Tolerances

Posted in Manufacturing , on Monday, September 29

We're not engineers, but making a physical product definitely required us to do some mechanical engineering. Working on DURR Beta's assembly, consisting of multiple parts and different materials, all fitting tightly together, really put us awe of all the thought and care that's put into every product... Even the low quality ones! We experienced a few things along the way that might help other novices:

Learn from other's work:
If you're not familiar with this, get right on it: Break open other products, and reverse engineer! It doesn't even have to be similar in size or in product category. It's mesmerizing to think of all the man-hours that have gone into the design and engineering of everything from ball point pens to a electric cable winch. Try to understand the thought process behind the different solutions and acknowledge both good and bad decisions.

Snap-fit was hard but rewarding:
SLS Polyamide for the back cover and screw-in battery cap had it's disadvantages . However, it enabled us in having multiple and replicable rounds of testing before we pushed it to the final production. After trials and errors we managed to construct a snap-fit geometry that held everything tightly, was reasonably easy to install , and spared us from using screws and/ or glue. The latter would most definitely prolong the assembly time, add parts and complexity, but could also make other components, e.g. the aluminum chassis, more expensive.

Don't stack too many deviations:
Every manufactured part deviates to some extend from the original drawings. In most cases, these variances are quite small, and they are usually found in the part's datasheet (often in a parentheses with a '+/-'). E.g. a part's height of 3.0mm can have a +/- 0.1mm deviation, which means its height ranges from 2.9mm to 3.1mm. At first glance this might not look like a big deal, but can in fact become one – especially if the deviations are stacked . We encountered this several times in the process, but perhaps most evident in regards to the coin cell's connectivity to its battery contact on the PCB. Adding up the possible vertical deviations on the aluminum, tape, PCB, battery contact, coin cell battery, + 2x Polyamide SLS, gives a range of +/-0.7mm. A height greater than the battery contact itself. Most likely, the variances would not add up only one way or the other, but if so, the battery would either be unreliably connected to the PCB, or not fit the enclosure at all.

Reiterate your thresholds:
Working with tolerances can easily give you tunnel vision; Everything have to be 0.15mm less to fit into its mating part. Or so it may seem. Not all parts are CNC machined out with the same precision, and not all materials are as rigid as it looks in your 3d software. So it's important to reiterate your threshold, and to keep in mind the material properties and manufacturing techniques. We weren't able to stay fully on top of this, but luckily avoided disaster by chance; Upon examination the hand made metal die (for the straps) deviated +0.5mm. This exceeded our unnecessary tight estimate of 0.2mm by quite a lot. But, because of the leather's flexibility (and durability), it only resulted in a slightly tighter buckle.

---