Is 10% good enough for tube matching?

I ordered a tube from Upscale Audio and received T1:12500 T2:13000. I ordered another one to match it and gave them the T1 and T2 and received T1:13500 and T2:13700. I called them and was told that as long as they are within 10% then they consider them matched. Is 10% close enough?

 

Informative read on tubes by Roger Modjeski




Mu (voltage gain), Gm (transconductance) and Rp (plate resistance) are the three electrical characteristics that make up the overall electrical characteristics for a vacuum tube. . There is much confusion in the audiophile community about these factors as they affect tubes and their performance in components. Written by world-renowned amplifier, Roger Modjeski, this article, will look at what each of these characteristics mean, how they are measured and what each of them means to the sound of a component. We will also look at tube testing, tube matching and which characteristics are most important when matching tubes for a given circuit.

How these measurements are applied and their relevance differs for triodes and pentodes. Triodes and pentodes both possess the characteristics Gm (transconductance) and Mu (voltage gain). However what is important in triodes is Mu and what is important in pentodes is Gm as these parameters are the dominant characteristics affecting performance when each is used in a typical circuit.

Triodes
Triodes have an interesting relationship where two parameters relate to make a third. These parameters are locked into a relationship so that knowing any two will yield the third. Just like ohm's law. In the case of triodes, Mu = Gm x Rp. Gm has the units of mhos, (amps per volt) and Rp has the units of ohms (volts per amp) and when multiplied their units cancel, making MU (a measure of voltage gain) unit-less as it should be. A Mu of 30 means that what goes in comes out 30 times bigger.

Triodes are most widely used for voltage amplification in our preamps. Here, Mu (voltage gain) is the more important parameter to measure and most directly affects what you hear. If you have a tube with a Mu of 30 in one channel and a MU of 33 in the other, you will have a 1-dB imbalance in your preamp. So let’s say you get two triodes matched for Gm. There is no assurance that these tubes will have equal voltage gain. Voltage gain (Mu) is the product of Gm and Rp. So if Gm is matched to, get matched Mu (the important characteristic in this circuit) Rp must be matched also and Rp is almost never measured. So as you can see although the famous Hickock tube testers are sought for their ability to measure Gm, this was more of a marketing ploy and less useful for real-world testing. If the vendor measuring Gm would give you the Rp then you could calculate the Mu by simple multiplication. The reason you don't get the Rp data is that it is hard to measure and no commercial tube tester ever measured it.

Modern audiophiles are more concerned about "matching" than ever before. They want everything to be matched, though they rarely know what it should be matched for. This is a dangerous situation as many of you are buying because the vendor says it is matched, though it may or may not be matched for something unimportant while something important is not matched. Gm and MU are a perfect example.

Rp or plate resistance is the second most important parameter as it directly affects the output impedance of the preamp. Circuit designers are looking at Mu primarily, Rp next and we hardly care about Gm at all. Why then do most vendors measure Gm and Gm alone? Because most testers only measure Gm! No commercial tester ever measured Mu either. And even the Hickock, treasured for measuring Gm, doesn’t measure it meaningfully (due to current dependency), accurately (due to calibration difficulties) or consistently (due to internal circuit drift).

Since Mu is what controls voltage gain, and voltage gain is what we listen to, then lets measure that directly. That is just what we do at RAM TUBE WORKS. Of course we had to build our own tester to do that. A few years ago I heard that one of my competitors tried to duplicate the RAM tester and gave up after spending $100,000. It took me about a year to develop the small tube tester and another 6 months to develop the power tube tester. These were no small efforts.

Pentodes
Pentodes are an entirely different matter. For Pentodes, as used in our power amp output stages, Gm does indeed matter and must be matched for good performance. Not only must Gm be matched, but it must stay matched over the range of bias currents typically found in amps.

In pentodes, Rp is very large and is swamped out by the load impedance. Therefore, gain is Gm times the load impedance presented to the tube (through the output transformer). Gm is the thing to measure here but the Hickok does not measure anywhere near the real world operating voltages and currents found in a typical circuit. Given I had an accurate way to measure bias and Gm at real operating conditions in a tester of my design, I found that tubes matched for bias and matched for Gm would "track" each other over the wide range of bias currents encountered in power amps. Without this accurate "double match" there would be no guarantee that tubes measured at 50 mA (where we measure) would match at 30-70 Ma where sometimes used, or match over the range of screen voltages from 300-450 V encountered in the range of Pentode circuits (RM-200, RM-10, ARC) to Ultralinear circuits (RM-9, MANLEY, CJ).

