In https://www.diyaudio.com/forums/sol...d-mosfet-amplifier-wireless-world-1982-a.html I mentioned that the Hitachi circuit posted by Nigel reminds me of the BUZ-Amp from a Siemens Appnote 1982/83. Very similar amps were published in France and in a German electronic magazine. Here is the simplified schematic: "Siemens-BUZ-Amp.pdf".
What I like about this is the simple 2-stage design with only 2 LTPS, and no VAS. It needs no compensation apart from 2.2pF over the feedback resistor since the dominant pole is the output mosfet's input capacitance. Since I once ordered a small batch of boards and had some left, I wanted to try a complementary output. This only needed a current mirror to invert the drive for the lower P-channel FET. With both source pins towards the output now, I re-arranged the 1k resistors to include the 0R22 bias resistors. They are there anyway for the U/I limiter (not shown for clarity, neither thermal compensation, Zobel, etc.). What came out is seen under “BUZ-Amp-N+P-MOS.pdf”.
Why it resembles the Hitachi circuit you can see in "BUZ-Amp-into-Maplin(Hitachi).pdf".
It has the advantage of a push-pull drive to the output FETs, and the bias is adjusted there (since it has no current sources, only resistor tails for the LTPs). As shown (also in the link at the top with the original schematic), it is only suited for lateral FETs (V-FETs = runaway bias).
It occurred to me that something originally designed to drive two N-Channels might also make a simple circlotron: "BUZ-Amp-into-Circlotron.pdf". The big question was what to do with the feedback? Taking it from one output node only like in the picture does work somehow, but it does not look right to me.
A way around was to take (negative) feedback to the input as well, and keep the positive feedback connected to the negative input of the first LTP like before: “BUZ-Amp-into-Circlotron_2xFB.pdf”. This actually worked, sounded good and with a pot at one side of the LTP the output offset could be adjusted too.
This was one more example to make a circlotron with only one (ground referenced) input. Most circuits you see posted have either two inputs, or opamp front ends.
Next idea is to make a symmetrical input, with IN- feeding into the bottom of C3 (instead of going to GND). This would make it fully symmetric. But will it work? What do you think?
What I like about this is the simple 2-stage design with only 2 LTPS, and no VAS. It needs no compensation apart from 2.2pF over the feedback resistor since the dominant pole is the output mosfet's input capacitance. Since I once ordered a small batch of boards and had some left, I wanted to try a complementary output. This only needed a current mirror to invert the drive for the lower P-channel FET. With both source pins towards the output now, I re-arranged the 1k resistors to include the 0R22 bias resistors. They are there anyway for the U/I limiter (not shown for clarity, neither thermal compensation, Zobel, etc.). What came out is seen under “BUZ-Amp-N+P-MOS.pdf”.
Why it resembles the Hitachi circuit you can see in "BUZ-Amp-into-Maplin(Hitachi).pdf".
It has the advantage of a push-pull drive to the output FETs, and the bias is adjusted there (since it has no current sources, only resistor tails for the LTPs). As shown (also in the link at the top with the original schematic), it is only suited for lateral FETs (V-FETs = runaway bias).
It occurred to me that something originally designed to drive two N-Channels might also make a simple circlotron: "BUZ-Amp-into-Circlotron.pdf". The big question was what to do with the feedback? Taking it from one output node only like in the picture does work somehow, but it does not look right to me.
A way around was to take (negative) feedback to the input as well, and keep the positive feedback connected to the negative input of the first LTP like before: “BUZ-Amp-into-Circlotron_2xFB.pdf”. This actually worked, sounded good and with a pot at one side of the LTP the output offset could be adjusted too.
This was one more example to make a circlotron with only one (ground referenced) input. Most circuits you see posted have either two inputs, or opamp front ends.
Next idea is to make a symmetrical input, with IN- feeding into the bottom of C3 (instead of going to GND). This would make it fully symmetric. But will it work? What do you think?