+15V supply shorted

[(*steve*)]

I have a power supply which generates a +/- 15V rail. The -15V is fine, while the +15V measures around 50mV. The 7815 gets very hot.

I suspect it's a short

Unfortunately, the PSU is on a fairly large board, and I can't disconnect it from the rest of the board.

I've tried my calibrated finger all over the board, but the only thing which I've found that is hot is the regulator.

There is a 10uF tantalum (through hole, dipped epoxy style) on the output of the regulator.

So, the question: I know that these tantalums have a habit of failing short circuit. However, I'd expect it would fail more spectacularly (it looks fine). Does anyone have experience of a tantalum failing short circuit and pulling down a supply rail without appearing damaged?

 

[Harald Kapp]

Does anyone have experience of a tantalum failing short circuit and pulling down a supply rail

Unfortunately yes.

without appearing damaged?

Usually they go 'bang' rather spectacularly. But that depends on the energy available at the moment of short circuit. Depending also on the specific variant of 7815 the short circuit current is limited to 1 A or less. Together with a 15 V supply this may be enough to keep the short circuit in the capacitor active but not enough to create the expected spectacular effect.

 

[shrtrnd]

I'd be looking at the circuit further down the line that the +15 feeds.
Is it drawing too much power, or maybe a short in the circuitry there?

 

[davenn]

So, the question: I know that these tantalums have a habit of failing short circuit. However, I'd expect it would fail more spectacularly (it looks fine). Does anyone have experience of a tantalum failing short circuit and pulling down a supply rail without appearing damaged?

not without obvious damage, no
Always a first time

 

[(*steve*)]

Thanks for the information. If it is that tantalum, I bet the side I can't see has an evil grin on it.

Unfortunately I'm going to be tied up all weekend (and my weekend starts now and ends Monday evening) so I'm going to have to leave this for a while.

The resistance across the regulator can't be much more than 50mΩ, assuming the regulator is still working. There's no indication of the mains transformer being under stress. The resistance of the +15V rail from the regulator to the far end of the board is about 50mΩ too (coincidence I expect). The low resistance of the short means it's probably only dissipating 50mW, so it's probably running cool anyway.

When I get back to it I'll probably pull out my 6.5 digit 4-wire resistance meter and try to use that to locate the short.

I'll keep you all posted.

 

[(*steve*)]

Yeah, I promised I would leave it alone, but I suddenly thought I could use my ESR meter to probe for small resistances...

Near the regulator (across that tantalum cap) I see 40 to 50mΩ. Further away I see higher resistance.

And yes, it was the tantalum cap. Removing that raised the ESR to 4.5 ohms, and the DC resistance to something over 1k. The tantalum reads as a dead short.

I replaced it with a 10uF 50V 105C low ESR cap that I just happened to have within reach.

Unfortunately the +15V regulator now only delivers 11.5V, but it doesn't get hot. I imagine that it has decided it wants to visit the parts bin in the sky.

And wouldn't you know it, I don't have any +15V regulators. Oh well, at least I have an excuse to do what I'm supposed to be doing now.

 

[Harald Kapp]

Always a first time

... or after a year in service. Trust me, it has caused us a lot of trouble.

 

[(*steve*)]

Damn!

On the way somewhere I accidentally bought an LM7815.

 

[HellasTechn]

There is a 10uF tantalum (through hole, dipped epoxy style) on the output of the regulator.

Why do you suspect the tantalum and not the Vreg itself ?

I replaced it with a 10uF 50V 105C low ESR cap that I just happened to have within reach.

Unfortunately the +15V regulator now only delivers 11.5V, but it doesn't get hot. I imagine that it has decided it wants to visit the parts bin in the sky.

Ok now i see why

 

[Harald Kapp]

On the way somewhere I accidentally bought an LM7815.

Accidentally, aha, I see...
S* happens

 

[(*steve*)]

And it all came to life with the replacement of the 7815.

Here are the dead bits:



And here is the other side of them (and you'll see what was hidden from me all that time...)



Oh, and the new regulator runs as cool as a cucumber (makes you wonder why there's a heatsink on it actually).

 

[Harald Kapp]

Good. Finally you'll have time for your other assignment

 

[WHONOES]

It is worth noting that some voltage regulators have a tendency to oscillate if its output capacitor is missing or not adequate. They tend to get fairly warm when doing so. If it is, it won't always be obvious and can lead to a wild goose chase.

 

[(*steve*)]

True. But in this case 20 seconds or so was enough to get the heatsink up to uncomfortably warm, and extended period got it to beyond that.

Considering the other rail had a heatsink that was not noticeably above ambient, and the average voltage was 50mV, that seemed unlikely.

And given that my next measurement was the DC resistance across the supply rails (and it was effectively zero), I decided to look for a short first.

But yeah, looking for an AC component on a DC rail is another really simple check.

 

[WHONOES]

Yup. I wasn't actually suggesting that oscillation was your problem, just adding my thoughts to the posts as they might be useful to someone else having issues.