Copyrighted by Bob Putnak, all rights reserved.
“Sometimes output tubes must be selected which will provide a satisfactory balance adjustment. A tube tester usually will indicate whether a pair of tubes have reasonably similar characteristics.” – Robert Middleton, 101 Ways to Use Your Audio Test Equipment, Howard Sams Inc. 
“Tube matching” is a controversial and complex topic, and there is no consensus regarding “what is best.” In fact, some people even feel the whole topic is a waste of time, arguing that matched tubes “kill the mojo” of what makes a tube amp sound special.
People ask me about tube matching using a tube tester. In most cases, they just want to do a reasonable job at matching tubes for themselves, and they have reasonable expectations. Others are not satisfied unless some guru tells them they need to spend big bucks buying “matching” gizmos that will magically take care of it.
I will try to provide an simple overview of the tube matching topic vis-a-vis a tube tester. Since the topic has no absolute answer, no conclusion can be offered…only opinions. It is important to remember that this discussion has nothing to do with the importance of a tube tester as a diagnostic tool. As a diagnostic tool, tube testers excel. The question is whether they also do a good job at tube matching.
Modern thinking regarding proper tube matching falls into four primary schools of thought (…while acknowledging that other opinions exist):
(1) one school of thought is that the output tubes should be matched for idle current draw at the amplifier’s working voltages. This measurement is either plate current, or cathode current, while the amp is on and idling, but not working. Both parameters are important — amp is idling, and the tube is seeing the amp’s actual operating voltages, which are often in the 400v+ range.
Analysis: No vintage tube tester will meet those criteria, so if this is your preference for matching, your tube tester will not assist you.
(2) another school of thought takes method #1 to another level and advocates that the output tubes should be matched for idle current draw at the amplifier’s working voltages, plus equal AC balance . The second factor is a measurement of the tube’s amplification capability, transconductance.
Analysis: We already established that the first factor is not something that a tube tester will perform. The second factor, measuring the tube’s amplification, is more complex. It is true that a mutual conductance tube tester (such as Hickok, many models of B&K, and others…) will measure this characteristic, but not at the operating voltages in a real amp circuit. In essence, just because a pair of tubes are matched for transconductance at a plate voltage of 150v or 200vdc and a particular DC bias in a tube tester, does not always follow that they will be matched at the plate/screen/grid voltages used in your amp. Tube characteristics curves are not linear, therefore matching at one spot on the curve does not guarantee matching across the entire curve. Hence, measurement in a transconductance tube tester is only a “very good prediction” that the pair of tubes will match closely in a real amp.
A sidenote to this point is regarding what qualifies as a transconductance tube tester. Some testers are obvious, but what about “Dynamic Conductance” testers such as Eico 666 or 667, Precision Apparatus 10-series, Jackson 648? These test methods are often named “Dynamic Conductance”, or “Dynamic Plate Conductance”, or “Plate Conductance”, or “Electronamic”.
Mannie Horowitz argues in his book “How To Troubleshoot and Repair Electronic Test Equipment” that dynamic conductance testers are in the same category as the classic transconductance testers, and they provide a valid measurement of the tube’s amplification capability, albeit using a different test circuit to accomplish a similar evaluation.  In fact, he uses the Eico 666/667 circuit to explain the principle. This opinion is also shared by the engineers at Simpson Electric Co  and Precision Apparatus Co . You can read these footnotes below for more information about Dynamic Conductance and Mutual Conductance.
(3) the third school of thought is to use a curve tracer to properly match tubes. A curve tracer will overcome the “characteristics curves are not linear” issue, permitting an evaluation of tubes based upon how they perform across several points of the tube’s characteristics curve.
Analysis: a tube tester is not a curve tracer, and still does not evaluate how much residual distortion will be present in your amp.
(4) the fourth school of thought is “total distortion” — meaning that output tubes are truly matched when the total distortion of your amp is lowest vis-a-vis another set of tubes.
Neither a tube tester, nor a curve tracer, measures an amplifier’s output distortion. For that, you need an automatic distortion meter and low-distortion signal source, and a lot of experience. The tubes are placed into the amp, biased correctly, and driven to max output before clipping. Distortion is measured. Tubes that do not pair well together will have higher residual distortion than a nicely matched pair. This method also allows you to measure the amp’s power output from each pairing of tubes.
In my opinion, this is the only real method to accomplish “matching”, although it is not strictly a “tube-matching” issue and therefore beyond the scope of what most end-users want to tackle. It is holistic health care for your amp. You are evaluating all aspects of the amp simultaneously and how they interrelate to each other. You are tweaking the entire amp (including the tubes) to achieve the cleanest output. You are tweaking the bias. You are tweaking the “AC balance” / “phase inverter”. You are swapping/mixing output tubes. You have objective scientific evidence (the distortion numbers) to prove how your amp is working. When you can’t get the amp any cleaner at the same power output level, you know that the amp is optimal within its design limitations. This requires a lot of experience as a technician and quality test gear.
Honestly, this is the “right” way in my opinion and anything else is a serious compromise. But since this would require everyone to become a technician, folks are looking for an “easier” or “simpler” way to match tubes.
