We’re used to reflow soldering of our PCBs at the hacker level, for quite a few years people have been reflowing with toaster ovens, skillets, and similar pieces of domestic equipment and equipping them with temperature controllers and timers. We take one or two boards, screen print a layer of solder paste on the pads by using a stencil, and place our surface-mount components with a pair of tweezers before putting them in the oven. It’s a process that requires care and attention, but it’s fairly straightforward once mastered and we can create small runs of high quality boards.
But what about the same process at a professional level, what do you do when your board isn’t a matchbox-sized panel from OSH Park with less than 50 or so parts but a densely-packed multilayer board about the size of a small tablet computer and with many hundreds of parts? In theory the same process of screen print and pick and place applies, but in practice to achieve a succesful result a lot more care and planning has to go into the process.
This is being written the morning after a marathon session encompassing all of the working day and half of the night. I was hand-stuffing a row of large high-density boards with components ranging from 0402 passives to large QFPs and everything else in between. I can’t describe the board in question because it is a commercially sensitive prototype for the industrial customer of the friend I was putting in the day’s work for, but it’s worth going through the minutiae of successfully assembling a small batch of prototypes at this level. Apologies then, any pictures will be rather generic.
Some of you reading this will now be asking “What on earth are you doing making this run of boards by hand, you should be doing it with a pick-and-place machine, or you should be hiring a specialist company!”. The answer to that isn’t really mine to give as the boards weren’t commissioned by me, but in reality it’s a nuanced decision based on a combination of cost, number of boards, and the eventual customer’s deadline for a trade show. Setting up a pick-and-place for a very large job is a performance in itself, and for a very small run of boards there is a hard financial decision to make over whether it is justified.
So there we were, setting out to make a batch of eight prototype PCBs. The story didn’t start on the build day, instead a few weeks ago the Bill Of Materials, or BoM, was exported from the CAD package, and the task of sourcing all the components began.
It stands to reason that the complexity of component sourcing increases with the number of individual component lines in the BoM. If your design consists entirely of generic components that every supplier has by the reel then sourcing is as simple as making the order, but sadly very few real designs are like that. So this step became an involved trawl through an array of suppliers for the elusive parts, sometimes ringing company reps to beg a few free samples.
The Great Gathering of Components
In the days running up to the build, a variety of packages arrived containing the components. There began the second major task, that of collation. It’s necessary to both ensure that everything has arrived and is the right component for the job, and to index and array them in a form such as to make the placement on the boards as easy as possible.
We started with a storage box of the type designed to hold hanging files. Each hanging file was labeled with a range of numbers corresponding to BoM lines, so 1 – 5, 6 -10, 11 – 15, and so on to the end of the BoM. Every line on the BoM spreadsheet was checked, that the component was present, was it compatible with the package it should be on the board, and was it present in sufficient numbers to populate the boards. Its line number from the spreadsheet was written on the label, and the spreadsheet was updated to show that it was present. The numbered bag of components was then placed in the appropriate hanging file for its line number, and the process was repeated with the next line number and so on until the whole BoM was covered.
At the end of the component collation, we had a box of hanging files containing numbered component lines, and inevitably there were a few lines which either weren’t quite right or hadn’t arrived. Some last-minute overnight ordering was in order, followed by collation steps for those parts.
It might seem like a lot of work to put in before making any boards, but this couple of days getting everything in a row will save you time when it comes to populating the boards, and will in turn result in better quality final prototypes.
On the day of the build, our components were all collated and we were handed a stack of bare PCBs from the board house. We set up adjacent workstations for the two of us, and set to work.
Become a Paste Aficionado
The first task when populating boards is to screen print the solder paste on the pads. We used the jig supplied by the board house for this task, with locating pegs fitted to align both board and stencil, and extra pieces of scrap PCB material taped down to support the stencil beyond the edge of the board.
Screen printing solder paste is in principle quite simple. Align the stencil with the pads, and using a scraper spread a layer of solder paste over all the holes. When you lift the stencil away the board should then be left with a uniform layer of solder paste on each pad, ready to have a component placed on it.
