Rebuilding electrostatic speakers, help me pick a modern HV rectifier?

[cvanc]

Hi all-

I am restoring an old set of electrostatic speakers and their HV supply needs new rectifiers.  It's a fairly straightforward double-doubler arrangement that generates ±3.2KV DC for the stators.

It's quite old school on the inside; these are 70's vintage and the rails run at powerline frequency.  The original rectifiers are unknown so I'm looking around for something modern to drop in.  It seems to me the requirements are quite similar to the rectifier used in microwave ovens.  I find a lot of possible p/n's and have attached a couple sample datasheets.

Be aware I do have inrush to consider as the filter caps are about 20uF.  I am OK with adding a resistor in series w/the transformer secondary to manage this.

I only need a few so cost is not too critical.  Ultimately I'll be driven by what I can buy in small quantities here in the States.  I would love to know if any of you have a suggestion for both the part and a good source.  Digi-Key and Mouser don't seem to have stuff like this.  Thanks.

[floobydust]

Microwave oven rectifier diodes are overkill, they are huge at 22mm and if you have space then use them, but on a pc board I would fit the same size parts to maintain HV clearances. What kind of room do you have?
I would use fast recovery as higher current diodes are slower, the originals surely were lower current parts. Note the CL family is rated 450mA avg. IN OIL and "blank" recovery spec. HVM family has no recovery spec. it's slow. Vishay GP-02 to 4kV 250mA and small 5mm, but slower 2usec trr. I get panasonic inverter microwave ovens out of the garbage and remove their fast recovery diodes, used to be Sanken UX-C2B. RP1H is 2kV 100mA 100nsec trr 7.2mm

20uF at 3.2kV is huge. Are you sure about that? I think the doubler caps must be small ceramics and end up limiting the current, with a big 20uF as the final filter cap?
Suggest Ohmite OX/OY series HV resistors for inrush.

Ceramic capacitors actually age, I would check the multiplier caps (without them arcing to your multimeter, discharge them all first). Modern ceramics are not as good, the cheapies from china on eBay have a huge voltage coefficient and end up less than 1/2 rated capacitance at 1/2 voltage, so upsizing them is mandatory.

[tggzzz]

http://hvstuff.com

[cvanc]

Thank you both.  I had missed that oil spec completely 

Here's the schematic, it has a built in tube amp running off the +HV rail.  This explains the big filter caps.  Interesting design.

[james_s]

Is there something wrong with the existing rectifiers? If they work then leave them alone, diode technology has not really changed much since modern silicon diodes appeared in the 1970s. They don't degrade with age so unless some of them have failed, don't mess with them. As the old saying goes "If it ain't broke, don't fix it."

[floobydust]

The M40 diode, can't find anything on it, a 1978 build this 2SW-1. D206 and D207 see much higher current than D204, D205. If you know the secondary's DC resistance you can estimate inrush and ripple current. I would expect the HV capacitors to be bad, they don't age well.

Unusual, never thought I'd be seeing a 35KD6 circuit that wasn't in a colour TV lol. I used to sell/replace those by the carton. Not easy to troubleshoot this at all especially with +/-3.2kV

I think the diode's operating PIV is 3.3kV (1,650Vx2)

[cvanc]

Is there something wrong with the existing rectifiers?... As the old saying goes "If it ain't broke, don't fix it."

See photo.  The original rectifiers are long gone and what's currently in there (4 strings of 3x 1N4007; on the right) is blown.  Hole spacing on the board is ~12mm but I have room to place a larger part.

If you know the secondary's DC resistance you can estimate inrush and ripple current.

  About 90 ohms.  Power transformer reads OK with an ohmmeter but it's unknown if it will break down when powered.  I hope not...

[NiHaoMike]

Since there's plenty of space, just use microwave oven diodes. They're cheap and robust.

[David Hess]

Mouser has the Rectron R5000F 200mA 5000V 500ns DO-15 30pF and they are cheap:

https://www.mouser.com/ProductDetail/Rectron/R5000F-T?qs=sGAEpiMZZMtbRapU8LlZDyxbre29nH7ScjESEi%2FfqK0%3D

I would also consider pairs of the tougher Vishay BYT62 350mA 2400V 5us SOD-57 avalanche rectifier but they are more expensive and twice as many would be required:

https://www.mouser.com/ProductDetail/Vishay-Semiconductors/BYT62-TAP?qs=sGAEpiMZZMtbRapU8LlZD88ZClzmvJPC3D9UQTWWnP4%3D

Or triples of the BY203-20S if I wanted something fast and tough:

https://www.mouser.com/ProductDetail/Vishay-Semiconductors/BY203-20STR?qs=sGAEpiMZZMtbRapU8LlZDxcRxaviIn1yTF8Dk6v3UhI%3D

The Vishay parts above are avalanche diodes so do not require balancing networks when used in series.

I am not a fan of microwave oven diodes.  They seem awfully fragile.

[james_s]

I don't recall ever having a microwave oven diode fail, in an oven or otherwise. I do think they're overkill for this application but if you can find them cheaply they will certainly work.

