Eliminating Push-Pull Crossover Distortion

Discussion in 'Technical' started by W5HRO, Sep 29, 2016.

  1. W5HRO

    W5HRO Administrator

    Does anyone have any good tricks to help reduce or eliminate crossover distortion like in your class B tube driver and/or modulator other than biasing the tubes more positive or using typical negative feedback? You see a lot of articles on the web for doing such using pre-bias, diodes, and etc., but using transistors and their bases always have a positive voltage applied anyway. With tubes you have negative bias voltages at the grids so it’s not positive. You can run the tubes less negative so they are not completely in cutoff, but that has it's disadvantages. I have seen some designs though where they used added transistor circuits in series with the cathodes to drive them and that's sort of what I'm curious about.

    [doublepost=1475156221,1474905115][/doublepost]Here is a patent for eliminating it, but its all solid state.

    US3564445-0.png US3564445-1.png

    Then here are two amps doing a similar thing with tube finals. They are cathode driven. What I’m wondering is if there is a way to do a similar thing, but using all tubes with tubes in series with the final cathodes or in a buffer configuration.

    1.jpg 2.jpg
  2. WD5JKO

    WD5JKO Member

    A simple option is to use a tube with a transfer curve appropriate for zero bias class B operation. The 838 and 805 are two possibilities. These are better than 811 / 572B types. These will keep a good load on the driver over the entire waveform. If you have to add some - bias on the tubes, then the load on the driver abruptly changes as a given tube starts drawing grid current. Here you need a low impedance driver (6B4's. 6A3's, etc.), and perhaps grid swamping of the driven stage. A little NFB helps too, but don't get carried away with that stuff unless the IRON is quality stuff good for at least an Octave above and below the audio range used.

    I had great luck with a quad of 808's at 2500V plate, and around -50v bias. The 808 is a very interesting tube. I used a pair of 6B4's, a Globe King 500 audio interstage transformer (brought out the CT. from within the bell), and P-P NFB back to the 12BH7 cathodes. The 12BH7's drove the grids of the 6B4's. I got a very clean 600 watts from that modulator.

  3. W5HRO

    W5HRO Administrator

    Hi Jim,

    I'll tell you why I'm asking these questions and doing a little research. When I go to build up the BA-1K transmitter which uses the two 4-400A's in push-pull which take "0" watts of drive power I want to drive them with two small tubes in push-pull like maybe 6C4's. Just really only need the voltage swing. Well, those two small tubes in push-pull will be amplifying the phase rotator board that I bought if I can get all of that to work correctly.

    Basically the D104 to the FET buffer installed at the input of the phase rotator board, and then the phase rotator board to the input of the two small tubes in push-pull. Well, I could use two small transistors and cathode drive those two small tubes and is what I may actually do. That would make the most sense I think. I don’t think it would be a good idea to cathode drive the 4-400A's though. It could probably be done if using two separate filament transformers and then drive the center-taps, but I'm not sure how much power the 4-400A’s would produce that way. At the same time it could possibly work. I'd like to not use any coupling transformers at all. The only issue is going to be coupling the two-small tubes to the 4-400A grids. That's the section or stage I'm wondering about.

    Anyway, just thinking over all of the options and what might work and what might not in an attempt to eliminate any crossover distortion that might exist and obtain the best fidelity.
  4. WD5JKO

    WD5JKO Member


    There are a lot of ideas from this approach:


    Keep in mind that although the 4-400 will be grid driven in Ab1, there will be a lot of gain to magnify the plate to g1 capacitance. You still need a beefy driver since the grid swing will be influenced by the tube Miller capacitance. The circuit I linked to drove something like a 1/2 dozed 6550's in P-P to make 300 watts audio with a lot of headroom.

  5. W5HRO

    W5HRO Administrator

    Well, the BA-1K audio drivers were MJE2801 transistors and those were 90W transistors at 10A :eek: That high current low voltage crap has to go!

    It will take a little power to drive the 4-400A grids, but in the end the result is still "0" watts in class AB1. They just need enough power to keep the voltage swing up to where it needs to be. As long as I get 130VAC from each grid to ground it should be ok. A 300V filtered supply transformer with a center tap connected to small push-pull tubes should work fine. If need be I could always go with a little bigger tubes.


    Below is the simulation I have been playing around with. All I need is the extra op-amp to invert the output of the phase board to drive the other small tube in the push-pull stage and maybe increase the gain of that last stage on the phase board by increasing that one 10K resistor. The collectors of the two transistors will just connect to the cathodes of that push-pull stage and will be linear. In the simulation I used a couple of resistors for the output to simulate a transformer, but those will be tubes. I can eliminate all of the crossover distortion in that push-pull driver stage. It's just the push-pull 4-400A stage I need to think about.


    The above simulation only uses half of the phase board because the circuit was going to be too big too manage on the simulator so I only used 4 poles just for a test reference. The board actually has 8 poles. I may need an extra amp between the output and the pull-pull stage transistors. What I can also do is add a non-inverting buffer on the other side to match the added inverting one since the added chip will have at least two op-amps in it. Then one more chip for added drive to the transistors if necessary. Either that or make one of the op-amps an amplifier then the other one the inverting buffer so it would only take one chip total.

    [doublepost=1475244305,1475206434][/doublepost]They only ran the audio amplifier and those transistors at +40VDC. They were just barley tickling the 4-400A's in AB1. The BK-1 runs at 2800V on the plates so that's an easy 1000 watts of modulator power and into 1200W finals :eek:

    I'm going to need to keep the drive turned down anyway otherwise almost 200% modulation if I run them up to the full peak AC grid voltages. Lucklly there is no DC load on the secondary of the modulation transformer and it has that big 50H audio reactor instead. I wIll still add the improved negative peak limiter though.

    I think T2, the audio driver transformer is a step-up transformer so using two tubes instead of transistors will make the step-up ratio not as great if that is the case. The question is can I directly couple them or should I just try to find a transformer? That's why I was thinking of driving the 4-400A cathodes instead, but that would in fact take a lot of drive power. I might be able to get away with direct coupling them if I use two medium mu triodes and do it right. Low mu triodes would work best, but it doesn't really need 6A3's, ect. I was thinking of 6C4's or maybe two dual tridoes with each one wired up in parallel. Maybe like two 6N7's, but 6C4's may work better down at 150VDC. At 150V I might be able to just ground the grids using 6C4's when driving thier cathodes. You can do that at 100V and it might still work at 150V or at 125V.

    [doublepost=1475254255][/doublepost]And here is the stock audio circuit designed to take a 600-ohm line input to the 4-400A grids. They only used a 40V supply. The MJE2801's went to the primary of the T2 driver trasnformer.

    [doublepost=1475277600][/doublepost]If I use a driver transformer then something like the below would probably work. If I need more drive power then I could just parallel in two more 6C4's, etc. I like the idea of grounding the 6C4 grids so it's basically a grounded-grid push-pull linear amp and those tubes are tiny. Just parallel more until it's there. And I guess because the 4-400A's are in AB1 crossover distortion may not be an issue.

    [doublepost=1475417895][/doublepost]I was thinking, something like a 6AS7 would work better. Those are twin much higher power low-mu triodes. I would just need to use big enough transistors to drive it's cathodes. I wanted smaller size tubes, but I may have one or two of them in one of my spare tube boxes. The 6AS7 will work down at 100V and has specs for 135V which is perfect. Each plate has a resistance of only 280-ohms with 135V at 125mA.
    [doublepost=1475423362][/doublepost]This topic as is pertains to the BA-1K transmitter will be continued here: McMartin BA-1K Transmitter Project

    Attached Files: