[tomb]

The difference is between a catastrophic failure and not. The last time I popped an electrolytic (coincidentally on the LNMP), I got a bulge in the aluminum can and some messy electrolyte oozed out onto the PCB. I wasn't digging out bits in my neck and chest from an explosion. Just saying ... but I won't say anymore.

 

[saraengelstad]

Yeah. I've seen cans dome and smoked a few resistors and diodes, that's where I came up with the lineup test. It's easy to see a backwards orientation or suspiciously small resistor that way, like five cheerleaders, only one is missing shoes... It jumps out at you. But I hated wondering what I might be digging out of me with a sewing needle. I worried over a glass of wine, getting into the baby, you know? (Was still breastfeeding then)

 

[saraengelstad]

I would be happier with that procedure if you had a known-good instance in the group for comparison. As you say, it is possible that you could get them all backwards.


I doubt it. I would guess that the regulator isn't regulating for some reason, so that the full rectified DC voltage is getting through, and that voltage is too much for the cap.

If that's an LM317 you've got bent partly out of the frame, I would cut the remains of C8 off the board then power it back up again — still being careful! — and check the DC voltage with at those pins or between TP2 and TP3.

If it's some other type of regulator, the cap may be necessary for its reliable startup, so you may not be able to do this test without replacing it first. If you do, be prepared for a repeat performance.

It is indeed an LM317. I'll do that, just for fun. The powerup testing went well, I ended up with 17 units that performed as expected (although I don't have my test procedure and results sheets handy) with all the test point voltages as expected. I have 2 incomplete boards, one of them I think taught me to check my diode polarity because there is evidence of desoldering. One unit is marked 'bad led', and the firecracker. I'm okay with my success rate.

You do have a point with the known good, because it is not unimaginable that I would repeat the same error 20 times, so I guess I'd have to rely on the first powered test to alert me... yikes. That's a lot of rework.

 

[saraengelstad]

The difference is between a catastrophic failure and not. The last time I popped an electrolytic (coincidentally on the LNMP), I got a bulge in the aluminum can and some messy electrolyte oozed out onto the PCB. I wasn't digging out bits in my neck and chest from an explosion. Just saying ... but I won't say anymore.

Hi, Tomb! Thank you. Please don't feel you are intruding. I can use all the ideas and perspectives I can get.

 

[saraengelstad]

I would be happier with that procedure if you had a known-good instance in the group for comparison. As you say, it is possible that you could get them all backwards.


I doubt it. I would guess that the regulator isn't regulating for some reason, so that the full rectified DC voltage is getting through, and that voltage is too much for the cap.

If that's an LM317 you've got bent partly out of the frame, I would cut the remains of C8 off the board then power it back up again — still being careful! — and check the DC voltage with at those pins or between TP2 and TP3.

If it's some other type of regulator, the cap may be necessary for its reliable startup, so you may not be able to do this test without replacing it first. If you do, be prepared for a repeat performance.

It is indeed an LM317. I'll do that, just for fun. The powerup testing went well, I ended up with 17 units that performed as expected (although I don't have my test procedure and results sheets handy) with all the test point voltages as expected. I have 2 incomplete boards, one of then I think taught me to check my diode polarity because there is evidence of desoldering. one unit marked 'bad led', and the firecracker.

 

[saraengelstad]

No offense at invading this thread, guys, but ...

She already said she had to dig out bits of it from her neck and chest!  Maybe this would be prudent advice: Do Not Use Tantalum Capacitors.  Replace that thing with a ceramic, film or electrolytic - whatever has a similar rating and can be pigeon-holed in those pads.  Perhaps Tangent can help with an alternative.

Just my personal opinion, but I think tantalums should be outlawed - certainly for DIY.  They're too dangerous.


P.S. Nice Rugrat, Sara!  I had two female versions of those, but it's been a while.    Anyway, good luck on the LNMP.  It's a great device and it looks like you built a great example of Tangent's design.

No worries, invade away. Actually, my left cheek caught a couple of bits as well - too small to dig out and produced pimples. Dudes, It was no fun at all. But I chose the tants on purpose, for both aesthetic and practical reasons. If they are good enough for military satellites, why not for me? Of course, that was long before I discovered that they can go flash-bang. But I'm not pulling 20 out now, I'm flying on the idea that was an uncommon phenomenon.

So, I'm interested in your conviction that the tantalum caps are dangerous.... ?

Thanks for the compliments, I have a couple of older daughters, too. I couldn't be more proud. This Christmas my oldest got an insane compound microscope from Amazon-Santa to go with her lab coat. (Mom gets to play with it too)

 

[saraengelstad]

We have so many different capacitor types because they each have their advantages and disadvantages. Typically the only thing that can replace a tantalum is an aluminum electrolytic, which is much more susceptible to heat, a serious consideration for that particular cap, being right next to a voltage regulator.

Cap types that aren't still useful typically get dropped from the manufacturers' lines. Tants are still being made despite the difficulty of getting the materials because they have some uncommon properties that makes them more useful in some areas than their closest alternative.

Tantalum capacitors have other properties that make replacement with an aluminum electrolytic difficult. The National Semiconductor LM317 datasheet says that you need about a 25 µF aluminum electrolytic to give equivalent performance to a 1 µF tantalum. If the problem here is overvoltage, though, that will also kill an aluminum electrolytic cap.

Fascinating. Thank you, Tangent. I started building these, wow, maybe seven years ago?

I did carefully consider all my parts, that's part of my art thing. It's weird, but part of my process. EVERY component I use is obsessively considered. I spent a fortune on MELF resistors from Germany for my DACs - they were unavailable here in the quantities I needed back then. I've spent thirty hours choosing a resistor and have even had some manufactured for my collection. But this decision was made so long ago that I had forgotten what alternatives I decided against using.

 

[tangent]

it is not unimaginable that I would repeat the same error 20 times

 
It's especially easy in this case, since solid tantalums typically have their positive lead marked, while alum elecs have their negative lead marked.
 
I've also heard that some LEDs are marked "backwards" compared to the standard way.
 
Bottom line, always check the data sheet.

 

[saraengelstad]

You're kidding. Well that sounds like just about every other 'standard' I've dealt with.

 

[dsavitsk]

I've also heard that some LEDs are marked "backwards" compared to the standard way.

This is definitely the case for surface mount LEDs. For instance, if you look on the bottom of page 2 of this datasheet which covers several devices, there is a note saying "Polarity for HSMH-Cxxx will be the opposite of what is shown on above drawings."

 

[wakibaki]

I've seen products massaged through test with copper tape. All the test houses have large rolls of the stuff, it has conductive glue. Of course any modifications have to be carried through into the product.
 
The problem is gauging the effectiveness of what you've done. In a test you can bombard the object with high power across a range of frequencies with various modulations.
 
There are loads of proprietary products for the control of RF for boxes, sprung copper fingers are common to connect lids to enclosures. You can get some nice little tinned boxes that solder to a PCB and use a layer as one face of the box.