Help Identifying Vintage Component

[Drjaymz]

I was testing an over voltage relay and it was not working (its from 1970s home built aircraft).  I reverse engineered it and its reasonably straight forward, potential divider, NPN transistor activating the relay when the supply > 16V.   The potential divider is actually 3 resistors, with the middle one being a trimmer, the transistor base tap comes from the top of the first resistor that is grounded. 

This was a dead short to ground (hence why it wouldn't activate) with a component in parallel which I assumed is a capacitor to prevent it triggering on noise etc.  When I remove that component the relay works correctly switching at just over 16V.

Mystery component attached.  What made me question my assumption was that I wondered if it was a zener or something else and I wondered if anyone else could identify it.  It looks like a polystyrene cap and the positive lead is offset which is consistent with some capacitor designs.  15V75 sounds a bit like a zener voltage but we have to remember that its also from around 1975 and may just be a date code of something like week 15 1975 maybe. 
The 4070 possibly the value 40n *10^7 or something buy thats a weird value. 4.7 something would be more normal.

Its worth checking over voltage relays because when they don't work the first you know if when your avionics explode or your battery literally boils.

PS I already had an account on here but it says I didn't exist so just had to create account again.

[floobydust]

It looks like a tantalum capacitor, probably wet 75uF 15V.

[factory]

Looks like an axial wet tantalum capacitor to me as well, they do seem to use odd values a lot.

David

[fzabkar]

I used to see these tantalum capacitors in old Control Data disc drives. When they failed, they would burn a hole in the PCB. They are definitely not suitable for aerospace applications.

[Drjaymz]

There are actually a couple of them in there.  I know that tants like to short when they fail. The relay module is a Prestolite X17621 and they are certified but obviously do fail.

Being certified doesn't necessarily mean it uses military or special grade equipment and in fact a lot of what I have seen is of a standard you wouldn't allow in a car.   I'd expected there to be some form of freewheeling diode or snubber but there is nothing at all, to be fair it shouldn't usually activate and it has done 50 years. The reason I took it apart was that the contacts had a fair bit of resistance, which with the ropey design of aircraft alternator systems results in strong oscillation in the charge current.

Are we saying its 75uF 15V?  Sounds high for a cap that isn't much larger than a 1/2W resistor and from the 1970's.  I imagine that the capacitance has probably escaped over the years.  I'd like to put another cap in, we don't want it spuriously activating when the radio transmits.
I cleaned the contacts carefully and got the 2 ohms down to the same reading as just the test leads so thats good and there's a rubber seal which used to seal it all up but that gave up sealing probably 25 years ago and let corrosion in, so I will deal with that too.

[coromonadalix]

hum  not sure at the time if it was 7.5 uf      missing dot scheme  or exponent value    can you find one not shorted out and try to pop it out of the pcb and measure it

[floobydust]

[...] Are we saying its 75uF 15V?  [...]

No - a 15V tant. cap in something running off >16V or even 12-14V makes no sense.
Do you have more info or pics of the board?

[factory]

Are we saying its 75uF 15V?  Sounds high for a cap that isn't much larger than a 1/2W resistor and from the 1970's.  I imagine that the capacitance has probably escaped over the years.  I'd like to put another cap in, we don't want it spuriously activating when the radio transmits.
I cleaned the contacts carefully and got the 2 ohms down to the same reading as just the test leads so thats good and there's a rubber seal which used to seal it all up but that gave up sealing probably 25 years ago and let corrosion in, so I will deal with that too.

They are often smaller in size than regular aluminium electrolytic caps of the same age.
The rubber seal types have a high failure rate, the seal perishes and leaks out the corrosive contents. Those with a glass hermetic seal are much better and rarely fail, unless the voltage rating wasn't high enough.

David

[shakalnokturn]

[...] Are we saying its 75uF 15V?  [...]

No - a 15V tant. cap in something running off >16V or even 12-14V makes no sense.
Do you have more info or pics of the board?

OTOH if it's between transistor base and ground as it must have been to not go up in smoke 15V is plenty. As for the value I too find that 75μF is a little optimistic considering voltage rating and age.

[floobydust]

It would make sense having the cap across the transistor input E-B or in the divider chain. I wondered if it was across power internally.
I'm expecting a zener diode somewhere because a voltage divider/trimpot and transistor alone does not work well to trigger from a certain voltage, it's very finicky especially over temperature. We don't know if this module latches (SCR) or has hysteresis etc.
The Prestolite over-voltage relay X17621 won't win any beauty awards.  It would be neat to see the insides.

