Calling the veterans - what is this?

[Fdeschen]

This was in a 1/4HP 115V DC shunt motor drive, dating probably from the 70's. There is 6 of them in a parallel configuration.
I initially expected them to be braking resistors, but (the surviving ones) they measures at ~2-3 MEGA ohms... I was expecting something in the 100Ω range for that application, so i'm confused.
They looks like high voltage isolators, but for what?(it's just 115V).

3 of them literally caught fire while starting the motor and I am trying to find replacement. I've got no luck searching for the part number (3D31O1T).

Thank you

[jdragoset]

Selenium diode stack.
Each plate is good for 20 volts reverse so ten in series roughly 200 reverse volts.
Forward current depends on plate size.

Universal Rectifiershttps://www.universalrectifiers.com builds them in the US.

[George Edmonds]

Hi

They are dual rectifier stacks, possibly selenium or copper oxide.

G Edmonds

[themadhippy]

3 of them literally caught fire

Ooh, that smell

[bdunham7]

I would expect that to be a bit earlier than the 70s unless someone was hanging on to old ways.  Can you sketch out exactly how these were connected in your circuit showing everything from line-in to the motor?  They can likely be replaced with just a few modern rectifiers, but the voltage drop will be less so your motor may become a bit more energetic.

[Fdeschen]

The Diodes! That explains everything... I was searching for diodes in the drive circuit and could not find them, never guessed they were in my face all along
Well i'm gonna replace all of those with more "modern" ones  I have a Variac anyway to adjust the supply voltage.

For the smell... yeah that confirms It's the right thing, it was like someone cooked Roblochon cheese for an hour on the whole floor of the building! I had to leave the room for the first 10 minutes it was disgusting.

Thank you very much!

[DavidAlfa]

Yep, a Selenium diode.
Absolute crap compared to silicon, they're capable of low currents despite their giant size, that one might be rated for 100mA-500mA....or even less!
Also they can stand really low reverse voltage (20-25V max), that's why you see lots in series, each one losing >1V in forward conduction.
Play too hard with it, the place will have a terrible smell for days

That one seems half of a full-wave rectifier (two diodes in series),  each half having 10 plates, so about 200V reverse voltage, will generate quite a lot of heat since it'll drop >10V.

Probably 2x 1N4007 can do waaaay better job than that!

[Gyro]

Yes, the silicon replacements will be tiny in comparison. Beware of the much lower forward voltage drop of modern parts though. You might find higher than expected voltages on capacitors, higher surge currents etc. Worst case, you can add series resistors of a few ohms and appropriate wattage to compensate for the difference - You will have plenty of space!

Your Variac will help here.

[james_s]

I had a selenium rectifier burn up in my uncle's workshop when I was about 8 years old. I can still smell that stench in my mind when I think about it. It seems like the smoke it made was colored too but I could be imagining that.

[TheMG]

Yep, a Selenium diode.
Absolute crap compared to silicon, they're capable of low currents despite their giant size, that one might be rated for 100mA-500mA....or even less!
Also they can stand really low reverse voltage (20-25V max), that's why you see lots in series, each one losing >1V in forward conduction.
Play too hard with it, the place will have a terrible smell for days

Add to that the fact that the selenium rectifiers degrade even in storage. You can take a new-old-stock one right out of the box and find it's no good.

[George Edmonds]

Hi

You have had only one selenium rectifier burn up, how lucky you are.

In the 1950's/1960's there was a flat pack version from Philips used extensivley, they died like flies and the smell was vomit making.

At one time there was a mechanical and physically identical version available but using silicon diodes.

G Edmonds

[james_s]

Something to be mindful of is selenium rectifiers have a much greater forward drop than silicon diodes, you can't always just do a direct swap.

[TimFox]

One of many problems with selenium rectifiers is that they are nowhere near so passivated as good silicon devices.
There are many ancient technologies for which I have no nostalgia.

[strawberry]

couple years ago sold selenium power rectifier 200V 5...A
have some selenium rectifiers but not voltage you need

why everyone tell that selenium rectifiers smell??

[james_s]

why everyone tell that selenium rectifiers smell??

Run one long enough or overload it and you'll find out.

[George Edmonds]

Hi

You will also never forget it.

G Edmonds

[Wallace Gasiewicz]

I bought a 1920's ( maybe 1930's??)multimeter that had a 1KV AC function. There were two "stacked" diodes in 2cm dia bakelite cases. One was malfunctioning, I took it apart and found round 2 cm dia round "crystals" stacked inside. They looked like like the crystals inside big old crystal cans.
I suppose these were selenium

I put in a few modern diodes and the Multimeter works.

[floobydust]

I've never been able to swap out selenium rectifiers with silicon ones and not have hassles. Because of the much lower forward voltage drop and lower on resistance, and lack of reverse voltage protection that silicon parts have in comparison.

