very much a worthwhile goal IMO, but hardly anyone does it because a) it's difficult to do without a curve tracer, which involves money, possibly a computer w/attendant skill set, etc. b) you really need a huge quantity of 'raw' tubes to end up with a worthwhile number of AC-matched sets, and c) it doesn't always result in a closely-DC-idle-matched set, which can greatly increase the WTF!-factor on resale, especially when top dollar was paid. AFAIK the only big tube producer who AC-matches is Groove Tubes.

If you feel more comfortable going with tightly-DC-matched tube sets, that's great - it surely can't hurt, and at the very least it shows that some TLC was put into the selection of your tubes which IMO shows integrity on the part of the seller. I realize you posted that unmatched sets might still sound good, but I get the feeling that you'd much prefer a DC-matched set, and I'd hate to see you pass up some possibly great-sounded tubes (or having a bad feeling about a tube vendor) because they weren't within a certain arbitrary number of DC mA of each other at one particular Vp/-Vg. And as idle current (along with practically every other aspect of tube DC & audio performance) is dependant on interelectrode spacings - which are really pretty small, and can change during shipping - it's quite possible a perfectly-DC-matched set will arrive at your door unmatched.

I haven't really had too many "crusher" 12AX7's (ones that were so microphonic they made thunderous booming/rattling noises, or rang like a gong, and IMO had to be destroyed) but man have I encountered tons more that were borderline. IME high-gain combos are the worst for showing up loose 12AX7's (or beating good ones into pulp), followed by high-gain amps sitting on top of high-efficiency multi-speaker cabs. And of course all the other microphonic tubes I've seen/replaced over the years - 9-pins, octals, power tubes, rectifiers (!) - they're out there, believe me. :)

Ray

[Amp Kat]

[QUOTE=Ray Ivers]Regis,


IME high-gain combos are the worst for showing up loose 12AX7's (or beating good ones into pulp), followed by high-gain amps sitting on top of high-efficiency multi-speaker cabs. And of course all the other microphonic tubes I've seen/replaced over the years - 9-pins, octals, power tubes, rectifiers (!) - they're out there, believe me. :)

I 'm with Ray on this one and some 12AX7's that are microphonic in high gain amps may be perfectly quiet in another amp which brings me to another point. It's not always the tube that causes micrphonics. I've seen small ceramic disc capacitors used that are microphonic and even lead dress that can cause oscillation leading you to believe it's microphonic and other issues as well like pots. For the power tube mismatch, are you sure that your OT is evenly matched ? Peaveys are notorious for lobsided trannys so 5 to 10 ma's apart is the difference from side to side impedance wise. I really don't let it bother me if they are a few milli's apart unless they hum then it's a problem. Some amps with the bias balance can make up for the difference bringing them closer together. Then of course theres also individual bias pots for setting them like you want it but a rare thing in current production amps and they are getting more and more cheaply made by the minute. Theres also intentional tube mismatching for getting more asymetrical balance for producing a more even order harmonic content to the mix so for playing clean IMO you do want them somewhat close but for distortion depending on how you get it there could be some disadvantage to exactly matched tubes.

[Regis]

Dammit Ray and KB, I was up here on my soap box spouting truth to the masses, and you had to go and make sense and stuff. :D All of sudden my soap box is damn rickety and look like a fool, deservedly so.

Ray, I never heard a more clear explanation about the why's and wherefore's of matching. That's not hard, I always considered it a number but not what the numbers meant. Now I do.

KB, I also hadn't considered OT mismatching, how do I measure that? Come to think of it, one of my SF Bassmans has always been 5-8ma off when I measured it, maybe that's the problem.

I'm gonna save this thread and post it on my website, it's been very educational.

