Matched Tubes?

[Emetal]

I got "Matched Quartet" of JJ EL34. When installed they read: 30mA, 30mA, 28mA and 35mA Is that considered "Matched" because 28mA and 35mA are quite apart?

[Rockwell666]

are they burnt in or is this the initial reading? also where did u get the tubes? some places match better then others.

[Reeltarded]

I never buy matched tubes. I just buy tubes.

Have a pair of "matched" (LMAO) 6ca7s in Petey drifted 11ma within 6 hours.

We are nearing the end of viable tube production, me thinks.

[Phil_S]

I'd try the 30+30mA on one side and the 28+35mA on the other side. If it sounds OK, then it's OK. If it doesn't sound right put one 30mA in each side and see if that sounds better. If it sounds bad, see if you can get one more tube closer to 30mA. 7mA may be more difference than you want, but it could be OK. Run the tubes for a while and check them. You are likely to find they've drifted anyway.

[Emetal]

are they burnt in or is this the initial reading? also where did u get the tubes? some places match better then others.

It is initial reading, yes they might get closer later but also they might get further apart. Got them from Thomann

[pdf64]

Thomann offer a 30 day 'satisfaction guarenteed' money back, plus a 12 month and 3 year warranty!
These are TAD selected tubes and I get the impression their process is pretty good. But nothing lasts forever, anything can fail at any time, tubes are a case in point and power tubes especially so as they are such mechanically fragile structures and so liable to a bash during transit causing change / damage.
https://www.thomann.de/gb/helpdesk_moneyback.html
Perhaps give them a few more hours to assess how stable they are, and get back to Thomann if you're not happy.

[Rockwell666]

[/quote]
It is initial reading, yes they might get closer later but also they might get further apart. Got them from Thomann
[/quote]

I have seen many tubes drift closer to matching after a few hrs of burning in. also what PDF64 said

[R.G.]

You can worry about matching, or you can force it. One bias pot per tube will go a long way toward eliminating matching charges and issues.

As you've found out, "matched" may not always mean the same thing to different people.

There are a couple of valid reasons to get matched tubes. In guitar amps, it's generally because you need to have equal DC bias currents in each half of the OT's two primaries. This cancels out the DC offset in the iron and gives you the most use of the iron and copper. It was once so common for there to be zillions of near-identical tubes made that tube to tube matching was assumed. The amp makers then put in only one bias control, assuming "matching" was good enough.

That's no longer true. They're different, and they drift. Sometimes a lot.

If you're going to measure the idle current per tube, why not put in a control that lets you set it per tube?

[Littlewyan]

So back in the 60s and 70s tube matching wasn't an issue? You could just buy two separate EL34s and they would be very closely matched? I've often wondered this. My grandad told me he never used to worry about matching them.

[R.G.]

Like all casual assertions like that, there are a batch of asterisks that go with it.

It seems to be true that the bigger the batch you make things in, the more alike they're going to be, one to the next. Tubes were cranked out on hard-automated production lines back in the "Golden Age", and once a manufacturing line gets tweaked in and running well, it tends to be consistent. The companies that made the things worked HARD at making every single unit off the end of the line be consistent. So when GE, or Sylvania, or RCA, or Muller, or Raytheon, or [insert your favorite maker here] shipped out thousands of crates of tubes, there was a reasonable expectation that any two tubes on a given truckload would be really close.

That doesn't mean that Muller tubes were the same as Sylvania, for example. The two different manufacturing lines would have different distributions, all falling inside the range of the data sheet. So while you probably could put any two EL34s in an amp and have it work, it would be smarter to put any two of the same manufacturer and approximate time of manufacture.

With some notable exceptions, you could probably put any two new, correct-part-number tubes in an amp and have it work. Of course, the batches are much smaller, and there is a pressure to ship every single tube, not to compete on quality in all your tubes, so more "outlier" tubes get shipped, the ones that work for a few minutes and then short, or simply quite working or suffer from too much gas. Not a lot overall, but a guitar player doesn't have to have one of his output tubes go up in flames very often to get picky.

The variation in characteristics between tubes is wider today than it was in the Golden Age. So we need to compensate for it by being ready to tweak in the wider variations in tubes.

[Reeltarded]

I never bought an incapable tube until maybe 2000 or so. A preamp tube would die, an output set would wear out after many, many hours of hammering them and bouncing around in cartage, but until the age of Russo-Chinese production.. meh. The worst I ever had to go through was putting the weak one in the last hole. I bought a lot of GT remarked Phillips for a while.

I take some of that back; RFTs I had some issues with around 1990.

The consistant production thing is real. I believe GT remarked set rating put every tube on the street. Now, DO NOT PASS tubes end up in stores matched with other failures instead of a bad glass pile to the sky.

[Stevem]

Let's talk about the sonic effect of output tubes or tubes in PI sections .
Once either of these gets far enough away from matched output you get increasing levels of phase cancelation taking place , this is a form of distortion which in and of itself is not a bad thing for guitar tones, but it starts to cut deeply into your guitars sustain level be it clean tones or distorted!

While on the subject of sustane and distortion , the reason we players want distortion / clipping / compression IS for the sustane it provides , so killing some of that sustane by running too out of match tubes is self defeating if you ask me!!

