Reeves Custom 50PS non-polarized electrolytic replacement?

[Helmholtz]

With power tubes pulled and the NP cap disconnected, what DCV voltage do you measure at the point where the non-grounded end of the cap was connected?
Positive or negative?

[R.G.]

If the suspicions, intuitions, etc. are correct(ish) and it's part of a variable B+ or variable screen supply setup, I can guess that some certain conditions of power on/off, broken wires, broken controls for the variable voltages, the main power caps might be able to generate a transient that reverse-biases the cap.
Maybe. Stranger things have happened.
An NP would be proof against most of that. But I can't figure out why, if that supposition is true, there would not be a normal polarized cap and a 1N4007 reverse-protection diode paralleled with it to catch transients. Maybe the designer didn't know why it was failing and just tried an NP part, which (mostly) fixed it. Deadlines make designers do strange things, sometimes. I have been in situations where management was yelling "don't go study the problem to death!! Just get it fixed NOW!!"

[Helmholtz]

Could also be part of the variable bias.

[Beyer160]

With power tubes pulled and the NP cap disconnected, what DCV voltage do you measure at the point where the non-grounded end of the cap was connected?
Positive or negative?

71.5v DC, positive

[x-pro]

this is getting more and more interesting...
such potential can only be here.

[R.G.]

It might help to see the voltages on the output tube pins are in operation. Next test I would like to see (failing a schematic or more tracing) would be the output tube pin voltages. That would nail down whether it's cathode biased or fixed bias, and also whether the output grids sit at the same voltage as the capacitor voltage.

We're still faced with the original question - is NP needed, or was it stuck in even though a polarized part would work. Or indeed, if it was ever non-defective.

The devil that sits on my left shoulder is poking me with his pitchfork. For boutique designs where the maker is being cagey about schematics, a modestly clever thing to do would be to build the unit with a part in a special place that is defective in a special way. That way it would take a very skilled and experienced copier to ever get a copy to work, given that most copiers would "repair" the defective part and now nothing would work well in the original or the copy.

Just because you're paranoid doesn't mean they are not really out to get you.

[Helmholtz]

this is getting more and more interesting...
such potential can only be here.

Please elaborate on attachment (I'm lazy).

[Helmholtz]

71.5v DC, positive

Does the DCV vary with the vary-power control?

Did the amp work fine until the cap blew?

[x-pro]

Please elaborate on attachment .

https://hiwatt.org/tech2.html

[Helmholtz]

I see, so you think the cap is part of the PI bias supply.
Could be. At least a cap from grid (or cathode) to ground probably wouldn't hurt.
But no reason for a NP cap.

[g1]

I believe he said the top of the cap goes to the power scaling board. Also it was described (posts #1 and 4) as a simple voltage divider coming off one of the supply nodes, with a 330K and 100K in series. The NP cap is across the 100K.
With a 330K in line, it's not going to be the supply line for the PS board. I think that still leaves supply tracking (probably screen supply) to the PS board the best guess.
Some of the London Power PS kits also scale the bias. I think that may be happening on the PS board here as well, being that there are no bias components visible where they usually show up in the DR504 circuits.

Still nothing that would call for an NP cap, which is possibly a 'red herring'.

[Mick Bailey]

Due to the range of voltages, tube types, and bias arrangements the power scaling module needs to be configured to work effectively across a number of different situations. There are also verious versions of the module that have different features. I don't have the TuT books, but power scaling and how it's implemented is described in some detail.

One feature is that some modules can accommodate half-power switching and can track B+.

I recall Kevin O'connor posting here a few times, but that was quite a while back.

[R.G.]

I read through this again. It occurs to me that we're not solving the problem, we're redoing the design. At this point, if I had the original poster's problem, I would get two 100uF polarized caps and two diodes and make a two-cap-series non-polar cap from them.

[Mick Bailey]

R. G. - I think you have a typo. Original cap is 10uf.

I saw the posts myself as trying to rationalize the purpose of the NP cap and learn more about the circuit. When I see something unexpected I'd rather know why, if possible.

Of course, sometimes if you have to just get something repaired time may be against you.

[R.G.]

Doh!! Yep, 10u.
And I agree, knowing why is always better - but sometimes just getting it fixed is critical.