OP I would expect the drive to experience much higher stress due to the current-limiting and voltage-limiting the old diodes provided. It depends on the motor's load but you might want to add series resistance and MOV's to protect the windings and drive from back EMF. You'll get much more torque and speed- if the drive doesn't croak.
For a replacement, I would put in an oversized 25A 1,000PIV bridge rectifier module (a few dollars) like GBJ2510 or GBPC2510 has crimp terminals which might be easier to deal with.
edit: there's 6 diode stacks? Then you need individual diodes like 6A10's 6A 1,000PIV each. I'm not sure how the motor is wired up.

Even today there are customers specifying selenium rectifiers for use in cathodic protection systems, they find they are tougher when it comes to lightning. The diodes act like a huge MOV is built-in there. There is some fungus that attacks selenium diodes causing their downfall, and selenium smoke is highly toxic.

[TimFox]

I bought a 1920's ( maybe 1930's??)multimeter that had a 1KV AC function. There were two "stacked" diodes in 2cm dia bakelite cases. One was malfunctioning, I took it apart and found round 2 cm dia round "crystals" stacked inside. They looked like like the crystals inside big old crystal cans.
I suppose these were selenium

I put in a few modern diodes and the Multimeter works.

The first meter I owned was a pre-war Simpson 260.  I think the diodes in the AC voltmeter section were copper-oxide.

[free_electron]

selenium rectifier. a.k.a. a skunk. (when they die they stink like a skunk )

[Fdeschen]

Wow, I did not expect to trigger so much memories with that thing!

Thank everybody for the specifications and suggestions. I have a box full of big bridges (like the GBPCs), I'll throw in a ±15A 1kV one, change the caps for "post 1950" ones and add some circuit protections (there was a fuse at least...)
 
The variac control the supply voltage and the motor is a shunt DC, so it will just be set lower to get to speed (there is a tachometer on it)

[coppercone2]

They have strong diffusion parameters which make them age poorly I believe. So they get a greater voltage drop and burn up? I think it has to do with selenium being 'soft' but I need to look it up.

Oh the Mohs scale, silicon is 3x harder then selenium. So I think this means there is more diffusion that occurs over time in doped regions.  Selenium has a higher thermal conductivity but the voltage drop made by the diode is greater which is not enough to offset the increased power generation, so the parts are already running hot and require pretty tricky thermal design to make work normally. As that voltage drop drifts upwards they get toasted (just my understanding of the situation, I only have one selenium rectifier and I never turned it on).

Maybe it was OK when it was brand new but if you study power electronics, you will see its all about reducing the voltage drop. Manufacturers prefer generating less power then trying to figure out how to sink it (i.e. old generation IGBT have massive thermal dissipation capabilities and are designed to dissipate heat, while newer ones have lower voltage drop and they focus on electrical parameters since they can get away with the heat).


The rectifier is built like a heat exchanger because I think its possible given the 2x greater thermal conductivity compared to silicon. I don't think it looks like a radiator for any electrical reasons. And I also want to stay its probably really fragile since that is a glassy metal formed into a crazy shape.

When new I don't think there is anything intrinsically wrong with it so long it does not fail (but is the smoke worth it?). And does it have intrinsic reliability problems because of the odd mechanical design?

[Kleinstein]

Hardness does not correlate very much with diffusion speed. If at all is the melting point, but even there die different crystal structure can make a big difference.The high melting point of silicon helps, but the relatively open diamond structure of silicon hurts it. Some elements (e.g. copper) can diffuse through silicon crazy fast.

Selenium diodes are really poor performance in most aspects.  Due to the different characteristics they still got used even in times when silicon diodes were available. Somehow they were slow to do the change over.

The slenium diodes need the shape to get rid of all the heat. They do have a reliablity problem, especially the older units and this is not due to the mechanical shape. It is just normal aging that makes then ticking time bombs with a distingt smell. A bit like the old Rifa caps from the 1980s, but with a different odor.
So using them in a power application is a risky thing.

[George Edmonds]

Hi

The reason they were used for some time post WWII and into the silicon era was that they have similar voltage drops to Valve (Tube) rectifiers and were a simple drop in replacement without changing the input voltages.

G Edmonds

[floobydust]

Selenium rectifier diodes seem to be common starting out in the early 1940's they did duty in WWII and it looks like a 30 year run for them. Mostly in power supplies - electroplating to 10,000A, radio and TV, HV, battery chargers etc.
Problem is there are no decent IV curves for selenium diodes. Every time I've replaced selenium with silicon diodes, I've had way too voltage and the device (battery charger, power supply etc.) ends up unusable without a bunch of mods.
They were also used in aircraft, military too and when they shorted the smoke was quite serious i.e. B-52's (400Hz 28V 50A) you can't just roll down a window but there were apparently no incidents thru 1957, but I did see urgent gov docs to change them out.