Effects of Tube Aging
Regarding a tube’s performance as it ages, these parameters will remain stable as long as the cathode emission is 70% or more of its original value. If Gm, Rp and Mu are right and matched when new, they will stay right and matched through the usable life. Given that, all we need to measure is the emission to know the tube is still good. Only in rare cases will this not be true. At RAM, our job is to check up on the tube makers and verify that they keep parameters within the published range (we reject tubes outside that range) and match for variations within that range. In the large batches of tubes we measure we see exactly what would be expected: Bell Curves for each parameter. See my article "Why the sound of your preamp changes when you change the tubes" for more on that.

 

So what are the T1 and T2 values? How do they relate to Gm, Rp and Mu?

 

My guess is that T1 is Gm of the first triode in the tube, T2 is Gm of the second triode in the tube. So, given the above discourse, is Gm in preamp tubes even worth matching?

 

oregonalex,
In answer to your question, I don't know but I do know that Roger Modjeski
knows what he's talking about when it comes to this subject..

Just got a "matched" pair today.
Telefunken Platinum E88CC, 16500/14500, 16000/14500
Whether these are matched or not is a matter of opinion.
I don't know exactly how they measure their tubes but
they seem to maintain good quality control.

 

It is important that your output tubes match to reasonably close others are not as important as I have read.

 

In a hand-wired, discrete component amp, dang near everything in there has a 10% tolerance or higher (resistors, caps.....). Getting better (stated) tolerances than that gets pricey, quickly, in my experience. Having read Reb's post above, it seems even sillier to me not to have balance controls on amps. Maybe I'm missing something.

 

I have a push pull KT88 amp and one tube went out. Is it fine if I use a totally different brand of KT88 that is not matched with the other three? I don't want to buy another tube just to match the appearance of the other 3.

 

Depends on the amp. Some are okay without matched output tubes; others will perform poorly. Your Line Magnetic amp (216IA?) looks to want two matched pairs, so swapping in a non-matched tube in one of the sides may give you poor performance or increase the possibility of red-plating one of the tubes on that channel. You can try it, but reduce the current to the tubes to a minimum and then work up to the usual bias point -- watching the tubes closely for any problems. Then keep watching them for a while afterwards. Ideally you'd swap out both tubes on that side.

 

I'm wondering if it is better to use one JJ KT88 and one of the stock Line Magnetic KT88's for each side? I have two blue tube JJ's. Or is it best to use the two JJ's on one side and two Line Magnetic (Shuguang) one the other side? But I wouldn't like the imbalanced appearance

 

Using different brands of tubes in a pair is a singularly bad idea IMHO.
Using different brands in each channel is not going to do your system any sonic favors either.

 

The matching of electrical characteristics between the two tubes in each channel is more important than who made each tube. It's probable that your JJ and Shuguang tubes aren't matched though, so I probably wouldn't do it.

 

Okay, to both of you and thanks. I'll just wait and get two more blue JJ's. I have other amps to use in the meantime

 

Seconding Steve, Yes.

 

I buy my 6C33Cs from Lamm - they are matched to the serial numbers of my amps (although I honestly don't know what he does). All I know is they come marked on the boxes with the S/N of the amp they should go into and whether they are in position V1 or V2. Peace of mind.

 

I don't see the problem with trying out different brands of tubes, if a replacement isn't currently at hand - surely only the sound could be compromised (or not - you should be able to hear for yourself) , but nothing else will be damaged ?

 

Of course, using different tube pairs in each channel cannot cause any damage, only sound issues.

But, using unmatched tubes in a push-pull circuit (and two random different brand tubes with different internal construction are likely to have significant differences in parameters) could unbalance the circuit enough to cause one of the tubes to work much harder and fail prematurely and who knows what other components could be taken out with it. I don't know how likely it is to happen, but I think it is not worth the risk even if the chance is small.