I wondered that, too. I wonder whether he actually reflowed solder in the oven. Wikipedia gives a sample profile that heats to over 200 degrees Celsius for a minute or so. This uses around 170 degrees Celsius, which is used for preheating in the Wikipedia example. Also, a Google image search gives me reflow heats above 200 degrees for the 5 images I checked.
Also "The speed holes worked: The boot chime rang. The screen glowed. The fans blew." doesn't make sense to me. The extra holes may help and may even fix his problem, but I find it hard to believe that successful booting shows that. Yes, it is a 'little' early for a final verdict.
From what I know about heat treat ageing of metals something similar may be happening. Because he is pushing his system through above average temperature cycles, this causes microcrystalline changes in the solder. Ageing at even a temperature below reflow/melt temperatures can help restructure the grain.
If he didn't run his system at such a high temp and then eventually turn it off to cool so much then he might not have the problem to begin with. As a side note, personal experience building pcs for many years has shown me that pcs that sleep/get powered off don't last as long as ones that stay powered on 24/7.
Why do you doubt the temperature did not reflow the board?
When I solder electronics, I'm usually somewhere between 500F to 650F, but that's so things get hot enough quickly enough to not burn. I don't doubt a slow bake at 340F would do the same thing... some online googling shows people doing reflows with toaster ovens in the same temperature range (340F-450F).
It's likely the solder joints were not completely disconnected, but rather simply cracked. So heating them a bit would allow the crack to reconnect, at least for a short while (seems to explain his symptoms of repeated failures after extreme system temperatures during operation).
Is it not true that baking it makes any "cracking" in the soldering smooth out since this is similar to what is done to boards at the factory during construction?
Right. And the reason why it keeps breaking? Well, the solder on the board might not have any flux left, so it might have trouble flowing properly. There could be a cold joint, etc.
If it fails again, one possible next step would be to grab a multimeter and test the resistance across each component on the board. If the resistance is infinite, that might be your problem component.
There's more than resistors on logic board. Also, by even touching component leads with multimeter probes is sometimes enough to connect the joint back thus fooling you to think connection was ok.
Also, the problem might as well be one of the numerous chips in BGA packages whose connection points you can't even see, let alone- probe.
Don't you have to isolate components from the circuit to check resistance? Otherwise, you're looking also the other components look like the other arm(s) of a parallel circuit.
You know what, you're right. I saw a video of someone testing that a chip had been soldered on correctly with a multimeter, but he was testing for the lack of a connection. Testing for a connection would not be done this way.My bad.
Got a source for that? According to Wikipedia, 340F is not far from the melting point for solder. Depending on the solder used, it may have melted, or just gotten soft enough to heal any gaps or cracks.
The link you attached lists a few different lead-free solder alloys with the lowest melting at 412F(211C) well above 340F(171C). Eutectic tin/lead solder has a melting point of 361F(183C), also above 340F.
I'm not sure whether or not cracks can heal at temperatures below the melting point, but I would be surprised if that were possible.
So it turns out that home oven manufacturers don't care too much about precision when it comes to oven temperature (and it doesn't matter usually).
A high end home oven will easily swing between +- 25 degrees of the temperature that you have dialed in, and that's if it's well calibrated.
If he didn't dial in "340" quite right, and his oven was average, I can easily see it hitting temperatures of 412+, and that could also account for his varied results.
It's possible that he did reflow the board. Consumer ovens usually reach 250C, slightly more when in autoclean mode. Lead-free reflow happens above 250C (normally 255-275C).
The whole procedure is wild, though, and I wouldn't recommend trying it out. He was lucky that the board didn't get destroyed. Most SMD components can withstand temperatures above 250C for tens of seconds, not necessarily for nine minutes.
But he writes "and turned the oven up to 340 F". That's about 170 Celsius. The oven probably isn't very precise, but I doubt that's in the 250 Celsius range.
I'm guessing he put the board in and then started the oven which took a while to preheat. Industrial reflow heats up the board gradually based on the solder paste manufacturer's suggested heat profile or experimentation/trial and error (e.g. when dealing with small pitch BGAs). 250C for even a few minutes would do quite some damage even to the fire retardant laminate and smell horrible.
He stated he was using a temperature of 340F, which is roughly 170C. Would that temperature also be harmful to SMD components when exposed for several minutes?
Also "The speed holes worked: The boot chime rang. The screen glowed. The fans blew." doesn't make sense to me. The extra holes may help and may even fix his problem, but I find it hard to believe that successful booting shows that. Yes, it is a 'little' early for a final verdict.