On optimizing manual soldering

• Miscellaneous
• Optimization

They say that the difference between 'clever' and 'wise' is that a clever person can get out of situations which a wise person would have avoided in the first place. Being neither wise nor clever, apparently, I found myself facing the task of soldering, by hand, dozens of boards, with dozens of components on each. Yes, soldering is a relaxing activity, but I don't need to relax that much, and I'd rather finish this job as fast - and therefore, as profitably - as possible.

To clarify: PCBA and reflow-at-home, for various reasons, were not an option. Neither was delegating the work to someone else. The components were mostly SMT, and for quality reasons, most of the job had to be done under a microscope. I've been doing such work for almost ten years now, so I had some idea about the time it should take, and I didn’t like the prospects; so, I got a coach-style stopwatch, took a statistically significant pile of boards and started measuring.

The idea is to isolate and measure batches of distinct, meaningful "sub-jobs". For example: T_R is the time it takes to solder all red LEDs (one per board); T_B is the time needed to solder all blue LEDs (two per board); 2T_R – T_B is therefore the time saved by soldering LEDs two at a time, and knowing that, I can start thinking where this saved time comes from: for the second blue LED, the board is already in place.

In this fashion I got some useful information regarding the duration and breakdown of common sub-tasks: rotating boards to comfortable angles, placing polarized vs. non-polarized components, the "marginal utility" of soldering additional components of the same, or different, kind each time, the effect of the number of pads, etc. I will not give the numbers here – not because they're secret but because they're specific to me. Instead, I'll present the general conclusions that other solder jockeys could use.

When things go well: minimizing context switching

All happy families, wrote Tolstoy, are alike; each unhappy family is unhappy in its own way. The same applies, with some imagination, to bulk manual soldering: Once you have the basic skills, and while everything goes according to plan, there's surprisingly little to optimize. The interesting stuff and the major optimization opportunities are related to problems and mistakes. But let's suspend the real world until the next section, and imagine that my work is perfect: I never drop anything, my hands don't shake a little, solder only goes exactly where I want it, etc. Under these favorable conditions, what's the best approach for optimizing the task?

Abstractly speaking, the job consists of many different subtasks which are done in different contexts. Some are obvious – for instance, setting up the components on the desk vs. doing solder joints – and some are more subtle, such as holding tweezers vs. holding the soldering wire. As in firmware, context switching takes time, so the less we switch, the better. That’s why placing a few adjacent components on the board and then soldering them is faster than placing and soldering each component in turn. Optimizing soldering work is, therefore, about finding your context switches, thinking up ways to accelerate them, and eliminating as many of them as possible.

Here's an example. I solder an SMT component by, first, wetting one pad; then I tack the component to that pad, and complete the soldering of the other pads. This requires context switching between tweezers and soldering wire, which is quite uncomfortable when working under a microscope (you have to either look away or grope each time to find the tweezers/wire!) So I figured, why not wet all those anchor pads in one go under the microscope, and then do the rest of the work? With enough components that have low pin/lead count, this idea alone reduced the overall work time by around 6% (YMMV).

Overall, optimizations of this kind reduced my work time by almost 20%.

When things go wrong: the cost of mistakes

Let’s return now to the real world. Every once in a while I’ll be putting too much, or too little, pressure on the tweezers, and a component will fall or fly off. If it’s a cheap resistor, who cares – I’ll pick another one. But with more expensive components, this becomes a problem: sooner or later I’ll have to find it. This alone, let me assure you, can take more time than soldering two or three components, and it also disrupts the work flow.

Or, let’s say I’m in the optimization mindset, and after finishing 20 boards I decide that the next batch can be done faster if I put them in a different orientation under the microscope – but then, out of habit, I solder a couple of LEDs in what’s now reverse polarity, and I have to go back, find and fix all these mistakes. This will easily wipe out any benefit I might have gained from the optimization. Ask me how I know.

The important lesson is that before we try to make things better, we should ensure that we’re not making them worse. Prepare a comfortable, well-lit work environment with minimal distractions; use high quality tools and materials; set up a workflow and stick to it. There may still be long hours of soldering ahead, but they won't be so painful.

Did I miss anything? Can we (and should we bother to) make manual soldering even easier?

Sign in