[SeanB]

I have seen them fail, but in general what kills them is an arcing magnetron, or a dead short capacitor, and in that case the transformer would also be rather cooked as well, because most people will run it till the fuse blows, and the smoke stops, though there could be a nice fire inside by that time. The variants with the overvoltage protection double diode though that often is a cause for fuse blowing, in either the HV or supply side. Otherwise the diodes rarely give trouble, except to arc over from grease and oil deposits collecting enough dust on them.

[jmelson]

Thank you both.  I had missed that oil spec completely 

Here's the schematic, it has a built in tube amp running off the +HV rail.  This explains the big filter caps.  Interesting design.

I can't figure out this power supply.  Something is missing or wrong.  Note that one diode D207, provides the +3200 V, but it takes a
4-diode multiplier to provide the -3200 V.  Also, one end of the transformer secondary is connected to the +1600 V point.  This doesn't make any
sense at all.  Is this an original manufacturer schematic, or one drawn up by the person who so badly mauled the original
(looks homemade, too) design?

Jon

[floobydust]

The schematic is confusing around the transformer, two 1/2 wave rectified rails providing +/- 1,650V but ground is taken at what would be -1,650V point.

I kind of think secondary resistance seems low, I have a 1,250VAC 300mA (much bigger) transformer and measure 85 ohms compared to your 90 ohms on what should be much thinner wire. I would do a test of the tranny with a variac or light bulb on the primary. I use old CCFL tubes+ballast resistor for a weak load. Just to check for shorted turns as it had to cook parts under an arc and the main fuse did not blow. What parts burned up?

Some q-tips and IPA to clean off the flux and soot. I see the pc board is carbonized from arcing. That region you can't use it anymore at 3.2kV I would build a mezzanine board out of FR4 or perfboard and mount it on standoffs or spacers above the main pcb with flying leads to connect. Or drill new holes in the PCB far away and move the parts there.

I think the rectifiers operate around 3.3kV PIV so three series 1N4007's is not a great idea especially without equalization resistors or caps. I'd put in a single fast-recovery part 4kV+ for less hassle.

Someone was in there doing mods, adding the two sets of 8 mica caps 1,000pF 500V//a smaller part across the 47uF 450V electrolytics, meh that could be done better if trying to filter RF there.  The solid-state section someone has upsized the electrolytics from 100uF 80V to 220uF 300V which seems silly. Even 160V parts could be used there. The 36KD6 47uF 63V looks upsized to 220uF.

[cvanc]

Thanks everyone.

Yeah, these are in pretty rough shape right now.  They are not homemade although I get they have that look.  They were made in small quantities and are supposed to be quite good.

I re-checked my notes and the secondary winding is 85Ω.  I suspect the tube amp draws significant current, at least when driven hard.  The datasheet for the 36KD6 says it can peak over one amp - yikes.  I also don't know what the standing bias current will be, there is no adjustment and no notes about typical values.

I think different rectifiers are needed for the +HV and -HV rails.  The negative rail D204/D205 is clearly low current and seems suited for one of the numerous microwave oven diodes.  They represent cheap overkill so why not?  I'll see what I can source.

The positive rail D206/D207, on the other hand, should tolerate much higher current.  It's a gut hunch but none of the parts discussed here - all topping out at about half an amp - feel good to me for this job.  Not if those tubes dig in hard.

I think I'm back to a string of 1N4007's, certainly more than 3, and with equalizing resistors.  Except... I have like a hundred 1N5408's.  6 amps 1KV.  Five or six of those, with resistors, would also represent cheap overkill.  If I can make it fit I might just do this.  That's a big if.

This will be a long slow project but I'll post the occasional update.  Again thanks for the discussion.

[cvanc]

It's a fairly big transformer

[james_s]

That doesn't look too bad, just dirty. I think I'd run that PCB through the dishwasher before I even started messing with it though. Same with the plate the transformer is bolted to, probably would hand wash the transformer though and then set it on a heat vent for a few days to thoroughly dry.

[floobydust]

I say no to using 1N5408's, tried that and they failed where the 1N4007 (string) survived. I think due to bigger diodes being slower, more capacitance. UF5408 I have not worked with but it has such fat leads and higher capacitance.
For high voltage multipliers, I found bigger (higher current) diodes have more leakage current which can make ripple worse. So bigger is not always better, at some point.

Wow that transformer is huge, I forgot it must take a lot of energy to move the panels, to move air.

[cvanc]

Wow that transformer is huge, I forgot it must take a lot of energy to move the panels, to move air.

Yeah it's pretty big but I don't think that is why.  I think the tube amp is the reason.  After all the negative supply is quite lightweight, and it takes exactly the same energy to drive the membrane forward as backward, right?  So the panel bias seems to just be a trickle, as you'd expect.  And everything else goes to the amp.

To be honest I wonder if even 1A rectifiers are appropriate.  It's unknown what the amplifier draws, but both the tube & the transformer appear capable of significant current.

So... what does 'just the right part' look like here?  How about something like a string of BY228?  Datasheet attached; I'm just more comfortable with this higher current rating (3A).  And avalanche diodes do not need equalizing resistors at all, right?  That makes the retrofit neater if nothing else.

Thanks again.