[Tom Phillips]

All,
I will suggest that anyone who sees a mismatch may proceed by swapping the push & pull side tubes. If the mismatch does not follow the tubes then you will know there is something else going on and you can investigate further. Could be the PT, the bias circuit, leaky PI coupling caps etc.
Regards,
Tom

[Ray Ivers]

Regis,

FWIW, you came across as anything but foolish, and you certainly have every right to set your own tube-matching tolerances! I just wanted to take a look at the DC-matching thing from a different perspective... I wish it were more indicative of a truly well-matched set then it really is, but it's by no means a worthless waste of time, either. Lots of good info in the other posts here, too - for some reason I always forget about OT primary DCR imbalance; wishful thinking, perhaps. :)

I've got two GE 6550A pulls here, idling 17mA apart (50/33mA), and yet they pull 198/197mA with signal applied; go figure. ;)

Ray

[Tom Phillips]

Hi Ray,
Re: I've got two GE 6550A pulls here, idling 17mA apart (50/33mA), and yet they pull 198/197mA with signal applied; go figure.

Maybe the tube with the lower idle current has a higher Transconductance (gm). Seems like that could account for it "catching up" to the other tube when signal is applied. In addition the drive from the phase inverter could be a little higher on one side. That would also affect the total measured current when signal is applied. Seems to me that the combination of these two factors could easily account for the measurements that you observed.

What do ya think?
Tom

[Ray Ivers]

Tom,

Yes, I bet that the 33mA tube does have a higher Gm, which would give a lower idle current for the same -Vg. You'd think it would have lower current across the board, though, if the tubes were otherwise identical and were receiving identical drive... the tubes probably have differences in their Gm curves, and/or it could well be phase inverter imbalance as you mentioned - I'll check that out tomorrow (my last set of tests, then it gets buttoned up and returned to dustcatcher-status ;>).

Ray

[Roberto Lasco]

Ray, could I infer from your comment that matched sets that idle at lower currents (at same Vp/Vg) perform and would be considered "better" (due to higher transcondutance) than sets that idle at higher currents ? That would be interesting if true...

[R.G.]

I have a mildly heretical view on matching.

Matching is what you do when you can't design the circuit to be tolerant of mismatching. Since we can't redesign amps (very much, at least) we're stuck with matching.

What is a matched tube? In order:
1. matched tubes DC-bias at the same currents or close so we don't have magnetic offsets in the OT causing us one-sided saturation grief or out-of-current grief on one tube.
2. If tubes are matched at DC, that does not mean they're matched for AC gain. So second order matching involves gain matching as well as DC bias matching.
3. If tubes are both DC matched and AC matched, they might also be distortion matched, so that the little irregularities in AC response are also matched. I'm not sure if anyone has ever done this or not. 1 and 2 are easy.

You can do 1 and 2 by brute force measurement and selection. Or you can do it actively.

By actively I mean matching DC bias by matching the tube current with individual bias voltage settings. That largely eliminates the need for DC matched tubes.

You can also AC-gain match tubes by putting in a single sided post PI master volume of limited range. By turning down the available signal to the hotter gain tube, you can match the overal PI-to-output-tube-plate gain.

So for two added pots to tune in a new set of tubes, you can ditch buying matched sets.

I don't have a good solution to the matched-distortion squiggles.

Yet.

[Ray Ivers]

Roberto,

Ray, could I infer from your comment that matched sets that idle at lower currents (at same Vp/Vg) perform and would be considered "better" (due to higher transcondutance) than sets that idle at higher currents ? That would be interesting if true...

I think I'd have to say "yes". ;) Given two otherwise-identical tubes, the one that idled at a lower current at some fixed negative bias voltage (i.e., the one with higher Gm) would actually be considered a "better" tube by the OEM, a tube tester, etc. In an audio-amp application, the higher-Gm tube would be easier to drive to full power, and would have more apparent gain.

