Thanks a lot for the clarifications.
[quote author="PRR"]Unless stated otherwise, I assume "matching" means two tubes of the same brand and decade, or maybe (if the seller really cares) two tubes that give similar current at some specific (not necessarily realistic) bias condition.[/quote]
If people do just the former then that sounds a bit like people selling a matched mic-pair: trying to convince you of good matching by mentioning that the serial numbers are near. :wink: But sure, same brand & decade would be some minimum requirement.
The latter was probably what I saw being mentioned (those mA-figures).
Or.... two tubes showing similar readings on the "Good-Bad" emission test, the only test on an old cheap tube "tester". You wire all the non-cathode electrodes together, put about 20VAC between this and the cathode, and read current. For oxide cathodes, this should be 5 to 50 times any normal operating current, but it really makes no difference if it is 5 or 50.
Is this perhaps the test that's at hand when it is stated that a certain tube is 95% on this or that tester ? I'm not familiar with tube testers, so I'm puzzled by what that % is actually indicating.
Matching for more than one parameter will usually give either no matches or very poor matches from a crate of tubes. You need to sort a LOT of tubes to give good matches for more than one parameter: instead of a one-dimensional problem, it becomes two-dimensional, three-dimensional, maybe 5-dimensional. A box of tubes can scatter all over 5-D space.
I see what you mean, it'll become hopeless fairly quickly. Jumping to discrete opamps, matching BJTs for Vbe & beta (fwiw) is already some burden.
Just the I/V plot is 3-D (plate voltage, grid voltage, current). Look at a plate curve plot and ponder all the ways a specific tube could vary yet still not be defective.
There do seem to be a few shops that do better than a 1-D match.
Mostly, I say if you "need matched tubes", you should probably redesign your circuit to be more tolerant of tube parameters.
Probably best is to build the circuit & plug a few tubes and measure the actual performance of the tubes. Taking a stereo G-Pultec gain make-up amp as an example, I expect differences between tubes will show up as differences in gain of the SRPP-amps. As long as that 'matches' well under identical conditions for both channels, I guess the other eventual differences will be small enough to worry about.
> mu, S, Rp
These three are inter-related, as you say. If you know two, you know the third.
And Mu is a geometric property. You can measure it with a ruler. While most tubes actually act like several tubes with several Mus as the main section goes into cutoff and stray paths begin to dominate, over the normal working range Mu is either darn close to nominal or the tube is screwed-up. So it may only be necessary to measure S, but over a wide range of voltage and current (unless the matchmaker knows the specific condition you will work at).
Interesting, didn't know that about Mu.
I'm getting me some PCC88's, if the SRPP-gain test mentioned above still is unsatisf. I could measure S.
Thanks,
Peter
