Inspired by Geek's quote on his tubes forum, *"Let's hear some tube-tales of how you got 500 watts out of that 6L6..."*

A.K.A. - Vacuum Tube Drag Racing!

- As much power as possible! No holds barred, damn the torpedoes, melt as many plates as you wish in trying.
- Must be real measurable
**AC**power. You can use a pure resistor load, but just counting its dissipation includes a useless DC component; for convienience, you might want to put the load resistor across a choke or transformer instead. A "true" wattmeter of some sort may be a handy tool. - Tests can take place at any frequency, though preferably audio range. If you want to play a 1kW note at 1kHz, that's fine by me. RF is a bit figety but if the results are true I see no reason to deny them. If you find a frequency range has a power output peak, by all means send data at the peak response!
- Continuous power would be neat, but impossible to qualify - does it have to run full power output for one minute? Hour? 500 hours? So I'll just go by peak readings (NOT peak instantaneous power!). It should run for at least a few cycles, and probably a second or longer if you don't have a storage 'scope. A tone burst generator would be another convienient option.
- I don't care about distortion. It might make for messy readings, but if you get a clean squarewave, that's no problem. Again, points for a true power meter.
- Since it would be downright easy for a 4-1000A
__OTL__to beat a 6C4, the record will be based on power output vs. rated plate dissipation. (Power inputs will be recorded but not considered.) This isn't entirely fair because some tubes are rated for more capacity than others which are almost equivalent, so you'll have to keep a reference handy if you are unfamiliar with the tube in question. - Topology doesn't matter, you can go SE or PP. I am only concerned with the amount of power each tube is producing: say you get 300W from a PPP quad; the number recorded will be 300/4 = 75W per tube. I will of course record topology for posterity.
- I'll sort the records by plate dissipation and power output, so you can see who got the most power from the least tube as well as the absolute biggest amplifier on record (hi Nick).
- E-mail me your most impressive results. It must include an attached schematic of the output stage and recorded data on voltages, currents and power output. An oscillograph (oscilloscope capture/photograph) of any waveforms, particularly output, will be helpful.

In tune to Gregg's quote, I've drawn up some load lines for the 6L6. It will be class D (in other words square wave, but would be class B if driven linearly) and use an MOT secondary as a choke load. I'm curious to see what the flyback waveform will look like on that...

So here's what I did. First of all, I know I'm going to be well outside the ratings, so I'm going to have to extrapolate the graphs to get the operating points I need to find a good starting point. First off, screen voltage and plate current: the graph goes to 400V_{g2}, all of 320mA at saturation (about 100V_{p}). Remember that this is the zero-bias curve, i.e., maximum plate current -- if I were doing a class 1 amplifier. I will of course use positive grid voltage (and grid current) to help melt this thing.

If I went with that and stayed within the maximum rated plate voltage, I could switch a maximum of 500V * 0.32A = 160W if the plate voltage dropped to zero, which can't happen. (At 100V_{sat}, it only puts 400V across the load and thus only 128W in the load*.)

**128W if the load were a resistor and the instantaneous flow of power were the goal. But it isn't; I want AC power. Best case then would be a square wave of 50% duty cycle between zero and maximum current, which is 64W output. Not bad for a single 6L6, which here only dissipates 16W plate, but we can do better!*

The blue line runs constant plate voltage to find transconductance of the screen. (Voltage is arbitrary and matters little due to the high plate resistance.) The lower dot is at nearly 300mA while the top one is at a bit over 350mA. The difference is thus 55mA, for a difference of 50V_{g2}, therefore Gm = 0.055A / 50V = 1100μmhos (not bad for a screen). Let's say 1200, since it rises a bit with current. Now, I want to bump it up by 100V to 500V, so plate current will rise 0.0012A/V * 100V = 120mA for a total of 470 or 480mA.

That said, we can move on to graphing the operating point. We will transfer the above point to this graph (which shows curves for grid voltages instead). 480mA 300V lands right between the +8 and +12V_{g1} curves, so we can call it +10V -- __ IF__ screen were at 400V as specified on this graph! Now this is where it gets complicated -- we will redefine this graph for a new screen voltage. The blue curve represents our new zero-bias (V

But that's not the worst we can do -- there's plenty of grid voltage to be applied here yet. The purple line is again vertical for determining Gm, this time of the grid. This time I'm doing three points, 10V apart each. The bottom is at 300V, 355mA; middle is the blue curve and dot (our 0V bias line at 500V_{g2}) at 490mA and the topmost at +10V is 640mA. The lower difference is 135mA/10V = 13,500μmhos and the upper is 150mA/10V = 15,000μmhos, so we can expect 17,500μmhos or so up to the next point at +20V_{g1}, which will thus bring 815mA at 300V. Our new saturation voltage will be close to 200V in this range (you'll notice from the graph that, inherent to the tube, no matter what the screen or grid voltages, the saturation voltage rises with plate current at a rate of around 200 ohms), so our power output is now Δ400V * 0.8A / 2 = 160W (into a load of 400V / 0.8A = 500 ohms at a dissipation of 200V * 0.8A / 2 = 80W).

So that's how I intend to find top operating points. I'm still in search of 500W from a single 6L6, probably not possible with a 6L6GC but one of its brothers with plate caps, the 807 or 6BG6 for example, are able to handle much more voltage and should be able to switch the 1kW easily. Anyone got a 1.2kV 500mA power supply?...

Web page maintained by Tim Williams. All rights reserved.