Learning about reflow soldering profiles the hard way...
Sorry for the lack of photos in this article, but we didn't think about it being worth reporting until after it had all happened. Infact, we were concerned about retaining any evidence of the events. Inspired by the many toaster-oven reflow projects floating around the Internet, we set out to acquire the power to solder all those tiny SMD components ourselves. Toaster ovens aren't that common in the UK and we were aiming for the lowest cost possible, so we hit ebay in search of a bargain. A very cheap Black and Decker Toast-R-Oven was on offer. The description was "only used a few times", and the grainy photos showed an indeterminate "foreign" plug with an adaptor. Now, we aren't generally in the habit of assuming that "some bloke on ebay" must know what he's doing with electricals, but with the auction starting at £5 (and not going up a huge amount from there by the end) it seemed like a good idea.
WARNING: the following paragraph contains extreme stupidity and highly dubious justifications. We do not recommend us trying it, never mind you!
Of course, there was a reason for the discount. Its a US device. The label on the bottom says "110V". Technically, it doesn't say "only" but that is pretty much implied by the manual that we were able to find online. The fact that we are even debating the implicit "only" should tell you just how dubious was the thinking that followed. Rather than do the sensible thing and put it down to ebay experience and move on, we contemplated the implications. As far as could be seen the oven is sufficiently cheap that all the controls are mechanical. The thermostat makes a nice, mechanical thermostat "click"; the timer is clockwork - to the extent that it has a label saying "wind past 20mins and then back to your desired time"; the heating element selector is a big, clunky four position switch. The heating elements are just wired through these mechanical devices. Assuming that the wiring and connectors are all safe to 240V there seemed no reason why it couldn't work on 240V, it would just get hot faster, but the mechanical thermostat should still switch it off. Besides, the bloke on ebay says he had used it "a couple of times".
Eventually, there was nothing left to do but plug it in and try it. Well, we could have put it down to ebay experience and moved on, but Who Dares Wins... The most likely failure was that the elements would just burn out, so we resolved to take it slowly. We set it to the "warm" setting, set the thermostat to a nice low 150F (65C) and turned the timer on. There was a nice warm red glow and no flashes or bangs or sparks. We watched. It glowed. After maybe 20sec the thermostat clicked and the elements went dark. So far, so good. Next, we tried the higher settings and verified that they just selected different combinations of elements. What was apparent was that on the other settings the elements received more power since the red glow became somewhere between orange and white. Toaster ovens really aren't that common in the UK so we have no idea if this is normal or if this indicates that they are running way over their design spec. Either way, they didn't melt and destroy themselves. At this stage we decided that the oven probably wouldn't destroy itself, so it was time for a proper test.
We have some PCBs and SMD components sat around that would have been hand soldered, so one was "volunteered" for testing the oven. Our pasting skills could use some work, and the vacuum pen that we had turned out to be far too big for the smaller components, but everything ended up on the board and the board went into the oven. Reading the various reports of other peoples' reflow experiences had introduced the idea of "reflow profiles" where the board is brought up to a temperature where the flux is activated and is allowed to "soak" for a while to let the flux do its fluxing and let all the metal on the board warm up to a reasonably equal temperature. Then the profiles call for a quick foray up to solder-melting temperatures and a steady cool down. Some people seem to have gone to a lot of trouble to produce complicated control systems with thermocouples and the like, but that all seemed a little excessive. After all, we had a thermostat that went click, that should be fine, right? Also, some of the reports suggested that toaster ovens didn't heat up that fast anyway, so at least one reporter had just ignored the profile and heated stuff. His oven was probably running on the correct voltage, though.
So, board in the oven, oven set to 120C, power on. Nice glow, nice click from the thermostat, elements off, bit of cool down, click, elements on, its holding temperature at a soak temp. We watched. The paste looked a little runnier but not a great deal was happening. Has it had enough soak time yet? How long has it had? Dunno, two minutes ish? Ok, its got to be time to do this properly *crank*. The thermostat was set up to 230C for a nice reflow. We would watch it reflow, admire the much-youtubed self-alignment of the components, then crank it right down to let it cool off. Not a lot happened. We looked at it. We had a quick chat about what might or might not happen, then we looked back. It seemed to be soldered. It was all shiny and located nicely. Look at that bit there, its aligned itself nicely...hey, is that bit bubbling..OH CRAP.
At this point the entire PCB began to bubble and drip (onto the heating elements). The liquid turned black and released black smoke in the time it took us to rapidly crank the thermostat down and rip the plug out of the wall. At this point we discovered that "having the window open a bit" does not constitute a well ventilated area... The window was flung open and a large, nearby fan was directed to disperse the smoke. Luckily, the "FR" in "FR-4" stands for "flame retardant", so there was no serious fire. However, the smell was somewhat noticeable and the sound of "hey, is something burning?" rippled its way down the corridor.
The result of all of this was a very scorched board and a very smelly office (which still persists two weeks later as I write this). Two explanations are possible for the problem. It seems that the "glass transition temperature" of the PCB material is only in the 140-150C range, and it could be that the reason all those reflow profiles spike in such a controlled manner is to do the soldering before you destroy the board. Alternatively, our 110V elements running at double their rated voltage may be heating things up way faster than the thermostat can handle - especially since its just a mechanical device in the corner of the enclosure - so it may be that we actually reached a much higher temperature than any reflow profile suggests.
The sensible conclusion to this is: don't use 110V devices on 240V. Ever. Obviously, our dubious reasoning leads to a different conclusion:we need to build one of those fancy reflow oven controllers to manage the duty cycle of the elements to keep them in check and to control the temperature more carefully. Look out for follow up posts describing the design and build of the control board. There is one obvious problem: how will we solder the control board?...