530V low-ish dropout regulator?

[dsavitsk]

I have D3a->300B based DRD amps with a woefully inadequate power supply. It is RCLC where the first cap is 10u, the second cap is 250u, the resistor is 30R and the choke is 15H/400R. It uses solid state rectifiers. B+ is about 540V, and unfortunately there is not a lot of headroom in the power transformer.

In an attempt to get the last bit of hum out of the circuit, I though I might work a regulator of some sort in there. Any ideas for something that will work at this voltage and not drop a lot of volts? Would it be ridiculous to try to work a LM317 in there? I'd be within the 37V in to out voltage, but I am not sure of any absolute voltage limitations. Maybe an LR8 with a pass transistor instead? Ideas?

[DougL]

My first thought is to use a cap multiplier, rather than a regulator.
If you can give up absolute voltage regulation, and only use the active device for AC reduction, you can live with significantly lower voltage drop.
I think TCJ has several blog entries on this.

If you have the loaded AC of the transformer and current draw, I could play in PSUDII with the unregulated portion of the supply.

Doug

[ikoflexer]

What would be the max acceptable drop?

I too think that you'd do well with a gyrator/cap multiplier type circuit. You could get a lot of filtering this way.

[wa2ise]

Back in the olden days of vacuum tube color TV sets, they used to regulate the 25KV (for the picture tube) with a shunt regulator. This was a tube that would draw current from the supply to load it down if the picture tube wasn't drawing the supply down (like when it's a dark scene in the TV show, vs a bright sunny day in that show). This tube, a 6BK4 was a funny looking triode with a mu of 1000, was set up something like a voltage divider from the power supply to be regulated, to feed a divided voltage to the grid, and the cathode was held at some fixed bias. Supply voltage goes up, the grid goes up, and the tube draws more current which in turn loads the supply down, to make the supply voltage drop. Problem is that the energy you didn't consume in the desired circuit function gets wasted by this regulator. But in the TV set, there was no reasonable way to series regulate 25KV.

A scaled down version of this topology might do what you need, if you don't mind the wasted power.

[dsavitsk]

Thanks for the suggestions. Here's what the current supply looks like. Because of the Ultrapath output cap, the ripple is being multiplied by the mu of the 300B (~4) which is too high overall. However, it looks like just increasing that 10u cap to 30u or 40u will bring this into a reasonable range. So, that seems like the easy first step.

[athos56]

Paralleled sections of a 7236 tube would be good till 600v. Rated for 15watts of dissipation. I'm thinking of putting some in the T-reg that Janneman designed.