Using ST-70 for tube matching (w/o tube tester) - comments?

[dcriner]

My tube tester is just a simple emissions checker, so I can't really match tubes. However, I came up with this procedure (which may not be a new idea); please tell me if my logic is sound.

Pull all four of the EL34 output tubes. Select one socket for testing. Plug in each tube into that socket, one at a time, and measure the bias voltage in the usual manner. (What we are really measuring is the current through the 15.6-ohm cathode resistor.) Don't touch the bias pot during this testing.

If the tubes are perfectly matched, we should measure the same voltage for each, correct? (Because only one tube is installed, the voltages wouldn't be expected to be the normal 1.56V, but should be the same or close for all four tubes.)

I'm running four JJ EL34s that were supposedly matched by the tube vendor. Well, using my test procedure, I measure anywhere from 0.89V to 1.06V -- a significant difference in cathode current.

Please comment. Thanks.

Doug Criner

[erichayes]

Hi All,

Your idea is basically sound, Doug, but it points out the flaw of single-point testing. Unfortunately, even a Hickok 539C, the holy grail of tube testers, can't do more than a single point test without lots of external componentry.

Ideally, tubes should be matched both for cathode current and transconductance (aka mutual conductance). But for our purposes, matching cathode currents is the more important criterion. What you should do, Doug, is set the bias control with one tube in the amp so that you do get 1.56 volts across the cathode resistor. That will at least establish a baseline in accordance with what Dynaco had in mind. Check the other tubes at this bias setting and see how much deviation there is. Go back to the original tube and decrease the cathode current by, say, 10%, then retest the remaining tubes and record the current. Repeat the process, this time increasing the original tube's cathode current by 10%. Match the tubes according to whichever test currents were closest for an arbitrary set of tubes. For instance, tube A might be 1.56V, 1,40V and 1.72V; tube B is 1.58V, 1.25V and 1.66V; tube C is 1.57V, 1.25V and 1.69V; tube D is 1.60V, 1.35V and 1.75V. I would pair tubes B and C, and A and D, even though the pairing points differ from the "benchmark" 1.56V.

I hope that makes sense.

Those of you who have read my past rants on tube matching know that the only way you can truly balance the output tube current is to put a balancing pot in the bias network.

[dcriner]

Thanks, Eric. That all makes sense to me.

However, I'm thinking that with just one tube plugged it for test, I should initially adjust the bias pot to give 1.56/2 = 0.78V. The 1.56V corresponds to the combined cathode current of both tubes on one channel, correct?

Doug Criner

[EWBrown]

With only one tube, 1.56V equates to 100 mA, which is far in excess for a single EL34 / 6CA7. 0.78V would be correct. for 50 mA.

/ed B in NH

[erichayes]

Hi All,

You're correct, of course. Brain fart. I like to run my output tubes heavily toward class A, but 100 mA on a single EL34 is a wee bit excessive.

[EWBrown]

I've used an LM317 current regulator set for 35 mA, (actually two tubes set up this way), for matching and measuring EL84s / 6BQ5s. Bypassed from catode to ground with a 470uF 35V cap. For matching, I measure the voltage drop from cathode to ground, across the regulators.

This is part of a mono "half an ST35" that I use for testing purposes. Tube matching developed later on, and comes in real handy. The "generic" Russian 6P14Ps run anywhere between 10V and 16V @ 35 mA, I aim for 12 to 14V for my matched sets, and set aside the low and high ones for experimental stuff.

This may not work well for EL34s, as the voltage may exceed the LM317's
input voltage rating, maybe an LM7824 24V regulator with 480 ohms, 5Watt loading (for 50 mA) on the output would work in this application.

Just a SWAG....

/ed B in NH

[mesherm]

The TL783 is a 3 terminal regulator with 125 volt maximum rating good to 700ma.

[dcriner]

OK, I ran the cathode current matching test of my four JJ EL34s, using one socket of my ST70:

TUBE# BASE TEST -10% TEST +10% TEST
...1..........0.78V...........0.70V...........0.86V
...2..........0.75.............0.66.............0.84
...3..........0.73.............0.66.............0.83
...4..........0.67.............0.58.............0.76

(Pardon the dots -- I couldn't figure any other way to get the tabular columns lined up.)

From the above data, I would say that Tubes 1 and 2 are pretty close, and also the hotest tubes. So they should be paired together on the same push-pull channel.

That leaves Tubes 3 and 4. Tube 4 seems exceptionally low. I wonder if I should maybe buy another JJ and hope it lies closer to the others?

My voltage measurements were quite repeatable when re-running the test (within 0.01V), so I'm confident that the above data are correct.

I've got two other quads of EL34s that aren't being used (not JJs). Those extra quads were not purchased as matched. (My JJs were purchased as matched Yellow_Light_Colorz_PDT_21 ) I also have a quad of =C= that are presently being used in another amp; I may check them for matching in my ST70 just for grins.

Comments?

[mesherm]

A vendor stating that the tubes are "matched" is sort of like stating that a product is "natural". It really doesn't mean anything unless it is further quantified. One person's definition may not be another's.
I have heard from other people who have purchased "matched" pairs and found them to have up to 15 ma difference when installed.
One solution is to modify the bias circuit to adjust the tubes individually.
The other solution is to learn by trial and error which vendors to buy "matched" tubes from.

[TomMcNally]

My 2 cents worth ...

If the vendors are matching tubes with a tube tester - keep in
mind that the tubes are being worked VERY lightly, probably
at a fairly low plate voltage, and hardly any current. The tubes
will certainly act differently under load, as you've discovered.

Shannon's designs let you tweak the bias for an exact match,
so that's the way to go. The real bias tweak freak will do it
every day, others like me, will do it when they put new tubes
in after repairing the amp after a tube does some pyrotechnics.

... tom

[Blackburn Audio]

Toms right, we use a Hickok TV7 for alot of stuff at the shop and it doesn't hit the tubes with near as much voltage as in a real amplifier, most testers don't. The matching is really only true if they can be matched under the conditions of the amplifier the customer wishs to put them in. Usually though the tubetesters will get close enough to not cause any problems. I am working on a design for a Homebuilt Tube Tester we can use that will allow tubes to be matched under real amplifier conditions but its hard to find time to get any extra work done around the shop as we are always so busy. Hopefully soon though, gotta dig around the shop for parts and see what I come up with.


-Matt at Triodeelectronics.com