Matched For What?

I have written several times on this topic, anyone interested in the finer points of tube matching should read them. There is no such thing as a tube matched for everything, so the buyer needs to inform the tube seller what parameter needs to be matched. The problem is that the customer often does not know what to ask and the seller probably could not match for it anyway. However, all is not lost. The manufacturer of your equipment should be able to tell you what kind of matching is needed. Unfortunately, most tube sellers only match for one thing, Gm, which is basically useless. I imagine the conversation between the customer and tube seller goes like this:

Customer:  "Are your tubes matched?" 
Seller:  "Yes, how many pairs would you like?”

Here is another conversation about matching, albeit with some embellishment on my part:

Customer: “Are your tubes matched?”
Seller: “Yes, of course, we always match tubes. Would you like them the same size or the same color.”
Customer: “Huh?”
Seller: “Just kidding. What do you want them matched for, what parameter?”
Customer: “I don't know, I was told to ask for matched tubes. How do I know how you match them?”
Seller: “Of course we do? We have a calibrated Hickok tube tester. We make sure the needle goes to the same place.”
Customer: “Same place for what?”
Seller: “I don't know, we have some guy in the back that turns the knobs and looks at the needle. The numbers are not important as long as they are the same. The meter has some big numbers and with the word transconductance under them. There are lots of numbers, we make sure they are the same.”
Customer: “How close to the same?”
Seller: “The guy writes the number on the box then he puts similar numbers together. We are getting too technical now, they are matched, I assure you.”
Customer: “Well okay if you say so, I will take a pair”
Seller: “How about the color?”

This lesson can be summed up nicely with a quote from the tragicomedy, Rodencrantz and Guildenstern Are Dead, written by Tom Stoppard:

"We only know what we are told, and even that isn't true"

Here is what is true though unfortunately a bit complicated.

The simple and correct answer is: Tubes should be matched for Gm and Bias (two-point matching) which is something I believe I was the first to discover around 1980. Now everyone claims to do it, though most do not do it right because they really do not understand it. Some have even instituted three-point matching because that must be better than two. I can imagine matching for a great number of points (parameters), but it gets silly to match for things that normally do not matter like matching screen current (that is what the third point often is).

Traditionally power tubes were stuck in a fixture with a set plate voltage and screen voltage (often the same as the plate to make things simple) and the plate current was measured. Two tubes with the same plate current were then considered "matched". However, if the tube is used at a different plate and screen voltage than the test, what would happen to that match? The answer is the tubes would no longer draw the same plate current, in fact it could be very far outside of a match.

What I discovered is that if both the plate current and transconductance (Gm) were matched in a pair of tubes they would "track" each other over a very wide range of plate and screen voltages. Thus, a pair of output tubes used in a Dynaco ST-70 where the plate and screen (ultralinear connection) were at the same 475 volt potential would remain matched at a much lower screen voltage of 300 volts in regulated screen amps like Audio Research. Since screen and plate voltages, not to mention bias voltages and cathode currents, are completely at the command of the amplifier designer one had to find a way to make a group of tests work for all.

Perhaps I was lucky to have seen this or perhaps I saw it because I like to measure things and design amplifiers. To my knowledge, most, if any, tube sellers do not design amplifiers. They either buy a Hickok or have someone make them a traditional tester, possibly with a computer to record the data, but my computer does much more than that. It sets each tube to a constant plate current before the test is made. Thus, the bias and transconductance are measured at the same plate current for every tube.

The problem with the fixed grid bias test is that the tubes will settle at different plate currents. Measuring the Gm at these random plate currents will not produce the same results as measuring them at a constant plate current as Gm goes up with plate current quite sharply. I use 50 mA for most tubes as that is a good bogey value for many amps. Even though I run tubes in the RM-9 at 30 mA the matching holds very well. Proof of concept.

Other than for special applications like differential amplifiers or RM-9 input tubes, matching dual triodes is often a fool’s errand. What you want in a preamp tube is low noise and low microphonics. Of course, we can match for gain (mu) if you know how the sections are used. In some preamps the two tube sections are used left and right. In other preamps they are used one complete tube in two stages left and one right. In any case if there are two stages in cascade (one stage following the other) the gain is the product of the section gains. 

If you want to know more about tube matching click here:

[Source: circa 2011]