by Bryan Bergeron, Editor by Bryan Bergeron, Editor
DE VELOPING
A MAKER'S DOZEN:
Designing Around Failure
Ijust finished a major production run of simulators for training physicians on how to examine the eye. I needed to deliver
10 units, so I started with 13 mannequins — each with high-resolution LCD screens for the backs of the eyes (retinae), an
Arduino, and a custom interface board. In addition, I printed
holders and mounting brackets for the LCD and each of the
two boards. Add to that about 30 wire-wrap connections for
power, an LED indicator, and interconnecting the major
boards, and you have a recipe for guaranteed failure. The only
question was the success rate.
For this project, there were several sources of failure. For
starters, two of the custom boards were DOA. It turned out the
manufacturer used an NPN transistor in the output circuit when
a PNP was called for. I had to contact the manufacturer and
arrange for replacements. Not a major problem, but it took
several days for the supplier to provide the replacement parts.
Then, there was operator error. I flipped a green and blue pair
when wire-wrapping the LCD to the Arduino on one of the
simulators. With a wire-wrap tool, this was an easy quick fix.
Slightly more problematic was my MakerGear M2. It
performed superbly when printing the circuit board holders in
PLA, but when I switched to ABS plastic, output quality
became erratic. After a bit of experimenting, I found that
increasing the platform temperature a few degrees solved the
problem. As I've learned on several models of 3D printers,
these problems are to be expected. Still, each failure cost time
— each component required two to three hours to print. As it
was, the printer was cranking nearly 24/7 for a week to make
the delivery deadline.
In the end, I delivered 10 units on time. Within a week of
that, I had repaired the additional three units, which I maintain
for backup and quick replacement. Based on my failure rate,
I'd say that a proper maker's dozen is 13. 5 units. That is, if you
need to make a dozen relatively complex devices, order
enough spare parts for 1.5 additional units.
You might be wondering how any business could succeed
with such a high failure rate. After all, if one in 10 or so
consumer electronics products failed, then there wouldn't be
much of a consumer electronics industry. A difference
between DIY and commercial electronics is that DIY
components are often seconds and the overall design
specifications shift over time.
Consider the 3D printer, for example. After the first print
of a given component, I inevitably modified the model to
improve it. Adding a brace to a thin wall, moving a mounting
hole, or simply changing the color of the PLA filament has
risks. In addition, not all PLA is created
equal. I've found that the inexpensive
bulk PLA on eBay simply isn't as good
as the more expensive PLA filament
from the print manufacturer. I don'
know if it's the chemical composition,
the diameter, or variation in some other
parameter, but I do know that the PLA
from MakerGear produces consistently
better prints.
If you're building one-offs, you
might not notice this relatively high
failure rate. It's worth considering if
you're building a number of units for,
say, a club or classroom. Plus, when it
comes to DIY, it's not really a "failure
rate" but a learning opportunity. After
all, you're not going to learn much if
your experiments always work. On the
contrary, if you're DIYing is 100%
successful, then you're not pushing the
edge hard enough. Good luck with all
your DIY projects. NV
PERSPECTIVES
6 April 2014