HV SS Regulators?

[mhuss]

I have an old Conrad-Johnson "hi fi" tube amp on the bench with a toasted power supply. It has a really old-school setup - a PNP constant current source feeding a string of twelve 33v Zeners. Two points on this chain are tapped to Darlington NPNs that provide regulated power to the preamp stages. All the transistors have a max VCE of 300v and the raw supply is close to 500.

The problem is, if anything goes wrong you end up with a popped PNP and a dozen toasted Zeners (plus other casualties). This is what has happened - and not for the first time. The PCB pads for all the Zeners are in sad shape (the PCB is an old-school tinned plain board with no coating, etc.)

Is there a better way to provide regulated voltage with modern parts? I've used 900v MOSFETS to do voltage /reduction/, but this is not regulated. Is there a better way to provide a (say) 8 x 33v reference?

[R.G.]

None of this is any problem. I can zap you up a circuit.

I have an old Conrad-Johnson "hi fi" tube amp on the bench with a toasted power supply. It has a really old-school setup - a PNP constant current source feeding a string of twelve 33v Zeners. Two points on this chain are tapped to Darlington NPNs that provide regulated power to the preamp stages. All the transistors have a max VCE of 300v and the raw supply is close to 500.

The problem is, if anything goes wrong you end up with a popped PNP and a dozen toasted Zeners (plus other casualties). This is what has happened - and not for the first time. The PCB pads for all the Zeners are in sad shape (the PCB is an old-school tinned plain board with no coating, etc.)

Is there a better way to provide regulated voltage with modern parts? I've used 900v MOSFETS to do voltage /reduction/, but this is not regulated. Is there a better way to provide a (say) 8 x 33v reference?

Sounds like it was marginal to begin with. Show me the original schemo or tell me what the various output voltage levels are, and any guesses you have as to the currents.

When you say "regulated", do you mean that the outputs are actively regulated, or just a fixed (zener) voltage on a base?

[mhuss]

Totally marginal to begin with, but in the 70s there weren't a lot of HV SS options.

Just a fixed Zener voltage on the bases. ~1ma through the diode string and probably 2-3mA max for each output.

[R.G.]

OK, got it. How close to the nominal output voltage do you think it needs?

MOSFETs have "infinite current gain", but the Vgs, analogous to Vbe in bipolars, is a few volts, and increases at about 1V per amp of drain current. This builds in a volt of two of slop if it's just tied to a fixed gate voltage.

I have a couple of ideas in mind, depending on how close you need to get.

[R.G.]

Mmmmm. Forgot to say - what's your budget constraint? A quick look at MOSFETs that will (probably) do the job looks like they run to $1.20-$2.00 each from Mouser.

[mhuss]

I *might* be able to afford up to $1.50.

Exact voltages are not important. This is a tube amp, lol.

Thank you very much for taking the time to do this.

[R.G.]

OK, first pass design done. Two options:
1. Active current limiter and string of 5W zeners. Works OK, but probably sloppy voltage tolerance. May not be cheaper than the more complex variant, but is mentally simpler.
2. Active current limiter, two active zeners. Each has low power, low cost zeners to actually set the voltages.

MOSFETs suitable to the task cost $1.20 each. First version uses one of them plus not-special other junk. The zeners in the first version cost about $0.50 each and you need 5-10 depending on what voltages you select.

Second version (and my favorite) uses two more active zener circuits. These each have one MOSFET to do the heavy lifting, a string of little, cheap ($0.15 or so) zeners to set the MOSFET zenering voltage. The two circuits differ only in the little-zener strings they use.

The current limit is resistor-settable. 39 ohms gets you 15.5ma nominal. This current flows into the two active zeners. Each of the active zeners is designed to flow from zero to 15.5ma through it without burning up. Each is designed to deliver 5.5ma out to the load at the design voltage.

Power in the MOSFETs in version 2 is about 1.5 - 3W. This is marginal for a TO-220 with no heat sink. Better to bolt a 1" square aluminum fin on them.

The rest is all jelly-bean Rs, Cs, diodes and so on.

Does that sound reasonable?

[mhuss]

Either sounds fine, but two things to keep in mind: 1) space is tight and 2) the chassis has no ventilation, and has several glass heat-producing items attached to the top. Thanks again.

[R.G.]

You know you're going to get an engineer's response to that, right?

How tight is the space? Got pics and measurements of the volume to put it in?
and...
To a certain extent, you're going to be wasting the same power any way you go with linear regulators. The circuit that's in there now will waste about the same power as what I'm proposing. A linear dropping string from B+ that eats current and lets some out to the rest of the parts in the amp will always waste the same amount of power taken from B+. Your choices (other than deliberately or accidentally making it worse) are only which parts you push the power into.

Here's a thought. Can you send a pic of the existing PCB space?

[mhuss]

I'll try to get a pic later, but it's basically an approximately 2.8 by 1.8 by 1 inch volume I have to fit it in. The original PS circuitry is on one big circuit board with all the audio circuits, which takes up most of the interior space not occupied by transformer bells, etc.

[R.G.]

That's good enough to start with. I think I can suggest a layout that keeps it in that volume.

Is there some chance of putting a baby board on top of the original circuit board to hold this?

[R.G.]

Hmmm. Next question: Can you solder surface mount parts? I can make it a lot smaller that way.

[mhuss]

Possible for the low-altitude parts (diodes, resistors).

[R.G.]

Can it stand 0.4" high?

[mhuss]

0.4 - I think so. SMD - ugh, really rather not!