Asleep at the wheel soldering

This is getting quite annoying. Are there no process guys around anymore?

My Handspring Visor died last summer. The problem, COLD SOLDER JOINTS combined with inadequate amounts of solder.

My Linksys 802.11b Ver 3 PCMCIA card died. Most people, me included would just buy I new one. However, trying to find a Ver 3 is a little difficult. I did finally get one in. So… its time to see what died on mine.

So, off comes the case. Its a pretty novel idea of using micro snaps to hold the cover on. They work really well, and after microscopic inspection, they are overkill. Most likely the front cover will never be removed. However, these snaps are well designed. I think the guys designed their tooling for ease of engineering… rather than lowest possible cost. Its something rare to see in consumer products.

Next the challenge was removing the metal cover. A little work with dental tools, and sure enough off it came, again a great design.

The bad news… I heard something fall out of the case. Upon further inspection, no I didn’t break anything… but a 0402 resistor on the antenna switch is now gone. The solder joints were cold, and a proper fillet was non-existant. In fact, other 0402 components looked pretty fishy. Some guy was asleep at the wheel either in process, or in inspection.

Otoh, some manufacturers look at assembly standards as giving them reigns to be sloppy. They shoot for the spec, not the center. The problem is gaussian distribution. If you are not centering, more than likely your process is going to be hosed, as you cannot control all variables. Now, some quality guys will take issue with this. It doesn’t fit there mold… but what they are missing is long term reliability. Process variations will occur, inspection and QC does not catch all variables. Marginal ones will show up as a product ages. This particular card was almost 2 years old. Decent assembly process control should not allow for stupid failures such as this.

Guys, you gotta learn to solder… and shoot for the center, not just the outside of the spec, if you want long term reliability. The costs of doing so are much lower than ticked off customers due to premature product death. So how to solder….
1. profile your process, the component manufacturers are dead serious with the profiles. Don’t assume, measure, its critical. Also, be sure to calibrate your sensors on your reflow and profiler. They are thermocouples, they are guaranteed to drift. Its a matter of physics. Also calibrate your controllers, they were built by someone somewhere. They will drift over time as well.
2. Run solder paste analysis religiously. Sure its a pita, but if you get a marginal viscosity, it will bite you.
3. Check your stencils, they do wear, they do get contaminated. Solder paste is not totally inert when it comes to stainless steel, add in the abrasive qualities, and change will occur.
4. Watch your date codes, they are there for a reason. If the parts are outdated, sell them on ebay.
5. Control your parts storage, as well as the bare boards. Temperature and humidty out of range can affect your parts even through the packaging.

Sorry for the lecture, but this stupidity has to end. Few people will repair things anymore… and when repair is needed due to process 101 issues, its frustrating.

Also, since we’re transitioning to lead free, watch out for high infant and long term failures in the latest and greatest consumer goods. In attention to process control in lead free will put many companies out of business.

So… the 0402 is gone, I could spend weeks trying to find it. The good news, its a diversity system. I’ll just copy the value from the other side and see if it comes back to life. Its been a few months since I last soldered a 0402, so it will be interesting.