Notes on Soldering at Three in the Morning

It is, at the moment of writing, three minutes past three in the morning. The iron on my bench has been at 350 degrees for the last four hours, and my apartment smells, faintly, of rosin. The board I am working on — a small, dumb thing, sixteen components, no microcontroller, just an op-amp and some passives, the sort of board a competent third-year would knock out in an afternoon — has defeated me, comprehensively, three times since dinner.

The defeats are entirely my own. I rushed the first one and put the chip in backwards. I rushed the second one, after lifting the chip, and tore a pad clean off the substrate. I rushed the third one, on a fresh board, by leaving the iron in contact with the QFN package for the better part of five seconds because I was distracted by a notification from my phone. The chip is, I am almost certain, dead. The board, in any case, is not going to work tonight.

I am writing this because there is a particular kind of lesson in bench work that I think is worth getting down, and the lesson is most legible at three in the morning, after the third failure, when the temptation to walk away is at its strongest. The lesson is: the discipline of small, careful work is not natural. It is a thing you have to relearn, slightly, every time you sit down.

The geometry of a joint

A good solder joint, on a through-hole pad, looks like a small, concave triangle. The fillet rises smoothly from the pad up to the lead, the surface is shiny, and the angle the solder makes with the pad is somewhere between thirty and forty-five degrees. A bad joint looks like a ball. A really bad joint looks like a ball with a small horn on top, where the iron pulled away too fast and dragged a strand of solder with it.

I have read this description in maybe a dozen books, and I have given it, in slightly different words, to four cohorts of workshop students. I can recite the temperature ranges from memory. I know that 327 degrees Celsius is the eutectic point of 63/37 lead-tin and 217 degrees is the liquidus of SAC305. I know that you tin the iron, then the pad, then the lead, in that order, and that the joint takes between two and three seconds and not four. I know all of this, and I have known it for ten years.

What I had forgotten, until tonight, is that knowing it is not enough. The hand has to know it too, and the hand forgets, the way the hand forgets a piece of piano music if you do not play it for six months. There is a particular wrist motion, a kind of small clockwise rotation at the moment of withdrawal, that you can describe in words but cannot transmit. You can only learn it by doing it a thousand times, and you can only keep it by doing it every few weeks.

The board on the bench in front of me was telling me, all evening, that my hand had forgotten the motion. Each joint was a degree too convex, a degree too dull, a degree too cold. I should have stopped and practised on a scrap board. I did not, because I had told myself I would have the board working before bed, and because I was tired, and because the iron was already hot.

A small taxonomy of nighttime failures

There is a kind of failure that comes from not knowing. You sit at the bench, you read the datasheet, you misread the pinout, you put the chip in backwards. This is the failure of the apprentice, and it has a remedy: read more carefully.

There is a kind of failure that comes from knowing imperfectly. You sit at the bench, you know the pinout, you know the iron is too hot for QFN packages, you do it anyway because last time it worked. This is the failure of the journeyman, and it has a remedy: do not.

There is a third kind of failure that comes from knowing perfectly and being too tired to act on what you know. You sit at the bench, you can see, in your hand, that the joint is going wrong, and you finish it anyway because finishing feels like progress. This is the failure of the master, in the precise sense that the only way to make it is to have first mastered the underlying skill. It has a remedy too, and the remedy is to go to sleep.

I am, on the evidence of tonight, somewhere between journeyman and master. The board would tell you the same.

On putting it down

There is a small, embarrassing literature, mostly published in the trade magazines of the 1970s and 1980s, on the dangers of working tired. The articles read like safety posters: photographs of workshops, of benches, of hands holding tools, with captions reminding the reader that the human eye loses contrast sensitivity after midnight, that fine-motor control degrades faster than gross-motor control, that the temptation to "just finish this one thing" is the most reliable predictor of an emergency-room visit in the entire trade.

I used to find these articles charmingly old-fashioned. I do not, any more. There is a reason workshop foremen, in serious shops, used to send people home at midnight regardless of what they were in the middle of. The reason is that the person who insists on finishing the joint at one in the morning is the same person who, at two in the morning, lifts a pad, and at three in the morning sits on a stool and writes an essay about it.

A note on the chip

The QFN, when I take a loupe to it, has a small discolouration on the underside that I think corresponds to a delaminated die. I will know in the morning, when I put it on the scope. If I am right, the chip is a $1.40 loss, and the board is a $4 loss, and the evening is a six-hour loss, and the entire failure has cost me, in the most literal accounting, less than my dinner.

In the less literal accounting, it has cost me the small, prideful conviction that I am, at this point in my life, past the kind of mistakes that students make. I am not. The hand forgets. The discipline of careful work, of three-second joints and one-task-at-a-time evenings, is not something you acquire and keep. It is something you keep acquiring, every time you sit down.

The iron is cooling. The room smells of rosin. I am going to bed.