HV shunt regulator

[ikoflexer]

This is a high voltage shunt regulator that I built and have been using for a bit more than a week without a problem. This is only a variation of the same topology that I built many prototypes in the past. The aim was to get good line and load regulation (low output impedance), low(ish) noise, slow start up.

The example circuit below was built to power two channels of a #26 tube preamp. Has an output voltage of about 160V, assumes a load of about 10mA; the shunt mosfet passes about 50mA. The real circuit gets up to 160V in about 35 seconds. The line regulation and output impedance plots are only simulations. I have not measured the real values and don't expect to be the same as the simulated values, but I expect them to be very decent. I recommend a film output capacitor of higher value but not higher than 680nF. Best is to check for ripple at the output with a passive (resistor) load or the real load and use a value that shows no ripple/oscillation. This is a high performance regulator and it will oscillate if not implemented properly (correct value for C1, short distances between parts, as short as possible connection to the load!!!).

This version is recommended for low current but the same topology can be used for higher current as well, with some minor modifications: different CCS and shutn mosfets to be able to dissipate the heat.

The max output voltage is limited by the use of MJE340/MJE350 and by the size of heat sinks on the mosfets.

I cannot stress enough the importance of using LARGE heatsinks. The IRFBC40LC is shown here only for illustration purposes. Better use large package like TO-247 n-channel devices with large heat sinks.

This circuit works for me but I make no claims and guarantee nothing.

WARNING/DISCLAIMER: THIS PROJECT USES DANGEROUSLY LARGE VOLTAGES! DO NOT ATTEMPT BUILDING IT UNLESS YOU ARE SKILLED IN USING LARGE VOLTAGES. I AM NOT RESPONSIBLE FOR ANYTHING BAD THAT HAPPENS WHEN/IF YOU BUILD THIS PROJECT.

[Salas]

How high OLG at 1kHz and 20kHz?

[ikoflexer]

About 67dB in simulation.

[Salas]

Using it for cans also? Could bypass that 33uF maybe. Is it low esr?

[ikoflexer]

You're probably thinking about stability issues (your question about the OLG). You know I've put a lot of time into that type of analysis in the past; I have good reasons not to trust the stability analysis that is simulated. That's why I built it and found a configuration that works well in reality. I guarantee nothing for other implementations. It's not a circuit to build by anyone with lack of experience.

Haven't built a higher current version yet (for a headphone or power amp), but I probably will in the future; will use large package mosfets and huge heat sinks, that's for sure.

As for the 33uF, the cap I used is fairly low ESR yes (chosen for large D from among a few electrolytics). In principle it can be bypassed with some nice film cap but I haven't done it so I can't say anything about the result.

[Salas]

I was thinking about noise gain actually and how good quality/low value a c2 can be afforded for the ear not to pick intrusive additional hiss depending on the application. For stability you mention, it only fully proves in practice, and again its a matter of certain or wider application use deciding safety margin compromising to an extent or not.

[ikoflexer]

Oh, I see. That's something the usual tweakers with a good ear can investigate and I'm happy to leave that issue to them. That part of the circuit is identical to your HV shunt reg and I imagine the same solution applies. For my needs the 33uF value was best (slow rise to spec-ed Vout).

[Salas]

That time can be trimmed with different Vref Iccs amount also.

[ikoflexer]

Here's an update, using TO-247 package mosfets, the IRFP450. This is what's actually employed right now in my #26 preamp and this is what I'd recommend over the use of the IRFBC40.