[Arm-netbook] Fwd: [eoma68 update] report back from factory on HDMI

[Paul Boddie]

On Wednesday 3. July 2019 09.03.44 Luke Kenneth Casson Leighton wrote:
> 
> Mike's staff began the PCB assembly of the run of 100, and had to stop
> at 36.  20 were ok: 16 of them, the HDMI connector refused to fit.
> The reason: the CNC machining on the edge of the PCB has not been done
> accurately enough: it's simply too ragged.  The staff did some
> experimentation, cleaning up some of the edges in the cut-out with an
> xacto-knife: this did the trick, even though it is shaving something
> like 0.01mm off the ragged edge of the PCB.

I can imagine how frustrating this can be, but it is good that some 
experimentation on the spot was able to identify the likely cause.

(My own limited experience with PCBs, albeit with "maker"-level producers, 
suggests that some PCB cutting/milling is not done as accurately or in as tidy 
a way as one might expect. There have also been some other interesting 
experiences with production quality with some producers, but that is another 
story.)

> The left and right edges do not matter too much, however where the
> HDMI connector comes in close, it definitely does.  Mike is going to
> talk to the PCB factory to see if there is anything that they can do
> in future, however with 1,000 PCBs already manufactured, the safest
> thing to do is probably to *hand-trim* that PCB edge, removing the
> burrs, on all 1,000 PCBs.

It is too bad that this adds another step to the process, but it at least 
hopefully remedies the issue. I guess it shows that minor faults can become 
more serious problems further down the chain of production.

> Again, to reiterate, because I am still seeing evidence of
> "complaints" out there, from people who believe this should be easy:
> these are absolutely ridiculously tiny components and tolerances, and
> the budget on which it's being done is equally as frugal.  0.05mm on
> the edge of a PCB.  0.2 mm wide pins, with 0.2mm clearance between
> them.

Having seen the measurements related to various parts, I can more easily 
believe how challenging it is. And it seems intimidating when considering the 
demands of production to such levels of accuracy.

> A "normal" Single-Board Computer product from any other
> well-funded Corporation would use large (Type A) HDMI, top-mounted,
> with plenty of tolerances and no need for the PCB edge to be
> accurately milled.

Yes, it seems that the larger but less ambitious players get off easy again.

> Again, to reiterate: we do not know what will need to be solved next.
> Therefore, a production date simply cannot be provided, and that
> really is the end of the matter.  Or, the answer is: the production
> date is "the production time plus the unknown time to solve unknown
> and unknowable future issues".

Well, I hope that you, Mike and his employees feel that valuable experience is 
being gained that will pay off in future. It would certainly be nicer if 
everything went more smoothly, but one might then wonder which potential 
problems are being missed.

> Mike is sending me the 20 "good" PCBs so that I can test them here, to
> see if they are okay.  The staff will continue with the rest by
> shaving the burrs on the PCB on every single one of the remaining 80
> with an xacto-knife, before putting them through the production line.
>  It is looking like I will need to do the testing of all 100 of this
> preliminary production run, here, at my home, in Taiwan.

And I hope that the rest of the run and the testing go a bit better than how 
things have gone so far. Keep up the good work!

Paul