Ray

[Amp Kat]

Regis. A simple resistance test across the primaries from center tap to each side will tell if each side is different. Find a 5150 OT and you'll most likely encounter it. It is an interesting point that Ray made because at quesient theres no signal and who knows what happens after it starts conducting and if frequency effects in the signal have any swaying one way or another. Tube testers really don't do justice because they will give you a gm reading and tell if there is a C.T.G short or heater short but they don't tell you how long you have left on the tubes or what happens in the real power transfer stage. I guess if you really wanted to scope it in depth you could hook up a Spectrum Analyzer and find some interesting things as that's most likely how they match them in the first place. Another thing is some tubes act much different when they've warmed up for a couple of hours and we know they aren't as stout as the NOS. This guy brought in a Siver Face Bassman the other day and said he hadn't ever changed the tubes (ten years!) so I yanked them out and of course an RCA GTA black plate and one gray plate and the bias on both sides is dead on ! Go figure,

[Enzo]

There is also the matter of expectations. When we arbitrarily decide that such and such amount of whatever parameter we are considering is the match standard, we should really consider also what the matcher considers matched. They may consider getting something within say 5% of another something is matched. We can then decide on our own that it has to be 1%, but that doesn't make 5% wrong.

If I want to match 100k resistors, how close do they have to be to be called matched? 2%? 1%? .05%?

[Regis]

There is also the matter of expectations. When we arbitrarily decide that such and such amount of whatever parameter we are considering is the match standard, we should really consider also what the matcher considers matched. They may consider getting something within say 5% of another something is matched. We can then decide on our own that it has to be 1%, but that doesn't make 5% wrong.

If I want to match 100k resistors, how close do they have to be to be called matched? 2%? 1%? .05%?
Ok, my head just exploded.

If an amp overdrives in the woods, and there's no one around to hear it, does it make a sound?

I guess it depends on whether it has a master volume or a power soak. :D

[Enzo]

If a tree falls in the woods, and no one is around to hear it, how do they know the tree fell down?

[steve]

If a tree falls in the woods, and no one is around to hear it, how do they know the tree fell down?

More importantly: if a man says something in the woods and there is no woman around to hear him, is he still wrong?

steve

[Enzo]

Count on it.

[Steve B]

Well, I've been alone in the house & playing on my SF Princeton Reverb for the past couple of hours at good volume (LOUD to other family members), & it sounds really good!

So 6ma differences on the power tubes doesn't make things sound any worse. In fact, I've been really enjoying the sound, maybe even more than the matched set that was in there previously?!?! Psychoacoustics or the real deal, I like it.

Steve

[Roberto Lasco]

Matching is fine, but I need to say the sweetest DR I ever heard had the 6V6s 18mA apart ! :eek:

[R.G.]

So 6ma differences on the power tubes doesn't make things sound any worse. In fact, I've been really enjoying the sound, maybe even more than the matched set that was in there previously?!?! Psychoacoustics or the real deal, I like it.
Steve
Oh, it's the real deal. Think about it - one tube with markedly different gain than the other? By definition, that's asymmetrical gain on the two half cycles, and a slight bit of even-order distortion.

Human ears LIKE small amounts of even order distortion.

This is another of the problems with a blind urge to have matched tubes - it may not be a good thing. This issue was pointed out in the Tube Amp FAQ years ago.

[stokes]

Copnsidering that all the components in a tube amp are +/- any where from 5 to 20% of their stated values or ratings or whatever I dont think tubes being a couple of ma apart means much.I have never measured the dc res from ct to either end of the winding of an OT and saw a match.Once everything in an amp gets cooking these small differences dont mean much.In a hi-fi situation you would want tighter tolerances and the hi-fi guys pay a lot more for the components just for this reason.Just look at the prices of Black Gate caps-$100+ for an electrolytic?Or some of the paper in oil signal caps.Wont ever see a guitar amp builder using high end parts like that cause we dont need them.

[Amp Kat]

I have never measured the dc res from ct to either end of the winding of an OT and saw a match.

Hmm I have and an exact to the very tenth of an ohm. The Jcm 900 OT's are very close and several other Fenders are too.