Describing this crucial step in those terms makes screen printing solder paste sound so easy, but of course it isn’t. The paste consistency is very important for a large board in the way it isn’t for a small one. The paste you print on the pads will spread out over time, and eventually your closely spaced tiny pads will be completely obscured by an amorphous blob of paste. If you have a small board you can get away with it, but on a large board it’s important to ensure that the spread is not too quick. You’ll then have a chance to place your components and reflow the board while the printed solder is still well-defined.
It’s best to describe the optimum solder paste consistency in terms of smooth peanut butter. Think of the perfect-spreading peanut butter, it’s not the runny stuff at the top of the jar where the oil has started to separate out, neither is it the stuff at the bottom of a jar that’s stood in the fridge for six weeks and goes all clumpy as you spread it on your toast. It’s the easy-spreading middle of a new jar, keeping its consistency and just about right for your knife. And so it is with solder paste, for a large board you need to very carefully ensure that your paste is well-mixed, and not dried out. Too runny and it will quickly spread out, while too thick and it may clump in the stencil and not stick to your pads. It’s something you gain a feel for with a bit of experience.
With the jig, stencil, scraper, and boards ready, and the correct consistency of paste to hand, there is one more step before spreading. Clean everything with IPA solvent; every single PCB, stencil, board, jig, the entire lot. It seems tedious, but it will make the difference between good and poor results. All drying paste residues, dirt, and oils can affect the quality of the job, and you need the best possible result.
After all that effort, the paste scraping itself is very quick. Make sure you have enough paste on your scraper, draw it across the stencil in one go covering all the holes with a firm pressure. Lift the stencil away, and inspect the quality of your paste printing. Don’t be afraid to clean it off and do it again if you aren’t happy with the results, we redid a couple of our board run.
Knowing Your Place
Towards the end of the morning then on our build day we had a row of boards ready screen printed with solder paste. They were lined up in front of my friend at one workstation, while at the other I had the box of parts and the BoM both on the computer and as a print-out. We both had large-scale print-outs of the component layout, and my job was to supply the tape of each component line in turn with the plastic strip ready lifted at the end, while he placed the components using a vacuum pencil and a magnifier. He has better dexterity for those 0402s than I do. The task of spotting component positions was shared, a game of Where’s Wally/Waldo on paper and computer, but without the stripey jumper.
Once you have started stuffing boards you are in a race against the spread of your solder paste, so there is no letting up until the task is complete. In our case the whole process took us all day and well into the evening, working through the BoM omitting any large components that would obstruct access for smaller one, then returning to fit them in a second pass. There followed a checking step during which a few inevitable omissions were rectified, no matter how hard you try there will be a few that you miss. Through all this process all that work collating components came into its own, when asked for any line I could pick it out in very short order, and I could return to it when any omissions were detected.
Bring the Heat
The final step was the reflow soldering itself, in this case with a small purpose-built reflow oven. Each run became a ten-minute anxious wait watching the display as the temperature cycled through its ramp, smelling the flux smoke and finally lifting the finished board from the drawer. There is something magical about the process of reflow soldering, watching messy spreading paste going in and bright well-defined solder joints come out. We examined each board as it came out, and it is inevitable that each one will require a small amount of reworking. There will always be a few bridges between adjacent pins or even components slightly off their pads, but this is the nature of prototype assembly.
After a very long day and a lot of preparatory work before that, we had a row of prototype boards. We should be able to commission them all as working devices, and from those my friend will have more than enough to satisfy his customer’s demand. All of the above describes a very long and tedious process but there is no reason why any Hackaday reader should not also be able to do a large board with a bit of practice. If we’ve inspired you to have a go at reflowing your own boards of whatever size then share the results with us on hackaday.io, meanwhile if you have any tips to further streamline the process with larger boards we’re all ears.
If you enjoyed this story of a marathon small-run manufacturing session you will also enjoy [Bob Baddeley’s] series on the topic called Tools of the Trade. You should also take a look at the process used to create the Hackaday SuperConference badge.