There is a project online that was posted a few years ago that I want to explore. Unfortunately the progress was never updated so I will have to design it myself from the information that was given. Unfortunately I can't for the life of me figure out how it was done.

The project stated here http://tubelab.com/articles/ideas/solid-state-opt/

enter image description here

My understanding of it is that it splices the signal amplitude at a constant frequency outside of the audio band through the transformer and then unsplices it at the other end.

It looks like an H bridge which would make sense in theory but I don't see how the mosfets won't just project their vgs voltage and signal onto the load completely negating whatever signal is present. I also don't see why the top mosfets are p channel wired in reverse. The more I think about the topology the more I don't understand. Can someone explain this to me?

Here is a pic of my spice file:

  • \$\begingroup\$ What do you mean "project their gate-source voltage"? Do you mean like a common drain amplifier? They're wired up the other way, like a common source amplifier. \$\endgroup\$
    – τεκ
    Dec 29 '17 at 16:42
  • \$\begingroup\$ the circuit is a high-voltage full bridge PWM modulated by audio at 70kHz switch rate with a step-down ferrite transformer from a ATX-PSU with a synchronous, lower voltage higher current Full bridge to rectify the audio down to DC. \$\endgroup\$ Dec 29 '17 at 17:06

You just have to treat the MOSFETs as switches. Whatever voltage exists between the power supply and the signal driver on the left is imposed across the primary of the transformer, first one way and then the other, just like any other H-bridge.

This creates an AM signal, in which the amplitude is controlled by the signal, and the square wave "carrier" is created by the MOSFET driver circuit.

On the right side, the switches (MOSFETs) are driven synchronously with the ones on the left (the exact synchronization mechanism is left as an exercise for the reader!), and this has the effect of full-wave rectifying the AM signal on the transformer secondary, which produces a replica of the waveform of the original analog driver signal. A little low-pass filtering will remove any remnants of the carrier frequency.

So, the FET driver circuit on each side must produce gate waveforms that switch the MOSFETs between fully off and fully on in the correct pattern, and the switching on both the "transmit" and "receive" sides must be synchronized.

  • \$\begingroup\$ I don't understand how the bottom mosfets on the left are not corrupting the load of the triode. In spice they just project the high frequencies at their gate through a low impedance, corrupting the triode signal. \$\endgroup\$
    – coinmaster
    Dec 29 '17 at 17:44
  • \$\begingroup\$ Show us your simulation. You're probably not driving the MOSFETs correctly -- the gate signals need to "float" relative to the source voltages. Also, you need to make sure that the transformer has the correct characteristics for the carrier frequency, not the original audio frequencies. \$\endgroup\$
    – Dave Tweed
    Dec 29 '17 at 17:55
  • \$\begingroup\$ Here is a pic of my spice file i.imgur.com/SzkulwX.png I'm not sure I get what you mean by floating the signal. \$\endgroup\$
    – coinmaster
    Dec 29 '17 at 17:58
  • \$\begingroup\$ I added the picture to your question, where it belongs. And it makes it clear that you are indeed driving the MOSFETs incorrectly. Try driving each gate with its own voltage source, with the other end of the source connected to the same transistor's source terminal. That's what "floating" means. Getting this correct with "real" components is what makes this project so complicated. One way to do it is to use a high-speed pulse transformer to drive each transistor. This would allow you to use N-channel devices throughout. \$\endgroup\$
    – Dave Tweed
    Dec 29 '17 at 18:07
  • \$\begingroup\$ Ohhhhhhh, so the entire driver circuit needs to be floating at a different potential than the amplifier, correct?. That's interesting. I've never thought of using it that way. \$\endgroup\$
    – coinmaster
    Dec 29 '17 at 18:19

The H-bridge switch would be set for precise and fixed 50% duty cycle, at a frequency well above the audio range (70 kHz is suggested).
chopped plate voltage
(chopped current waveform in green, audio in red).
Since the primary-side chopper, and the secondary-side chopper are driven from the same 70 kHz source, little 70 kHz filtering is necessary.

The MOSfet gate driver would be a complex affair, having to withstand a DC offset of the plate voltage of that Class-A audio tube. Nasty. Getting the chopping logic state into that primary-side MOSfet-driver is a bit tricky (your circuit doesn't show how). That 70 kHz signal might be opto-coupled. The secondary-side MOSfet driver would be far easier, being ground-referenced.
Both H-bridges are similar standard quads (two N-channel at the bottom, two P-channel at the top). A charge-pumped driver chip would be appropriate, but some of these are meant for all-N-channel MOSfets.

This circuit is a rather complex arrangement to avoid a monstrous iron audio transformer for deep-bass audio. But complexity might be worth the cost: high power chopping technology is well down the engineering curve.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.