"Once everything in an amp gets cooking these small differences dont mean much.In a hi-fi situation you would want tighter tolerances and the hi-fi guys pay a lot more for the components just for this reason."

That's true but they put more emphasis on tube matching than expensive caps. The Marantz 8B has 4 individual bias pots for each tube with an analog meter readout and switch selector for each tube. The THD % + N is a much greater concern in a HI-Fi amp at least back in the day because the music they were playing was distortion free. Nowadays with all the insane gain and noise related amps it may not be an advantage to have such a low THD % allthough that's the way it's supposed to be. Hell a mismatched set and some hum may fit in quite well with the musical content !

[Carl / Zwengel Amps]

I'm probably going to get crucified for saying it but in my eyes the whole argument is pretty much academic. Think about it for half a second! We're overdriving the hell out of the phase inverter, pushing the power tubes to the point of grid blocking and saturating the cores of the opt. So we're going to get our panties in a bunch over less than 1% thd from a set of tubes that isn't perfectly matched?! UGH!!!!

Having gotten that rant off my chest, I dont know where some people are getting their tubes but mine come matched to within less than 5% on current AND transconductance. And that'll get you some pretty damn symetric output up to the point of driver or power tube clipping...at which point as I've said previously, you're really not going to hear much difference.

-Carl

[Rob Mercure]

The Marantz 8B has 4 individual bias pots for each tube with an analog meter readout and switch selector for each tube. The THD % + N is a much greater concern in a HI-Fi amp at least back in the day because the music they were playing was distortion free

Yeah, but in the early 1950s the MacIntosh managed get superior numbers w/o any matching with their 50-W2 and 50 watts out of a pair of 6L6Gs! Split the load between the plate and cathode circuit and coupled one output tube's cathode winding to the plate of the other tube and vice-versa. No matching and THD and IM numbers that are still state of the art for tubes. More than one way to skin a armadillo!

Rob

PS: Personally I like individual bias controls for each and every output tube on any amp - who knows what I might want to throw in there!

[Enzo]

Armadillo? Possum on the half-shell.

[Steve Conner]

A few things that sprang to mind about tube matching.

AC signal matching: Most tube amps have negative feedback round the output stage. Won't this force the AC signal currents in the two output tubes to match even if they have different gm's?

DC matching: I fitted independent bias pots on my first (Toaster) homebrew amp. I just stick in whatever power tubes I fancy and tweak them to about the same idle current. I put quite a big range of adjustment on the bias, and played with weird power tube combos like one 6L6 and one EL34, or one 6V6 and one EL34. I run it at 475V plate voltage and a choice of 475 or 360V screen voltage, and 6V6s seem to hold together at the low screen voltage setting.

My favourite odd power tube combo was one EL34 and one EL37. It gave an excellent dirty sound at high volume, but the EL37 was an old worn-out one that began to run away and red-plate, so I couldn't use it any more. I'd love a pair of EL37s but they seem to be awful expensive >_<

[R ski]

Tube matching for guitar amps isn't going to hurt.

When using tubes for Hi Fi it does matter, distortion isn't
desirable.

I never found a output transformer balance out symetrically.

Due to that, swap tubes to balance out.

[sloblues]

Ray, this is excellent stuff. When you say "I've got two GE 6550A pulls here, idling 17mA apart (50/33mA), and yet they pull 198/197mA with signal applied" do you mean that you apply say a .125vac 400Hz signal into the amp, then crank it to max clean, or just max, output and then measure the DC current draw and it is nearly the same, despite the big mis-match at idle? And that is more an indication of a good set of power tubes than mere idle biasing might indicate? Also, if you please don't mind answering, what is the effect of the negative dc biasing voltage on the amp output as a whole, or the power output or balance of the amp? In other words, if I have seperate bias adjustment for each tube, and one tube requires -53vdc to idle at say 36ma ..then the other tube takes -38vdc to achieve the same idle. Does this negative DC bias voltage affect the quality or shape of the AC input signal, and, especially, does using seperate bias adjustments, so 'matched tubes' are not required, actually pose additional 'balance' problems since small to large variations on the amount of negative dc bias voltage will be required for each tube?

