Many tube sellers now use the Amplitrex AT1000 tube tester. This is a great piece of test gear. However, like any test equipment, it must be used correctly (manual mode with plenty of time for tubes to stabilize) and the results must be reported correctly for the data to be meaningful.

I am a tube buyer as well as a seller and I have bought many tubes from sellers who also use the AT1000 tester. When I get the tubes and run them on our AT1000 testers, I often find that the results I get vary dramatically from those reported by the seller. The most common cause for this is that the seller used auto bias mode to test the tubes, and we use fixed bias mode as our standard test mode.

The AT1000 tester’s two bias modes function very differently and can produce very different results. In fixed bias mode, the grid bias is fixed according to the specified value for each tube type, and the current draw (mA) and transconductance (Gm) results vary according to the strength of the tube. In auto bias mode, the grid bias is automatically adjusted according to the strength of the tube until the tube reaches the current draw (mA) specified for its tube type.

It’s worth taking the time to describe these two modes in more detail and take a look at some examples.

Fixed Bias Mode

In fixed bias mode, the AT1000 gives fixed voltages for the critical parameters much like a vintage tube tester. For example, the EL34/6CA7 tube specs we use are:
500v plate
450v screen
-38.5v grid bias

Under these conditions, the expected results for a new tube are:
50 mA current draw
7500 gm

The AT1000 also provides specs for a lower voltage, higher current test:
265v plate
250v screen
-13.5v grid bias

Under these conditions, the expected results for a new tube are:
100 mA current draw
12500 gm

You will notice that in both cases a tube that tests near 100% will be running at or near the tube’s maximum dissipation (25w). This is by design. Running a tube hard is the best way to ascertain its condition. In some cases, a tube will test quite a bit higher than expected. This can be harmful to the tube because the test is designed to run the tube at maximum dissipation. It is with these higher testing tubes that we use auto bias mode.*

Auto Bias Mode

In this mode, the same plate and screen voltages are used, but rather than use the “fixed” bias voltage, the bias value starts at double the expected value (i.e. -77v with the 500v EL34 settings above) and then steadily moves toward zero until the target plate current draw is reached. As the bias voltage becomes less negative (closer to zero) the current (mA) flowing through the tube increases. This is essentially the same process used to set the bias in fixed bias amplifiers.

Since the tester is adjusting the bias voltage to the tube’s strength, looking at the mA reading is now meaningless. Unless the tube is almost completely dead, the tester will be able to achieve a current reading very near the expected value. The Gm results must now be understood as how the tube would operate if the bias were adjusted for expected plate current. In auto bias mode the real question is, what bias voltage was required to get this tube to arrive at the expected current draw?

Sample tests for vintage Mullard EL34 tubes

Let’s look at some real world examples. I recently acquired a Dynaco ST-70 with a set of Mullard EL34 tubes in it. Two of the tubes looked pretty well used up. The flashing was a bit thin and had pin-holes in it. But the tubes still sounded good and biased up fine in the amp with some room left on the dial, so they were obviously not useless.

To illustrate the differences in results between different modes, I ran this same pair of tubes through the following tests: 500v fixed bias setting, 500v auto bias setting, 265v fixed bias setting, 265v auto bias setting, Triplett 3444, Hickok 539C. Here are the results:

500v in Fixed bias mode test results:
Tube 1: 20.4 mA (40% of expected value) and 4425 gm (59% expected value)
Tube 2: 20.0 mA (40% of expected value) and 4275 gm (57% expected value)

500v in Auto bias mode test results:
Tube 1: grid bias -33.5v (-38.5v expected) and 6975 gm (93% expected value)
Tube 2: grid bias -33.2v (-38.5v expected) and 6975 gm (93% expected value)

Here we can see that the AT1000 had to go about 5 volts positive to achieve the expected 50mA of plate current. Once the expected mA was achieved, the expected gm follows. The information about the grid voltage listed above is absolutely necessary to meaningfully report the scores for tubes tested in auto bias mode. Notice what happens if I simply report the scores from the auto bias testing in the given percentages from the tester’s digital display:

Tube 1: emissions 88% and gm 93%
Tube 2: emissions 87% and gm 93%

Now that looks a lot better doesn’t it? But of course that is also very misleading. Even worse is when only the gm number is given and it is not disclosed that the AT1000 was in auto bias mode. All the auto bias report can really tell you is that it took a 12-13% change in the grid voltage to achieve the expected current draw. And, as expected, the gm increased as the plate current increased.

Using the same two Mullard EL34 tubes from above, let’s see how they did at 265v.

