I've been in the habit of making vias on homemade pcb's by putting the leg of 1N4148 diode into a 0.5 mm hole, cutting it off to length and soldering both sides.

However, now my led controller broke and I traced the problem into a broken ground via. On closer inspection, the wire has separated from the solder:

Broken via

There was about 2 ohms of resistance between the wire and the ground plane where it connects. During operation, there has been about 1 ampere of current going through it for a few hours each day. The device internal temperature stays below 50°C at all times, and has always been stored and used indoors.

Are there any obvious things that I'm doing wrong? I doubt it is a cold joint, because a bit of solder remains also on the wire.

  • \$\begingroup\$ hard to tell with such low quality image and nothing known else, I would say its mechanical stress \$\endgroup\$ – PlasmaHH Aug 30 '17 at 20:00
  • \$\begingroup\$ Yeah, I don't have a better microscope. I don't think there has been any stress on that wire, it is a 1.6 mm thick FR4 PCB that's been sitting mostly untouched in an enclosure, and the wire is just 2mm long, going from one side of PCB to another. \$\endgroup\$ – jpa Aug 30 '17 at 20:03
  • \$\begingroup\$ Could be a mix of factors like metal types and solder type itself. \$\endgroup\$ – Chetan Bhargava Aug 30 '17 at 20:03
  • \$\begingroup\$ Solder is leaded, eutectic Sn63/Pb37 solder. According to datasheet, 1N4148 leads are copper clad steel, with tin coating. \$\endgroup\$ – jpa Aug 30 '17 at 20:06
  • 2
    \$\begingroup\$ You say it's not likely a cold joint, because of the bit of solder remaining on the wire. But at the same time, it's clear the rest of the fillet didn't wet to the wire, or get drawn into the hole. So I'm going to say, actually a cold joint is at least part of the problem. \$\endgroup\$ – The Photon Aug 30 '17 at 20:11

Looks like a cold joint to me- the slightly crystalline appearance, for example (though it's hard to tell with all that flux on it). It was probably fractured from the beginning and corroded a bit so it stopped working.

You can avoid this by preventing motion as the joint cools. You can catch such defects by inspecting the board with a 10:1 magnifier, and to some degree, by burn-in at elevated temperature.

  • 1
    \$\begingroup\$ Hmm yeah, on a second thought this is probably it. The joint being on a ground plane would make it harder to heat properly. \$\endgroup\$ – jpa Aug 31 '17 at 5:29
  • \$\begingroup\$ Maybe use thermal reliefs. They're better suppressed if the connections won't be soldered, but that not the situation with homemade PCBs. \$\endgroup\$ – Spehro Pefhany Aug 31 '17 at 5:38

I know that annular ring failures are common with any strain, mass and vibration and worse with crystalline solder joints. Relay THT solder joints are common in stoves, flyback transformers on large Apple monitor mobos etc etc. So even the major players have these quality issues.

Corrective Actions by design and process improvement.

  • Strain relief with rigid polyurethane , to prevent micro-vibration motion of attached parts.

  • improved solder process with adequate heat source with a quick response time rather than slow for crystalline growth which is structurally weaker. ie cold solder joint

  • verify board flex does not aggrevate vibration from resonance.

Looks like it has got a thump one second before the joint had cooled enough to stand that thump. Another possiblity is vibration. Sometimes a coil vibrates enough to break its own joints, but this is not a component that generates vibrations. Check, are there any other sources.


It's should be the problem with your solder wire or solder flux. Any ordinary solder wire or flux will make brittle solder joints, when vibration happen, it breaks!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.