Inverted Triodes - Investigating the 6AS7/6080 Family

Background

An inverted triode is one in which the function of the plate (anode) and control grid are reversed. That is, the signal is fed into the reverse bias plate acting as the grid, and the output is taken from the positively biased grid, acting as the anode. For more details see the original article: Inverted Tube Operation on these pages. In brief, an inverted tube provides a very low plate resistance and very low effective mu.

This report concentrates on the 6AS7 family of devices.

The 6AS7 Family

There are a few variations to this low mu triode: The 6AS7, 6080, 6080WA, 6080WC and "European" CV2984. There are additional variations, ones with 12 and 24 volt heaters, as well as "near" devices like the 5998. None of these are "characterized" for inverted operation; perhaps people didn't realize the use of a device with a mu of less than 1.

I'm in the process of evaluating the potential of using these to provide a reasonably powerful OTL SE amplifier; one with no feedback, no "iron" of any kind in the audio path, and directly coupling to the speakers. Such a device is possible using several 6AS7 tubes inverted and in parallel. This breaks the tradition of requiring a quasi push-pull circuit in OTL or resorting to various RF operations to make the iron required smaller.

I have characterized approximately 80 6AS7/6080 devices for inverted operation. Not overly suprising, the characteristics from device to device varied somewhat. The devices included current production Svetlana's, and various NOS devices; RCA, Sylvania, GE, Tung-Sol and some "England" stamped Sylvanias (not Great Britain) which are probably NOT Sylvania.

I found that the characteristics fall into three distinct characteristics. So far, every tube characterized falls into one of these three curve sets.

The Most Common Type

About 40% of the examined devices fell into this category (including the Svetlana current production), so I'm investigating the amplifier based on this type. Characterized by moderate mu (0.18) and reasonably low plate resistance (60 ohms), and a reasonably good cutoff characteristc. The "plate" (grid) of both sections paralleled appears to be capable of dissipating about 5 watts (2.5 watts per section) without overheating. Power levels over 8 watts appear to produce unstable characteristics, due to grid overheating.

Here's the "plate" characteristics of this type of 6AS7:

Note the "kink" in the zero bias line. That's real; it shows up in every tube of this "type". Notice the load line provides reasonable power (about 10 watts) with only slightly over 100V RMS drive in Class A1 (no -er- "grid" current). This implies a topology where the cathode is connected to a -33 volt supply (at 1.8 Amps), the speaker directly connected to the "plate" (grid), idling at 0 VDC, and a 56 ohm, 180 watt (!) resistor connected to a +100 volt supply for the plate load resistor.

This may be a little exotic, but proves the concept. The real world amplifier may use a slightly altered loadline!

The Second type of characteristic

There are other types of 6AS7 tube curves, though. One type seems to fit the 6080/CV2984. At least, most of the tubes that fell into this category were marked CV2994/6080, although a few 6080WA and 6AS7Gs fell into this category. This is a beautiful set of tube curves: sharp cutoff, great linearity, relatively constant plate resistance. The mu is a little higher (0.22) and the "plate" resistance is a little higher (80 ohms). These tubes likewise could handle about 5 watts for the combined power into both sections. Here's the tube curves:

There's something pleasing about these curves. If I had enough, I'd be tempted to use this type for the prototype test amplifier. Also note that there is no pronounced kink at zero bias either. 20% of my set of tubes fell into this category.

The third type

The third type came from all the 6AS7 family, but slightly predominated by the 6080WC devices. This is characterized by very low mu (0.1) and very low plate resistance (40-50 ohms). This type also has a "remote cutoff" characteristic, probably leading to relatively high distortion (but predominantly even order) as an amplifier. Here's the tube curves for this type:

Notice that the mu changes quite a bit and the cutoff is not very sharp. Almost the antithesis of the previous set of curves. There is also a quite prominant "kink" at zero bias. About 25% of my set of tubes fell into this category.

Well, which type do I have?

This is reasonably easy to determine. Apply -100 volts to the real plates (both in parallel) to act as the "grid bias". Apply 20 volts to the real grids (both in parallel) to act as the plate. Measure the current drawn from the 20 volt supply. If it's very low, either your tube is bad (I presume you eliminated this possibility by checking the candidate in normal mode) or its the "second type". If the current is 50-80 mA, it's the "first type". If the current is 150 to 200 mA, it's the third type.

Other members of the Family

I have a few 5998 tubes (higher forward mu; higher gm too). These provide an inverted mu of 0.08, and a plate resistance of 30 ohms. Despite requiring a few hundred volts drive, this would probably produce a nice amp, as the characteristics appear to be nice. This may be misleading, as I don't have a large enough sample size to tell if the characteristics are uniform or not.

And yet Other Types

I've tried 5687 (see first report). These all seem to have uniform and very nice characteristics. I've also tried 12B4 (mu 0.1, plate resistance 140 ohms, "plate" will take about 1.5 watts), 6B4 (mu 0.06: so low that you start to get arcing at the required "grid" biases), 7233 (very much like 6AS7 even considering it's one section per tube - you can only safely run about 2 watts into that "plate" before it gets too hot, otherwise, a good bet). I'm still looking for a good candidate in thoriated-tungsten.

Steve