Final Thoughts regarding tube matching and tube testers:
Again, it is important to reiterate that this discussion has nothing to do with the importance of a tube tester as a diagnostic tool. The primary purpose of a tube tester was “to find bad tubes”  — tubes that could be causing a problem in the equipment that the technician was repairing. Tube testers excel in this role. Just about every vintage repair book correctly suggests to “test the tubes” as a technician’s first diagnostic step. The question is to what extent a tube tester is useful for tube matching.
I have no strong opinion regarding what is the best way to match tubes, except for evaluating the total distortion of amp. Beyond that, the rest of these methods are all compromises. That said, I highly respect the books written by Bob Middleton and typically follow his methods. Use whatever method works for you, or ignore it completely if you believe that matching “kills the mojo”. I can say that in the old days this topic was nowhere near as important as it has become on the internet era. Many old-timer technicians were satisfied using a tube tester to match tubes — assuming that they bothered to make any attempt at tube matching whatsoever. Matching was not taken any further unless you could hear hum or distortion caused by grossly mismatched tubes.
Another practical factor to consider, for those people who only buy vintage audio tubes, is how likely is it that you will find enough vintage tubes to end up with pairs or quads that match using a method such as #2 above?
Likewise, can you hear the difference between a perfectly matched pair (however that is defined…) vs a pair of NOS random tubes? Have you ever done a blind objective listening test? And how long will your pair of perfectly matched tubes stay matched? They certainly will not age identical.
In summary, there is no consensus here. Some degree of common-sense and reasonableness must be given strong consideration.
I will leave you with these two quotes from old-school expert sources that I respect and mentioned above:
“The push-pull output stage of an audio amplifier should be balanced to within 5%, and preferably 2%, on a cathode-current basis. When tubes are unbalanced, hum, harmonic distortion, and intermodulation distortion increase. Sometimes output tubes must be selected which will provide a satisfactory balance adjustment. A tube tester usually will indicate whether a pair of tubes have reasonably similar characteristics.” – Robert G. Middleton, 101 Ways to Use Your Audio Test Equipment, Howard Sams Inc, see footnote 1 below. Bob Middleton was the author of approximately fifty books on tube-era electronics, many of them published by Sams. (Italics on the word “sometimes” added by me to emphasize the reality that often you can’t hear any difference whatsoever.)
“The dynamic Plate Conductance tests which you make with your Simpson Tube Tester Model 1000 will furnish indications which you may use to match the characteristics of several tubes of the same type for circuit applications which require balanced action. The same tests can be used to match two sections within one dual-type tube. Examples could be push-pull amplifier circuits in the amplifiers of an oscilloscope, and two sections of an electronic bridge network.” Simpson Electric Co., footnote 5 below.
Certainly you will find people who disagree with those old-time expert sources, so you have to decide what you believe, what is “reasonable”, and how far you want to pursue it. If tubes were cheap and plentiful, perfect matching would be more accessible. With scarce availability of vintage power output tubes, and pricey Russian imports, it seems as though some degree of reasonableness is necessary.
Footnotes and References:
Incidentally, Mr. Horowitz also has an excellent (and very technical) book published by Howard Sams entitled “Measuring Hi-Fi Amplifiers”.
Simpson Electric Co. was the manufacturer of the famous Simpson 260 meter, along with countless other quality electronic meters and equipment. I consider them to be a highly credible source. FYI, the Simpson 1000 tube tester is equivalent to an Eico 666/667, so I feel the explanation from Simpson is relevant to the discussion of comparing Dynamic Conductance and Mutual Conductance, and in regards to tube matching via a tube tester.
The manual states: “Plate conductance measurements are dynamic tests which indicate, in each test, a combination of the ability of the cathode to emit electrons, of each grid to affect the plate current in its proper manner, and of the plate to receive the regulated current.” – page 2.
“…Amplification Factor is equal to Mutual Conductance times Plate Resistance. …As the tube ages, its Plate Resistance will increase. This increase in Plate Resistance cause the Mutual Conductance to decrease, and a measure of either parameter will indicate the relative value of the tube. …The Simpson Tube Tester model 1000 measures Plate Resistance through its mathematical reciprocal, Plate Conductance. Thus it provides a dynamic test of the ability of the tube to operate in a circuit application.” – page 3.
“The percentage of Plate Conductance indicated on the meters of several Tube Testers Model 1000 are made to agree with the percentage of rated Mutual Conductance measured on each sample tube with a laboratory type Mutual Conductance Bridge. This correlation assures you of proper indications for evaluating all tubes according to standard rating systems when you test them on your Simpson Tube Tester Model 1000.” – page 4.
Regarding the topic of Tube Matching using a Dynamic Conductance tester, Simpson writes:
“The dynamic Plate Conductance tests which you make with your Simpson Tube Tester Model 1000 will furnish indications which you may use to match the characteristics of several tubes of the same type for circuit applications which require balanced action. The same tests can be used to match two sections within one dual-type tube. Examples could be push-pull amplifier circuits in the amplifiers of an oscilloscope, and two sections of an electronic bridge network.” – page 42.
5 Principles of Electronamic Tube Testing: More than just Mutual Conductance, Precision Apparatus Co. Inc. I scanned this article for download [HERE].
6 Principles of Electronamic Tube Testing: More than just Mutual Conductance, page 1. In emphasizing that the purpose of a tube tester was “to find bad tubes”, Precision used all capital boldface letters to emphasize the point.