[Ray Ivers]

sb,

two GE 6550A pulls here, idling 17mA apart (50/33mA), and yet they pull 198/197mA with signal applied"

The 198/197mA measurement was obtained with a clean 400Hz sine wave applied to the power amp input, just below clipping.

And that is more an indication of a good set of power tubes than mere idle biasing might indicate?

Well, I don't know about it indicating a "good" set of power tubes, but these two were certainly more closely matched at full power than at idle - and if I had to choose - which hopefully I'd never have to - I'd probably pick matched-at-full-power.

if I have seperate bias adjustment for each tube... does this negative DC bias voltage affect the quality or shape of the AC input signal...?

No, it shouldn't.

does using seperate bias adjustments, so 'matched tubes' are not required, actually pose additional 'balance' problems since small to large variations on the amount of negative dc bias voltage will be required for each tube?

If multiple output tubes each use a different DC bias voltage, they will each start to draw grid current at a different point on the driving waveform, resulting in moderate to severe AC imbalance reaching a maximum at full power (which some may like, and other's won't). I personally would still used matched tubes even if bias pots were provided for each tube.

My own preference is for a two-tube output stage with controls for a) DC bias b) DC balance and c) AC balance, but YMMV - this is just what has worked for me.

Ray

[Arthur B.]

I have a mildly heretical view on matching.

Matching is what you do when you can't design the circuit to be tolerant of mismatching. Since we can't redesign amps (very much, at least) we're stuck with matching.

What is a matched tube? In order:
1. matched tubes DC-bias at the same currents or close so we don't have magnetic offsets in the OT causing us one-sided saturation grief or out-of-current grief on one tube.
2. If tubes are matched at DC, that does not mean they're matched for AC gain. So second order matching involves gain matching as well as DC bias matching.
3. If tubes are both DC matched and AC matched, they might also be distortion matched, so that the little irregularities in AC response are also matched. I'm not sure if anyone has ever done this or not. 1 and 2 are easy.

You can do 1 and 2 by brute force measurement and selection. Or you can do it actively.

By actively I mean matching DC bias by matching the tube current with individual bias voltage settings. That largely eliminates the need for DC matched tubes.

You can also AC-gain match tubes by putting in a single sided post PI master volume of limited range. By turning down the available signal to the hotter gain tube, you can match the overal PI-to-output-tube-plate gain.

So for two added pots to tune in a new set of tubes, you can ditch buying matched sets.

I don't have a good solution to the matched-distortion squiggles.

Yet.

Balanced negative feedback and cross coupling would do the trick to match AC and DC characteristics.


Yeah, but in the early 1950s the MacIntosh managed get superior numbers w/o any matching with their 50-W2 and 50 watts out of a pair of 6L6Gs! Split the load between the plate and cathode circuit and coupled one output tube's cathode winding to the plate of the other tube and vice-versa. No matching and THD and IM numbers that are still state of the art for tubes. More than one way to skin a armadillo!

Rob


I can't find a schematic for the 50-W2, but it seems it must be different from the usual Mcintosh since they cross couple the plates and the screens, not the plates and the cathodes. The other is part is of course the unity coupling circuit, which feeds back all the gain.



I'm probably going to get crucified for saying it but in my eyes the whole argument is pretty much academic. Think about it for half a second! We're overdriving the hell out of the phase inverter, pushing the power tubes to the point of grid blocking and saturating the cores of the opt. So we're going to get our panties in a bunch over less than 1% thd from a set of tubes that isn't perfectly matched?! UGH!!!!

-Carl

What happens if the power amp isn't overdriven, and the distortion comes from the preamp, like most modern guitar amps?