Comparing Eight NOS 6AK6 Vacuum Tubes

Recently I've been designing an automated curve tracer for vacuum tubes. When it's finished, it should be able to measure the characteristics of almost any tube, producing graphs similar to those found in the datasheets published by the manufacturers. It's a very interesting project with some tough challenges, involving high-voltage circuit design, digital logic design, and a lot of software. It should be useful for many things, including:

I have a prototype of the curve tracer on my workbench, and it's working pretty well now. It's a bit ugly at the moment, but I'll build a nice version of it once the bugs are worked out.

I decided to use the curve tracer to compare a bunch of new-old-stock tubes. I wanted to find out how the actual tubes compare to the published datasheet, and how much variation there is between different tubes of the same type. The tubes I decided to compare were a set of eight NOS 6AK6 tubes. The 6AK6 is a power pentode that was used as the audio output stage in small AM radios. According to the datasheet, it can put out a thundering 1.1 watts of audio power with 10% total harmonic distortion. You wouldn't want to build a hi-fi around this tube, but I did use a pair of them to build a very nice bedroom guitar amp. I chose to analyze the 6AK6 primarily because I have a lot of them.

Click on the image below to see the plate curves from the original 6AK6 datasheet, followed by the measured curves for eight different specimens:

You might notice that the top curve of each plot is a bit ragged looking. That's an artifact of the curve tracer, not of the tube. The curve tracer's current measurement circuitry has two ranges. The low range covers from 0 mA to about 40 mA. Above 40 mA, the high range kicks in, covering currents up to about 500 mA. The top curve of the 6AK6 plots happens to be just barely into the high range of currents. The jaggedness you see in the curve is caused by the fact that the high range current circuit can only measure down to a resolution of about a half milliamp. On the scale of these particular plots, that's large enough to be visible.


I noticed several things about the measured plate characteristics. First, all of the GE tubes are fairly similar to one another. And they are all a bit "hotter" than the datasheet—which, by the way, was published by RCA.

The Sylvania tube is noticeably different from the GE tubes. Its slope at low voltages (below the knee) is less steep than the tubes from the other manufacturers.

The two Westinghouse tubes have curves whose shapes are similar to the GE tubes. One of the tubes is substantially "hotter" than the other.

I don't yet have software in place to plot the transconductance of the tubes. It will be interesting to see how the tubes compare in that respect.

Copyright © 2013 John D. Polstra. All rights reserved.