If the cap is across the input then it's just filtering B+ and EIA standard value of 82uF to replace the 75uF would be reasonable.
It is standard practice to derate these tantalum caps to 50%-60% applied voltage. So a 15V part max. seen should be 7.5-9VDC.

Axial solid tants are expensive ~$20 each. If it is close to the engine I would make sure it can take heat, although the original had epoxy end seals it looks like.

Mil-spec Vishay 150D series closest is either:
82uF 15V 150D826 and size S code is 0.786" L 20mm x 0.351" dia. 8.9mm
68uF 15V 150D686 and size R code is 0.686" L 17.4mm x 0.289" dia. 7.3mm
These are good -55°C to +125°C but harder to find. Digi-Key has no stock, Mouser has the 150D686 for $8.
Alternate is molded Kemet 322 series T322E826 size E code is 0.530" L 13.5mm x 0.280" dia. 7.1mm but still $21 each. Ouch, Mouser has stock 80-T322E826K015.

Today, this is an oddball capacitor so it might take some hunting.

[T3sl4co1l]

Nah not tant, looks more like M39018 series aluminum electrolytic to me.  Example: https://www.tti.com/content/dam/ttiinc/manufacturers/barker-microfarads/Resources/bmf-min-axial-alum-caps.pdf

Tim

[Swainster]

If it was a mil spec cap (and it certainly looks like one) then it would make sense that '4070' was the dash number... which in turn suggests that the capacitance is 6.8uF. However, if it was part of the M39018 series aluminium eletrolytics then the voltage would be 250V, with an obsolete tolerance code. On the other hand, if it was a M39003 tant then a '4070' makes a lot more sense - it would be 6.8uF 50V ultra high reliability, however the epoxy end cap doesn't make sense. Size wise, hard to say from the pic, but the tant option looks more plausible.

Conclusion, maybe 4070 is a red herring...

Cut it open to see what it looks like inside?

[fzabkar]

Modern tants in a similar looking package:

https://web.archive.org/web/20200925215647/www.mallory-sonalert.com/tantalum/articles/TS5%20Data%20Sheet.pdf
http://www.newyorkerelectronics.com/Mallory_Sonalert_TS5_Tantalum_Capacitor

[floobydust]

Very early, tantalums were known for high reliability and used in military and aerospace. So I say it's not an electrolytic.
Logo "H" OK I found the manufacturer is Hilton Capacitor but no joy on 4070 part number.

[fzabkar]

Logo "H" OK I found the manufacturer is Hilton Capacitor...

https://web.archive.org/web/20050210214206/http://www.tantalum-pellet.com/

[T3sl4co1l]

Logo "H" OK I found the manufacturer is Hilton Capacitor...

https://web.archive.org/web/20050210214206/http://www.tantalum-pellet.com/

Wow, looks like a 50s era photo; and the other identical but for air-brushing out the identifier (presumably after being bought out), didn't even bother to photograph new parts, haha.  Nicely spotted!

Tim

[Drjaymz]

Wow, I would say that these are exactly what they are.  So we know its a capacitor.
Picture attached.  The board is a fibre board, hand twisted blobs soldered and epoxied on - which looks a lot like 60's era stuff.  I can't easily remove the fibre board because the tiny relay windings are bonded to it at the bottom and I don't want to break it, especially as it now works.  I warmed up the solder and removed the cap with tweezers and any replacement will probably go on the outside.

I have traced out the diagram and will post that a little later.  The capacitor goes between the blue wire (ground) and the top most blob in parallel with a 6.2k resistor.  That then goes down the 20 degree slopey bit to the trimmer.  And that junction has an 11.1V zener that pops out below the trimmer and goes across to the transistor base (which you can see).  So that capacitor does see considerably more than 15V.  In fact we have to see > ~ 12V before the base lifts above zero at all.

At the end of the day, its just a ripple filter for something that shouldn't ever activate so its value isn't that important.  What's important is that it isn't open or short and has some capacitance probably in the 10 -> 100uF region.

I'll draw up the schematic in a translatable form because my mistake was drawing it as a standard 12V top 0V bottom schematic which confused the hell out of me.  It also latches once triggered and I couldn't work out how it did that last night but maybe will have better luck today.  Now I know they are caps that makes tracing it a little easier because when you're reverse engineering you tend to draw what makes sense rather than what you actually see.

BTW there is another tant filtering the RHS and I tried to measure that my disconnecting one end and it has no capacitance whatsoever, couldn't get it to read on my meter, so its either duff or something really low like a few pf.  But if you think about it they probably should be in the uF range given the time constant, because the relay isn't going to respond at khz or Mhz anyway its going to trigger or hunt in the hz region - which is my reasoning.

