• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Beam power OTL - shifting bias levels

Good day,

Some months ago, I finished an OTL based on 17KV6 valves, which puts out 10W into 72R. It is based on Futtermann, +-150V Power supply and auto bias with 220R and circa 90mA idle per valve. https://www.diyaudio.com/community/threads/beam-power-pentode-otl.391888/

I also have a stash of 22JF6 which I want to use for a more powerful AB OTL amp to drive a 50R load (9x ~6R woofers in series). I want more power and less heat in the amplifier, so this time I decided to use fixed bias on the output tubes, but hat do double on the number of output valves, to get close to 50W into 50R.

For 50W into 50R, I need 1A Rms, or 1.4A peak, which is 0.7A per valve. Furthermore, developing 50W over 50R requires 50VRMS at the output, or 140Vpp, or 70Vpeak from each valve. Using +-155V power supply (and circa 150V at G2), it means that each valve operates between 150V and ~90mA at idle, and 80V and 700mA at peak. The blue trace connects there two points, looks doable (and with that I mean the worst case for the valve, where the G1 is at 0V (same potential as cathode, so limit of class 1) and 150V at G2, the plate is able to go down to 80V and pass 700mA).

I built the amp, much like its smaller brother. I use a 500VA transformer with 2x 115V secondaries, that go into a diode bridge and CRCRC (1200uF per C, 3R per resistor) for the positive and negative rail. Output is about +-160VDC.

For the fixed bias I, for the moment, made three separate power supplies, A, B and C. I have a single transformer with 3x 27V AC secondaries, each fed into a bridge and filtered with 2x470uF, 20k and 390k bleeder. Close to the valve and the 130k grid leak I have 50uF local decoupling. These three supplies are completely independent from each other. Supply A is connected to the upper cathode of the 1st channel, Supply B is connected to B-, and provides bias for both channels, supply C is connected to the upper cathode of the 2nd channel. I feed the primary of this transformer with a VARIAC, to allow bias adjustment.

I built the amp (both channels) and let it run at idle, with al bias values at around -29V (in relation to the respective cathode). I apply 1kHz sinewave to the 1st channel and it starts clipping at circa 12W (measured 25V RMS into 50R). That means that each valve is only accomplishing 0.35A peak…

I measure around and finally put my DVMs across the bias supply A-A1, B-B1, and C-C1. All bias values where at -29V at idle, but when I start increasing the level, the value of bias A-A1 starts going up, at the 12W it is at -35V. The value at B-B1 goes to about -30V, the value at C-C1 (which has no signal applied, so still idle) stays at -29V. Why does that happen?

(20k and 390k filtering on the bias is a bit on the high side, I could well use 1k and 20k bleeder, current consumption of the bias would still be below 2mA.)

Looking at the scope it is clear that the top of the sine is clipped, implying that the top tube can’t pass more current, which is understandable at it is biased at -35V, while the lower is at -30V, with both getting the same signal from the concertina?

The driver (using a E180F) is fed from a separate 300V DC supply, the heaters are all fed from AC. At idle and up to 10W the amp is working very fine (I already got rid of oscillations, a different story…), only when requesting more power, it is somehow, I think, current constrained and altering the bias supply value.

I also thought about the resistance in series with G2 causing too high a drop, limiting voltage at G2, which is essential to pass more current – I will still measure that!

Originally, I had 10R resistors in each cathode to help equalize current over both valves. With peaks of 0.7A, these would develop 7V local feedback, so I (temporarily) bypassed them, but did not make a difference for the output power.

Many thanks for all of your comments and ideas!


  • 22JF6 loadline.png
    22JF6 loadline.png
    117.9 KB · Views: 108
  • 22JF6.jpg
    311 KB · Views: 120
I have been looking at the bias voltage sources again, and I think the combination of the 20k filter resistor and 130k grid leak may be adding, as a voltage divider, some of the 1kHz signal to the bias circuit, increasing its value. I will reduce the 20k to 1k and see what happens.
I worked on the three bias supplies, replacing the 20k and 390k with 1k and 33k respectively. This stopped the "additional" charging of the bias supplies - will have to look if this worsened the 100Hz hum now.

The top of the sine was still flattened, until I reached out for the voltage control dial on the 500V adjustable PS, and by turning it CW, the flat landscape slowly turned into a nice mountain.

I also included 2x18V windings in series with the 2x 115V, increasing the output stage supply to +-175V. I played around, but could not really get to the 50VRMS (50W), even with the driver to 500V, so I think I was reaching the max current of the tubes. I also don't really need that power level and still thinking about limiting heat, so I went back to the +-155V supply, and 450V for the driver stage (this is still a lot of heat...), reaching about 32W (40VRMS on the 50R load) at circa 0.9% THD. Increasing the level further clips both side of the sine equally, which to me is an indication of current limitation.

The trick with the FP230-50 as dual choke worked really well, the voltage over the decoupling cap is stable at max output.
I had this amp on the side for a while, as I needed to make an appropriate supply for the driver B+ and for the three independent bias supplies. I ended up using a VPT230-110 wired backwards (230V to the series connected secondaries), giving about 200V AC at the primary (unloaded). This I feed into a voltage doubler, and filter with a LC stage for circa 480V DC for the drivers. For the three bias supplies I wound 3x additional windings (circa 7V each) on one of the toroidals, feeding this into a voltage quadrupler and doing some filtering on that for a clean -33VDC for biasing.

The tests above I did as in the schematic, with the 50 ohms load from the output to ground. There was always some DC offset there (+-1.5VDC), but my idea was to use an output capacitor anyway (I want them to play from about 40Hz onwards), so I was not so worried about it and went for the 3x fixed negative supplies, knowing I could not trim them for 0V at the output. Once I built and installed the supplies described above and turned it on again (using Variac to power the transformer for the output stage), there was motorboating. I changed some things, went back to the regulated 450V PS for the driver stage, but nothing helped. Then I realized I had also put the 45uF DC blocking caps in there, which at the 1-2Hz motorboating frequency present quite a high impedance. Here https://www.diyaudio.com/community/threads/otl-amp-help.45795/ I read about similar problems with too high loads on a similar amp, so I took out the caps and kept the 50R load, perfect working and got to the same values as in the previous post.

Now I have to think what I do. Try if e.g. 1k from output to ground is enough to tame the motorboating? Make the bias supplies adjustable, trim for 0V at the output, and use the speakers without DC blocking cap. Lucky there is another amp playing nice music in the meantime :)


  • 2024-01-26 08.51.21.jpg
    2024-01-26 08.51.21.jpg
    440.3 KB · Views: 34
  • 2024-01-26 08.52.07.jpg
    2024-01-26 08.52.07.jpg
    359.1 KB · Views: 33
  • 2024-01-26 08.52.31.jpg
    2024-01-26 08.52.31.jpg
    503.4 KB · Views: 30