Tube Hybrid Shunt Regulator ... will this even work?

[dsavitsk]

I am just playing around with ideas for a HV shunt reg. I am not really sure that it will work at all. Any suggestion? Changes to make it work? Depending upon necessary current, a glow tube could probably be substituted. In fact, a zener is probably just as good, other than the lack of glowing. I'm also not sure what P channel fet to use for HV.

Edit: new pic added

[ikoflexer]

Originally posted by dsavitsk
[B I'm also not sure what P channel fet to use for HV.
[/B]

HV p-channel mosfets aren't that many to choose from. salas has used the irfp9240 in his HV shunt regulator. However, the safe operating area is quite limiting even with that device. So, depending on what you intend to use this for, you may get away with one mosfet, or you may have to use two in series. An alternative could also be to twist your circuit to use an n-channel mosfet for shunting. Or, if you feel particularly brave, use an igbt. At some point I was really considering this.

[sawreyrw]

You need to add some kind of reference device to your circuit. What you have won't work. Replace the tube with a resistor (to set the proper bias current for the zeners), and replace the diodes above it and the LND150 with a string of zeners.

[wrenchone]

Here's a couple of simple shunts to start with that will work. Output voltage is (VthQ1 + Vz)(1 + (R2/R1)). If you choos a high enough voltage zener with a positive tempco, it will help to balance out the negative tempco of the Mosfet threshold voltage.

[dsavitsk]

Originally posted by sawreyrw
You need to add some kind of reference device to your circuit.

Not sure what you mean. The mosfet is a P channel. Idss of the lnd150 is ~1mA and it will drop at least 1V minimum. In my head, the two LEDs are blue, so there is 6 to 7V. So, that means minimum 6V of bias between the gate and the source so the mosfet should not cut off below a couple of amps.

With a CCS load at 1mA, the triode will have 150V on the plate with a -1V bias. A 150V zener would work as well. Wouldn't this put the source of the P mosfet at 150V + Vgs and shunt extra current through the mosfet? Maybe it doesn't work, but your explanation of why this might be was not very helpful.

[wrenchone]

There's nothing about the circuit shown that really makes the output want to be constant voltage. You could probably tweak the pot to get a desired output voltage, but it would wander as soon as the output load varied, not a desirable characteristic for a voltage regulator. The simple circuit I showed gives an output impedance of around 75 ohms. Attached is a circuit almost as simple that did not change at all in output voltage when the shunt current was varied from 10mA to 50 mA. I used an IRF 9240 'cause that's what PSpice had in the library. Fairchild makes a couple of 500V P-Channel FETs (available from Mouser) that would serve well instead. You could tweak the circuit somewhat to substitute a tube for the N-channel mosfet, but the performance probably wouldn't be as good unless you used a much higher transconductance tube than the original 12AX7.

[wrenchone]

I meant IRF9620 for the shunt FET. I wrote the preceding post at 3:30 AM or thereabouts...

[sawreyrw]

dsavitsk,

I overlooked the fact that the MOSFET is a P channel, so my previous post is nonsense. Now that I know whats going on, here's what I recommend. Replace the tube with a zener of whatever value gives you the desired output voltage. If it's 150 volts, the bias current of the LND150 is adequate. The LEDs aren't doing anything, so you can delete them. Sorry about the mix up.

Rick

[dsavitsk]

I figured that was the issue. So, this sort of works but the impedance of the 12ax7 is too high? How about with a big bypass cap?

It really is that I am looking for a replacement for a CCS + 0D3 circuit with a lower Zout, and I want to keep the 0D3 in place. The adjustableness of a triode seemed clever, but I guess a bad idea. The overall circuit is from the TL431 datasheet but modified for higher voltage and with the addition of CCSes.