I’ve long wanted to try reflow soldering of surface mount components, using solder paste and hot air or a hot skillet, and I recently had an opportunity to try. Unfortunately the results were very poor, and I was left wondering what I did wrong. Compared to my normal method of drag-soldering surface mount chips with a standard iron and liberal amounts of flux, the reflow method with solder paste took more than twice as long and led to worse results overall.
My test case was the ROM-inator II SIMM, which has a couple of chips with a 0.5 mm pin spacing. With my normal soldering iron, I can tack down a few corner pins, drag solder the rest of the pins, and then use desoldering braid to clean up the inevitable solder bridges that sometimes occur between adjacent pins. It’s something of a slow and tedious method, but it works.
For this reflow experiment, I used a syringe of MG Chemicals Leaded Solder Paste. The needle on the MG syringe is quite large, and the bead of solder paste it dispenses is far too wide for detailed surface mount work. I dispensed about 0.1 mL of the solder paste into a second 1 mL syringe with a 22 gauge needle, which allowed me to create a nice narrow line of paste with very little finger pressure on the syringe’s plunger.
Solder paste has a limited shelf life, and you’re supposed to keep it refrigerated to help preserve it. I believe the flux in the solder paste evaporates more slowly at low temperatures. I made a conscious choice not to refrigerate, accepting that the paste would spoil faster than normal, because I wasn’t excited about keeping leaded solder where I keep my food. For this test, the solder paste was just one week old, and should still have been quite fresh.
I didn’t use a PCB stencil, because I didn’t have one, and because I was skeptical I could hand-position a stencil to sub-millimeter accuracy anyway. I also wasn’t excited about the mess created by squeegeeing solder paste across a stencil. Instead, following some advice from a tutorial, I used the syringe to dispense a very narrow bead of solder paste straight down the line of pads on the PCB. While it might seem that this would cause all the pins to be soldered to their neighbors instead of to the pads, the magic of flux and solder surface tension should have made it work OK.
Using tweezers, I positioned the chips on the pads, aligning them as accurately as I could. The MG solder paste had a nice amount of stickiness to it, holding each chip steadily in place after I’d positioned it. That magic surface tension of solder should have helped here too, causing the chip to “snap” into perfect alignment with the pads once the solder began to melt.
I used a hot air rework station to heat the pads, and before long the solder paste melted into a silvery and shiny color. I let it go for a few more seconds to ensure everything was fully melted, then removed the hot air and let the board cool.
Despite having put down a very narrow bead of paste, many of the pins were bridged together. In this respect it was as bad or worse than my hand-soldering efforts, which was disappointing. But while some pins became bridged to their neighbors, other pins had too little solder, or looked completely dry. Overall I’d say the row of pins looked like it had too little solder rather than too much, so I don’t think the bridged pins were due to applying too large a bead of solder paste.
A second and more serious problem was the positioning of the chip. When hand soldering with an iron I can usually align those 0.5 mm pins with their pads fairly accurately, but the gray solder paste made the task more difficult this time. After dispensing the solder paste and placing the chip, but before melting the solder, the row of pins looked like a silver and gray blur. It was very tough to see if it was positioned accurately. The result was that two of my chips were positioned with the pins misaligned far enough from the pads to create a mess I was never able to fix. It only takes about 0.2 mm of misalignment to create a major problem with pins this closely spaced.
Even when the chips were aligned properly, I still had to go back and forth over each row of pins with the iron and flux, to fix the solder bridges and flow more solder onto pads that had too little. In effect, I had to drag solder the chip after having already reflow soldered it. Definitely not an improvement over the old method.
I built four ROM-inator II SIMMS this way, refining my technique each time in the hopes that the results would improve, but they never really did. Two of the four SIMMs couldn’t be salvaged, and had to be scrapped.
The title photo above shows one of those reject SIMMs, with a poorly aligned chip, after I’d made about 50 passes with an iron, flux, and desoldering braid trying to clean it up. It’s a mess, which is why I like the photo, but don’t take the photo as being a literal example of the results immediately after reflowing. I should have thought to take some photos after the reflow process, but before I’d touched any pins with an iron. It looked much cleaner and nicer then, but was misaligned and had many solder bridges and many pads with low or no visible solder.
Why did this work so poorly? I don’t think it was the absence of a stencil. Attempting to position the chip smeared the solder paste around, so even if I’d had a stencil and used it perfectly, I still would have had smeared paste between the pads anyway.
I also don’t think my hand-dispensed solder paste bead was the problem. Using the 1 mL syringe, I was able to dispense a nice and even bead that was about as wide as a single pad, roughly 0.3 mm or so. Given the occasional low- and no-solder pads I observed, I probably had too little solder if anything, not too much.
Maybe 0.5 mm spaced pins are simply too small and close to be reflow soldered using this method? For comparison, I used the same reflow method to solder some 0805 sized SMD capacitors on this PCB, and it worked perfectly every time. No doubt larger sized components are easier to solder correctly, but I’m dubious that 0.5 mm pin spacing is beyond some threshold that can’t be reflow soldered using this method.
Maybe the temperature profile during reflow was far enough off to cause major problems? A real reflow process in a commercial oven would have a preheat phase about 4 minutes long, then an actual reflow time (above the solder’s melting point) of 60-90 seconds. My hot air method was much faster than that: point the hot air tool at the pads, wait maybe ~30 seconds for the solder paste to melt, wait another ~10 seconds to make sure it’s completely melted and reflowed everywhere, then remove the hot air.
Maybe the solder paste had already gone bad after a week at room temperature. That seems hard to believe, but it would explain the apparently poor wetting that I observed, with frequent solder bridges at the same time as other pins were dry, and the failure of the chip to snap into the proper alignment when the solder melted. The MG Chemicals syringe doesn’t have a date on it, but it’s marked with lot code 15-349 and was purchased new last week. If 15-349 means it was made in 2015 (on the 349th day perhaps), and the syringe has been stored at room temperature in an Amazon warehouse since then, that would certainly be a problem.
Maybe the way I heated the pads with the hot air gun caused a problem? This seems plausible. The heat was localized to a small area, so I would reflow the pins on one side of the chip, then reflow the other side 20 seconds later. Perhaps if the entire PCB had reached reflow temperature at the same time, using an oven or skillet, I might have had better results.
Any other good theories?Read 10 comments and join the conversation