So, there I am… sat at the workbench, head buried in my latest creation when I start to get a feint smell of something. I know that smell. It’s the smell of burning components. My immediate though was the project I was working on was about to go up in smoke, but it wasn’t powered on at the time. A quick glance around shows the room definitely has a smoke haze forming and I quickly trace it to my spectrum analyser. I switched it off, cleared a space on my bench and whipped off the lid to see what was going on. As it happens, it’s been playing up for a while now and I’ve been meaning to get around to having a look under the hood as it were. I was really worried as this piece of kit has three 6v SLA batteries inside and I thought if there was a short circuit, it’s not a quick job to remove them.

A close-up inspection revealed the culprit – a tantalum capacitor on a small voltage regulator board (front left and black).

For me, tantalum capacitors are a complete pain in the ass. They always seem to fail by going short circuit and produce the most horrible smell. They can also cause a LOT of damage to the surrounding PCB if you don’t catch them in time.

Anyway, PCB removed, both capacitors exchanged for standard electrolytics and everything now works a treat.

If you have ever wondered what happens if you don't cut the power in time, the pictures below are from the battery backup board in my Nixdorf 8870 mini computer. "ONE" tantalum failing caused all this damage.

I never did manage to get rid of the smell completely.

Personally, I don’t know why tantalum capacitors don’t have an in-built piece of fuse wire !!!

A friend asked me if I would have a look at his Thurlby bench power supply. It has three separate outputs, one switchable between 3 and 5v and two variable voltage also with variable current limiting. It’s quite a nice piece of kit even though it’s several years old now. Unfortunately, one of the variable outputs had stopped working.

Under the lid is a nicely arranged SMPS (Switched Mode Power Supply) and it didn’t take too long to identify the main PSU blocks and fortunately, the two variable output blocks are basically identical.

After removing the layer of dust that had settled on the PCB topside, I started with some of the obvious checks, looking for damaged, burnt or bulging components but there was nothing obvious.

Removing the PCB from the chassis and an underside inspection looking for broken tracks and dry-joints showed nothing obvious either.

It was only when I was poking at things with my screwdriver tip (I do this to check for things that are loose) that I saw there was a wire broken off from one of the main inductors; looked like a dry-joint had failed. I re-soldered that and powered on. Now the retched PSU refused to start at all.

Using my lovely Atlas testers from PEAK, I checked diodes and resistors. If readings looked odd, I would check the same component on the other channel; the Atlas testers are pretty good but obviously can be confused when testing components in-circuit.

Next, I removed and tested all the silicone devices from the faulty channel but still no faults were found. I then did the same for the three larger electrolytic caps and again, no fault was found but on powering up, the fault seemed to have rectified itself. I can only assume that in the process of removing one of the electrolytic capacitors, testing and then re-soldering it, I fixed a dry-joint I hadn’t spotted.

Now I’ve fallen for this once before.

My new rule of thumb is when presented with a fault; ALWAYS re-solder the joints just to make sure. This is doubly important now my eyesight isn’t as good as it used to be :(