So quick summary of what this is meant to do, in an aircraft alternator excitation is switchable, and because of some incident probably in the 1930s they decided they don't trust the regulator to regulate and so for safety they measure the bus voltage and if it exceeds 16V then the excitation is cut off completely. Because in flight the engine is 75% full power if the excitation fails at full then the alternator wants to produce ~ 300V and quickly becomes an issue because the batteries are lightweight i.e. 35Ah or less.  That and the satnav costs 15k and any instrument that fails is probably 10k.  You'd think if you bought a modern cirrus for £1 Million that it would be more sophisticated, but thats not how aviation works. No, we still have mixture control and magnetos.

[fzabkar]

Something like this?

[Drjaymz]

Something like this?

That is just the regulator.  I can see it uses the zener reference to control the field current (I say control I think its on or off).  So if you look at my attachment this circuit is in the regulator part.  Its more or less the same as that internally.  It actually has a neon indicator used as a TVS device which is quite neat, and there is one capacitor between the zener and ground.  In the actual aircraft that capacitor is also toast and thats the reason the ammeter bounces up and down.  I do actually have a new (June 2022) certified part for the regulator.  But I wanted to check the OV relay actually worked on the bench and I subsequently found out that it didn't!

[Haenk]

Some serious measures have been take to ensure the trimmer is not re-adjusting itself. Ever. 

[Drjaymz]

Reverse Engineering Complete (Big Clive would be proud).

But, the bit I don't understand is that if R3 really does take its supply from the -ve relay coil end then when the relay is energized doesn't that then turn itself off again.  Because TR1 is pulling R3+ve to ground.  I thought there might be a wire behind, but probing it suggests its as shown.

When it actuates what you actually see is around Vbe  ~ 0.8V T1 and about 0.7 Vce.  And it latches on.

If you then reduce the voltage to about 4V then the relay lets go but if you increase comes back on again.  Until you turn the circuit off again.
So could it be that ZD1 is actually not a zenner?  How does it stay on?  Its markings are 4545.

None of this is really relevant as I just need to replace c1 and c2 and get on with life, but I still wanted to work it out.

[David Hess]

It looks like a hermetically sealed axial tantalum capacitor to me.  It could be either wet or dry but this can be determined from what type of seal it has.

Without the epoxy coating causing strain in the tantalum pellet, these axial capacitors support up to 75 volts without derating and the risk of failing short so they are common in high reliability applications.  Based on the part number I suspect it really is a 75 volt part but 1.5 microfarads instead of 15, which is consistent with the size.

[Drjaymz]

That sounds about right to me, and 75V should be ok for the intended application.  You can in a 12V system see really huge spikes of any polarity but its not directly across the supply.  I would say its done ok for 50 years old.  But then again, we don't know when it actually failed because the relay will fail closed which is the non-fault condition.

So question comes to what to replace it with?  I have 2.2uF 100V multilayer - but they still feel far too modern for me and 2.2uF electrolytic, I think you just have to reckon that these will gradually get larger ESR and lower capacitance over time.  Capacitors - still the dilemma they always were.

On films they start up a million year old alien space ships and everything works fine, but we can't get capacitors to work for a couple of decades.  Good job those aliens didn't use rifa caps.

[Swainster]

Hilton SWT wet tantalums appear to be still in production, now made by QuanticTPC (formerly the tantalum pellet company). You can even "tour" the inside of their office/lab (factory?) via google maps. To be honest, it looks like each cap is being made by hand - you could probably get them signed by the operator

It's probably worth checking with them about the specs of the cap, given the application.

[Drjaymz]

Hilton SWT wet tantalums appear to be still in production, now made by QuanticTPC (formerly the tantalum pellet company). You can even "tour" the inside of their office/lab (factory?) via google maps. To be honest, it looks like each cap is being made by hand - you could probably get them signed by the operator

It's probably worth checking with them about the specs of the cap, given the application.

Nice!  but I can't see it being necessary to chase down the exact parts.

I looked up the relay and they did make them by hand until about 1990 about £400 and now they are cylindrical construction and advertised as with "integrated circuit" for reliability.  As if measuring a voltage and activating a relay was something difficult requiring pixie dust and that being able to trip within a millivolt is somehow superior when you consider that the system is already 2 full Volts over-voltage at this point.

And by the way, that relay isn't anything fancy, the field current is up to 4A and in reality a couple of amps.  But for anyone who doesn't know, DC is sometimes difficult to interrupt because the electric field never falls to 0 like it does for AC.