I have a LED-ring PCB, relatively new, that we sent through a few temperature cycle tests (-50 to +85*C), and it now shows signs of the solder-mask coming off some of the copper traces.

The LEDs weren't powered during the temperature cycle tests, and the tests were not oriented at the PCB but rather the weatherproofing of its enclosure. The PCB is mounted in an aluminum+glass enclosure (aluminum acting as heatsink) to be installed outdoors in the ground, so it'll be exposed to harsh temperatures.

The images show the slight damage to the traces on the bottom and the top of the PCB. A multimeter test confirms that the copper trace is exposed on some of the positions.

Soldermask damage on bottom of PCB Soldermask damage on top of PCB

I am trying to understand if the soldermask coming off is

  • a natural occurrence when dealing with PCBs getting hot or cold?
  • a case of a low-quality soldermask (ordered from seeedstudio, who as far as I can see don't offer any soldermask options)?
  • a sign of too-high-current on too slim traces?

If it's the first (typical temperature issue) would a conformal coating help to just keep the traces insulated (they do sit on top of aluminum)?

PCB specs
Layers: 2
Material: FR4 TG130
PCB Thickness: 0.8mm
Power track width: 0.4mm (max 1A current)
Default track width: 0.25mm (max 0.25A current)


1 Answer 1


I don't think it's as much of the soldermasks fault as the thermal coefficient of expansion between copper an aluminum, and the large PCB and features.

For aluminum its 21 - 24.0 (10-6 m/(m °C))
For copper its 16 - 16.7 (10-6 m/(m °C))

a aluminum feature that is 5cm will expand about 1.2um/C
a copper feature that is 5cm will expand about 8.35e-6m/C

Which doesn't sound like much, but if the temperature range was expanded to 100C (which means something like -20C to 80C or -40C to 40C) it would amount to ~0.7 mm or about ~3mil (which is probably around what is shown in the pictures above).

It doesn't really matter what you put on that PCB, those traces are probably going to bulldoze it right out of the way. A better way would be to design the traces to facilitate the expansion. Unforntunatley this is more of the relm of mechanical engineers (at least at this point I would turn the problem over to someone else in my group and let them run an FEM) they may have better suggestions of how to deal with this problem). If I were facing this problem myself I might try building in 2D 'springs' that could handle the stretching across the face of the pcb. Another option would be to contact the manufacturer as they likely have resources to deal with this (or have seen it before).

Another problem that you may face is eventual delamination of the trace from the PCB if this stretching continues, it depends on what the prepreg/adhesive layer (whatever the manufacturer is using) can handle strain.

  • \$\begingroup\$ Very interesting! Thanks for all the details. One question: You mention the expansion rate of Aluminum. But the PCB doesn't have an aluminum base, rather a standard fiberglass-laminate base. It is only mounted onto an aluminum fixture with screws. The copper traces popping out comes from the copper expanding in relation to the fiberglass-laminate base? Or it expanding in relation to the aluminum fixture it is mounted on? \$\endgroup\$
    – evsc
    Jun 29, 2021 at 18:33
  • 1
    \$\begingroup\$ All three will expand and contract with different rates, so you would find the thermal coefficient of expansion for that material and find how far it will expand or contract. If it is in between the copper and aluminum that could facilitate some 'wiggle room' to take strain out of the system. \$\endgroup\$
    – Voltage Spike
    Jun 29, 2021 at 19:03
  • \$\begingroup\$ meta.stackexchange.com/questions/126180/… \$\endgroup\$
    – Voltage Spike
    Jun 29, 2021 at 19:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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