Ramblings of an aging IT geek
← Ramblings of an aging IT geek
hardware

reviving a doorbell transformer board and learning to reflow

A second repair over the break, a board with surface-mount damage that taught me to stop fearing reflow and hot air.

A circuit board with surface-mount components

The first repair this week went smoothly because it was through-hole: big legs, big pads, an iron and a steady hand. This one did not, and I'm glad, because it pushed me into territory I'd been quietly avoiding for years. Surface-mount. The board was the controller out of a wireless doorbell receiver, and it had taken a knock that lifted a corner of a small QFN chip and, I'd later find, cracked a couple of joints under it. You cannot fix that with a fat iron tip and good intentions. I'd been putting off learning hot-air rework. The doorbell forced the issue.

Why SMD scares people, and why it shouldn't

The fear is mostly about scale. The parts are tiny, the pads are tiny, the pitch between pins can be less than a millimetre, and the first time you see a leadless package with the contacts hidden underneath it your instinct is that this is factory work, not bench work. That instinct is wrong, but it's an understandable place to start.

The patient, with the offending chip top left

What changed my mind was understanding that you're not really soldering each tiny joint by hand at all. You're managing heat across a whole region and letting surface tension do the precision for you. Molten solder wants to wet to clean copper and pull a part square onto its pads. Your job is mostly to get everything clean, get the right amount of solder and flux in place, and bring the whole lot up to temperature evenly. The chip will, given half a chance, drag itself into alignment. That's the bit nobody tells you and the bit that makes it possible.

The kit that actually matters

You don't need a reflow oven and a microscope to start. I didn't have either. What you genuinely need:

  • Hot-air rework station. This is the one real purchase. I'd bought a cheap combined iron-and-hot-air unit months ago and barely used the air side. Set somewhere around 350°C with moderate airflow for leaded solder.
  • Flux. More than you think. A good no-clean flux is the single biggest difference between a clean reflow and a bridged mess. It keeps the solder mobile and helps it flow where it should.
  • Solder wick. For pulling off excess and clearing bridges. Indispensable.
  • Tweezers and a steady surface. Fine-tipped tweezers to place parts, and something to brace your hands against.
  • Isopropyl and a brush for cleaning up afterwards.

That's it. The total spend over an iron you already own is one hot-air station, and the cheap ones are fine to learn on.

Doing the reflow

The chip was a small QFN with thermal pad underneath, which is about the least forgiving thing to start on, so naturally that's what I had. I'll describe what I did rather than pretend it was elegant.

First, flood the area with flux. Generously. Then bring the hot air in, kept moving in small circles over the chip, never parked in one spot. You're heating the whole package and its pads together, not blasting a corner. After fifteen or twenty seconds you watch for the moment the solder goes liquid, and you can see it: the chip settles, it shifts very slightly, and if it was sitting proud it drops flat onto the board. A gentle nudge with the tweezers and surface tension snaps it square. That self-alignment is genuinely satisfying to watch the first time.

station: ~350°C, medium air
motion:  small circles, keep moving, never park
watch:   for the chip to "settle" as solder liquefies
nudge:   tweezers, gently, let surface tension finish

The trap is impatience. Yank the heat away too soon and you get a cold joint hidden under a chip you can't inspect, which is the worst possible outcome. Leave the heat on too long and you cook the part or the board. You learn the window by feel faster than you'd expect, and a cheap thermocouple helps if you're nervous.

Inspecting what you can't see

The honest problem with leadless packages is that the joints are underneath. You can't put a loupe on them. So you verify indirectly. I checked continuity from each accessible point on that net through to where it should land, ran a careful eye round the edge for any bridge the flux had left, and then, the real test, powered it up.

The verdict

It worked. The receiver paired again, the chime came back, and a board I'd have thrown away now sits in the hall doing its small job. More than the fix, though, I came away with a skill I'd been ducking for years. Surface-mount rework turned out not to be the dark art I'd built it up to be in my head. It's hot air, flux, patience, and a willingness to let physics do the fiddly part. I'll stop binning SMD boards now, which means I'll start fixing a great many more things than I used to. That's the real return on an afternoon and a cheap hot-air gun: not one repaired doorbell, but every future repair I'd previously have written off as beyond me.