265v in Fixed bias mode test results:
Tube 1: 60.2 mA (60% of expected value) and 10,000 gm (80% expected value)
Tube 2: 58.6 mA (58% of expected value) and 9,750 gm (78% expected value)

Here it is still clear that these are significantly used tubes, but they look much healthier at 265v than they do at 500v.**

265v Auto bias mode test results:
Tube 1: grid bias -9.9v (-13.5v expected) and 12500 gm (100%)
Tube 2: grid bias -9.8v (-13.5v expected) and 12,750 gm (102%)

Here we can see it took 27% and 28% changes in the grid bias to get the expected current draw. That is a lot of change in grid bias! But calculated as a percentage (73% and 72%) it doesn’t look so bad. And of course, having forced the tubes to operate at these conditions we get the expected gm scores as well.

Out of curiosity, I ran these same two tubes on our calibrated vintage tube testers.*** Here are the results:

Triplett 3444 test results:
Tube 1: 31.0 mA and 7,100 gm (3800 min. / 7600 average)
Tube 2: 30.5 mA and 7,000 gm (3800 min. / 7600 average)

The test data provided with this vintage tester did not give an expected mA reading for this tube type, but these readings would certainly lead one to believe these are usable, good testing tubes.

Hickok 539C test results:
Tube 1: 28 mA and 6250 gm (3800 min.)
Tube 2: 27 mA and 6250 gm (3800 min.)

Once again, no test data for expected mA, and only a minimum reading for gm. Again these tubes pass the test just fine on this vintage tester.

The good news is that both of these vintage testers agreed that these tubes were well matched to each other. The bad news is that they both significantly overestimated the condition of these tubes because the testers themselves were not able to properly load power tubes of this type. Recall that the AT1000 test at 265v (reasonably close to the plate/screen voltage of the 3444) looks for 100mA for tubes of this type at this voltage. Neither of these vintage testers is able to do anything near that.

Conclusion

So how do we rate these Mullard EL34 tubes? Depending on which tests you use, you might say anything from “tests as new” to “reject.” At TC Tubes, we would describe these tubes as “bargain” grade and we would disclose that at 500v they test in the questionable range (40-60%) for mA and gm, but that they do bias up fine in a Dynaco amplifier. They would be priced appropriately for their condition and would come with our usual 30 day return policy.

I would not recommend these tubes for a cathode bias amplifier that runs at plate voltages of 500v or more. They would probably do fine in an amplifier that has the ability to bias them up correctly. After all, they came out of a working Dynaco ST-70 and there’s no reason they shouldn’t continue to serve in that application for some time yet.

The main point I hope you’ll take away from these examples is that when interpreting results from an AT1000 tester, it is critical to know what mode and what voltages were used for the test. It is easy to give results “straight from the digital screen” that do not directly reflect the condition of the tube being described.

If auto bias mode is being used, the mA and gm test results are not particularly meaningful unless the grid voltage data is also provided (expected grid voltage vs. actual grid voltage used). Significant deviations from expected grid voltage are problematic.

It is my opinion that fixed bias mode should be the default mode for evaluating a tube's condition because it provides mA and gm scores at fixed operating voltages for plate, screen, and grid bias. As established above, the highest voltage test settings available will give the most revealing test results.

---------------------------------
Notes

* The auto bias mode is useful for testing tubes with mA readings that are higher than expected, since these tubes risk exceeding their maximum plate dissipation ratings when tested under fixed bias settings. Auto bias mode adjusts for this and creates a safer environment for the tube. Another good use of auto bias mode is for testing tubes with mA readings that are much lower than expected, to see if they can still be biased up to work satisfactorily in your application. In this case, auto bias mode will give you a good idea of how much bias adjustment is needed for the tubes to perform correctly in your circuit.

** Another important bit of information about test results from the AT1000 has to do with the voltages at which a tube is tested. Generally speaking, I have observed the following:

• For significantly used tubes, lower voltage tests yield stronger results than higher voltage tests.
• For new tubes, higher voltage tests yield stronger results than lower voltage tests.

To illustrate this, a new production SED EL34 that tested at 82% mA and 83% gm at 265v (fixed bias) tested at 106% mA and 85% gm at 500v (fixed bias). New tubes respond well to higher voltages.

*** These two vintage testers both run the tubes at relatively low voltages. Because they do not have regulated power supplies, those low voltages are not consistent. The line adjust feature is used to partially compensate for the load of the tube during the testing process. (The better vintage testers have such a line adjust feature.) I say “partially compensate” because the set-up voltages are still not regulated—they will be affected by the presence of the tube in the test circuit and the current flowing through it. As more current flows through the tube being tested, the affect on the set-up voltages increases. This is acknowledged in the calibration documentation for the 3444. When doing the line test calibration sequence the 3444 plate voltage is measured at 250v with no tube under test, and it is considered acceptable for that to drop as low as 220v under load (provided by a 4.2k resistor during calibration). Please note this is 220v on the plate with the tube to be tested in circuit and the line adjusted properly. A larger power tube like the EL34 is likely to load this tester down quite a bit. The 539C starts with an unloaded plate voltage of around 170v and goes down from there under load as well.