Simple Circlotron with power Mosfets

[moschfet]

Hi guys,
I want to show you this simple Circlotron circuit. In the simulation with LT-Spice it works very well. The signal will be taken off from this circuit at the drain of the MOSFET. It is therefore a flawless transconductance amplifier.
This is the circuit diagram:

This is the FFT simulation:

These are the frequency responses, open loop an closed loop.


Ready to play

[moschfet]

I have simulated a little bit.
And I found out that the current availability is extremely important in this topology. With the 2SK1058 the distortions increase sharply at higher powers. The IRFP240 provide significantly better results.
Now I know why Thorenz use in their Circlotron such strong mosfets.

Now I'll build this circuit to test:

[Lazy Cat]

Hi moschfet

I like your circlotron sch, awaiting for real circuit performance.

If possible avoid input opamp, better use double j-fet unity gain buffer, IMHO it is much more transparent than OPA604.

[Osvaldo de Banfield]

This design is interesting. But, what about security in case of a MOSFET shortcicuit failure? I work with electronic and almost all times MOSFET and transistor fail shortcircuited. Keep in mind some fusing in the load???

[moschfet]

If possible avoid input opamp, better use double j-fet unity gain buffer, IMHO it is much more transparent than OPA604.

Perhaps, but as a prototype, it is easier and I get a good distance DC.
1kohm is very low, can a JFET buffer drive it with low distortion?

Keep in mind some fusing in the load???

Yes, but a simple fuse enough for now.

[Joe Berry]

Hi moschfet,

Nice to see that driver stage connection getting some use. It works quite well in practice; I've had similar designs running for some time (see US patent 6747513). You can easily drive it from a single differential voltage gain stage, if you like.

Also, as you probably know, the idle currents in the IRFP240 output MOSFETs will tend to increase as they warm up over time. If you provide compensation for that, you may not need source resistors in the output stage.

Best of luck with your design!

Joe

[moschfet]

Hi Joe,

thanks you I will thermally couple the current source at Q10 with the Mosfets.
There I will also set the bias.

I've just changed the OPA604 against a JFET buffer. The simulation also seems to work with very little distortion...but with higher DC-offset

[alex mm]

..... My contribution in this thread ...
Alex.

[moschfet]

Hi Alex,
very nice.
I have seen your layouts and I like them very much, but you are too fast.
I have changed some details. I think the Mosfets must be thermally coupled with the current source Q10. So I plan to put Q11 on the heatsink. R17 is a pot to set the bias. I estimate, the circuit at the bias must be tuned in reality.

[Joe Berry]

The DC offset problem noted earlier can be addressed by adding a current path from each output terminal to the -50V rail, to recapture the current supplied by Q10 from the +50V rail. Resistors will work, but an NPN current sink can be made to provide thermal tracking against the Q10 current source (though such precision may not be needed), and will better isolate the outputs from any noise on the -50V supply rail.

Also, speaking of 50V supply rails, I notice that the front end does not need nearly this much voltage to operate. It should work fine on 25V rails, or perhaps less.