[Stevem]

Let's talk about the sonic effect of output tubes or tubes in PI sections due to matching or not.

Once either of these tube sections gets far enough away from matched output you get increasing levels of phase cancelation taking place fast , this is a form of distortion which in and of itself is not a bad thing for guitar tones, but it starts to cut deeply into your guitars sustain level be it clean tones or distorted!

While on the subject of sustane and distortion , the reason we players want distortion / clipping / compression IS for the sustane it provides , so killing some of that sustane by running too out of match tubes is self defeating if you ask me!!

Also note that as tubes get weaker they rev up faster if you will as signal is applied, this means that even if you had a Separate bias control for every output tube to get them perfectly matched at idle your still going to get uneven output and in turn phase cancellation!
This is a very important point to me and to sound ,and even perfectly idle current matched new tubes may not rev up the same at different frequencies!

This to me is why the Groovetubes matching process is superior, as in part of there matching process they also test tubes at 3 different frequencies to test out how matched they are in that important regard!

[martin manning]

Matched tubes have the same Ia at a given Va, Vg2, and Vg1, and the same Gm. You can dial in the idle Ia for two miss-matched tubes by adjusting the Vg1’s, eliminating the dc off-set, but the Gm’s might still be different. What that produces is second order harmonic distortion. Miles is on it, the GT system just matches similarly off-spec tubes... something gained and something lost. Isn’t that what most vendor matching processes are doing now? They are making money doing this, of course...

[R.G.]

You're right Martin. That's the very next level of approximation to step to.

One useful approximation to tube behavior is to look at the curve of plate current versus grid voltage. This transfer curve is pretty commonly available, although not as common as the plate characteristics. The tube's gm is the slope of this line, the number of amperes change in the plate for a volt of change on the grid. But the line is a curve; gm isn't constant. It's close enough to make a single value of gm a useful approximation, though.

Gm is also different from tube to tube. This is why different tubes have different bias voltages, as down at the low current end of the gm curve, the grid voltage is nearly cutting off the plate current, and small differences from tube to tube matter. This matching problem is important especially in a push-pull amp because we need balanced DC in the OT. So we just pick tubes that happen to have similar currents at the same grid voltage. That's Matching 1.0.

Matching 1.1 says that we'd like to have them do the same amount of amplification too. That is, we'd like for one voltage of change in grid voltage to make the same change in plate current for our "matched" tubes. What this amounts to is demanding equal AC gains for the "matched" tubes. Since the gm curve is a curve, there's probably a Matching 1.1.1 that says to match the incremental AC gain at multiple points along the curve. I don't know of any matching process that does this, although with the rise of computerized matchers, it's entirely possible to do these days.

I can guess at a Matching 2.0 that says to match the frequency response of the tubes, amounting to selecting the tubes based on their parasitic capacitances, and a Matching 3.0 that matches their distortion responses.

However, the question then becomes where do you stop? Getting close-if-not-identical idle bias currents is a good thing, and is easy, so most "matching services" do this. Problem is, they almost certainly only match at one point on the curves. They set a grid bias voltage, measure the plate current, then label the tube with this current. The actual "matching" is to look over a box of tubes for two, four, six, whatever number of tubes with "close enough" current labels, then box them and ship them. I doubt there's much in the way of saying that they all have to be within X percent of one another. This makes me wonder how close the currents would be if measured at a slightly different grid voltage. I'm pretty sure that the "matched sets" at one grid voltage will change at another grid voltage.

Then there's the AC gain part of it. We approximate this by looking at gm. Gm isn't one number, but it's close-ish for similar production tubes. So now to get matched tubes, you really ought to measure each tube for bias voltage AND gm, and only match two (or more) tubes if they're within X% on bias voltage and Y% on gm. How many tube matching setups do you think do that? It's possible, of course, but the yield of matched tubes per cubic yard of raw unmatched tubes goes down dramatically. Bad economics there.

It's also a little futile. In Class AB amps, we deliberately let one tube handle mostly positive half cycles, the other mostly negative half cycles, and we obligingly separate these for them in a phase inverter. Phase inverters are not perfect, either, and the effective "gain" for positive and negative half cycles isn't the same. Using this imperfection, we can actually game the system, deliberately offsetting the gains in the phase inverter to make up for (or make worse, depending on taste) the differences in gm in the tubes actually in the sockets. That lets you "match" unmatched outputs much better than hoping that the guy in the shop really did pick good ones for you. And perhaps more importantly to guitarists, deliberately mismatch to taste.

I'm not familiar with the noted use of three frequencies for testing at ... um? Groove Tubes? That seems like a futile thing to do in my mind. There will of course be differences, but we already have power tubes that have responses above audio or we wouldn't be using them; there are differences in little oddities down in the gm curves, but I'm not sure what the three frequencies would tell you unless the test happened to be sitting on top of a little kink the the gm curve. But it is something you could do by looking at meters and labeling, and it makes for good advertising copy.

I'm still back at my first position. If you want matched tubes, buy tubes, bias them to the same offsets for your OT, then if you want to play with the AC gain match, diddle the phase inverter to match/not-match the gm vagaries. I have a whole polemic about how likely this is to happen, given the leanings of guitar players and amp hackers, but that's another story.