3
\$\begingroup\$

I just ordered 5 boards from a popular Chinese PCB manufacturer and (for the first time ever w/ with this manufacturer) some SOD-323 diodes and their pads and adjacent tracks were peeled off out of the box (1 or 2 per board in 3/5 boards).

I think it's a QA/assembly issue, but I also think my pads may be too small.
So I started "stress testing" the boards and, sure enough, with a modest amount of force I could take apart the diodes.

I also tried to peel off the other components (just with my fingers, not tools).
I failed with the STM32 and the crystal, but I was successful with 0402 capacitors and resistors.

Pad size for the diodes was about 0.65mm x 0.55mm (one of Kicad's default sizes; not the smallest but not the largest).
Track width is 0.2mm.

  • So, is this normal with PCbs in general, and with SOD-323 diodes in particular?
  • What is the recommended pad size for something like a SOD-323 diode?
  • What's a reasonable level of "physical resilience" that a PCB is expected to have? ( I mean, I remember dropping my electronics more than time throughout my life...)
\$\endgroup\$
10
  • 13
    \$\begingroup\$ pictures or it didn't happen! \$\endgroup\$
    – fraxinus
    Commented Aug 14 at 12:34
  • \$\begingroup\$ When you were designing the PCB layout, did you check to make sure the pad sizes matched up with what was recommended in the datasheet? If the pads are too small, then it will take less stress to knock them off the PCB, although it shouldn't be as easy as you are describing \$\endgroup\$
    – MCG
    Commented Aug 14 at 12:46
  • 9
    \$\begingroup\$ Are you managing to peel the components off their pads, or are you peeling the component + pads off the PCB substrate? \$\endgroup\$
    – brhans
    Commented Aug 14 at 12:48
  • 2
    \$\begingroup\$ It's normal to have the pad at least 0.2mm or so larger all round than the component pad, so that the solder fillet can be visually inspected for a good joint. It's not normal to be able to manually peel copper pads off of a cold PCB, unless you have very strong thumbnails. \$\endgroup\$
    – Neil_UK
    Commented Aug 14 at 13:25
  • \$\begingroup\$ Suggestion: Reheat / melt the solder on the joints of a few components of the sorts that are failing. See if this changes there robustness. || Inquiring minds want to know who the manufacturer was. \$\endgroup\$
    – Russell McMahon
    Commented Aug 15 at 2:41

2 Answers 2

11
\$\begingroup\$

I'm not sure if you refer to the quality of the PCB or the quality of the soldering, or both.

So, is this normal with PCbs in general, and with SOD-323 diodes in particular?

A properly done solder joint provides both a good mechanical and electrical connection. So in regards of soldering, you should not be able to do that by hand unless you use a lot of force. As in, if you press firmly against the component with a finger it shouldn't move, but if you do something extreme like press for all that you are worth, you'll probably damage things even with a proper solder joint. Note that SOD323 isn't the most rugged of components - the legs are relatively small in comparison with the component body. It isn't meant to take up force - most components are not.

You should be able notice the culprit if the component comes off: is there a square pattern in the solder where the component pad used to sit? Then the cause was bad soldering, most likely caused by a cold joint, oxidated component legs or bad paste. Did you tear off traces or damaged the component itself? Then you simply used excessive force or maybe the PCB is bad.

Them "well-known Chinese manufacturer" tend to have diverse quality depending on mood and manufacturer. I was soldering one such board just the other day and noticed that the solder mask started to come off close to the pads just from heating the pad during normal hand soldering. That's a bad quality PCB.

Similarly, you shouldn't be able to easily peel off pads from the PCB given that they are attached to traces/vias/copper pour (single "dummy" pads not connected to anything tend to come off and that's no fault of the PCB manufacturer). Some boards used for volume production are of low quality and can for example not stand desoldering/component replacement well at all. Doing such is an easy informal way to test the PCB quality.


What is the recommended pad size for something like a SOD-323 diode?

This should be mentioned when reading the friendly datasheet, or otherwise from manufacturer general recommendations. Apparently 0.6 x 0.6mm is a common land pattern recommendation for reflow soldering. So 0.65 x 0.55 sounds reasonable.

Note the difference between pad size and land pattern - pad size is slightly smaller, in this case 0.45 x 0.40. You should CAD according to land pattern.

In addition there will be a recommendation for paste stencils, in this case 0.5 x 0.5mm. That's an important measurement for the PCBA assembly/whoever orders the stencils. Get this wrong and that could explain bad soldering with wrong amount of paste.


What's a reasonable level of "physical resilience" that a PCB is expected to have?

I assume you mean a mounted PCBA or otherwise I don't know the answer. Well, it depends... it is subjective. Some components like crystals, LEDs or other optics, inductors, potentiometers/trim caps etc are sensitive to physical damage and if you drop a board with any of those on-board it might be broken now. While resistors and other passives can usually take quite a beating.

If you expect a rough treatment of the board, there's various tricks like only place components in a direction that the board will not be expected to bend in, or better yet use through-hole components all over. In general we shouldn't mechanically place the PCBA in such a way that it will take the blow from dropping the product etc.

There's standards for shock and vibration depending on what the end application is - design accordingly.

\$\endgroup\$
6
\$\begingroup\$

In my experience it is relatively easy to pull extremely small pads off of cheaply made boards. They're only held on by glue and if the pad is tiny the glue has little area to work with. In the case of SOD-323, you've only got ~300 micron square of contact area being held on by highly cost-optimized (i.e. cheap) glue.

Adding traces makes them stronger, but a 200 micron by 20 micron thick piece of copper doesn't add that much strength or that much contact area.

Conversely they never should have shipped you already broken boards, so I would complain about that. IMO that is a separate issue from the general strength (or lack their of) of copper pads.

\$\endgroup\$

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