[Arm-netbook] SPI-based LCDs, 3D printing, RISC-V

[Luke Kenneth Casson Leighton]

On Thu, Apr 27, 2017 at 4:10 PM, Bluey <bluey at smallfootprint.info> wrote:
> Hi All,
>
> For what it’s worth, I’d like to thank all of you involved for taking on
> this important project.

 thanks bluey.

> I started subscribing to this list just a few days ago so please forgive me
> if specific design requirements/parameters have already been decided for the
> EOMA computer and my comments are not germane.

 all appreciated.  the only thing not "up for negotiation" is the
EOMA68 standard revision 1.0, and any future revisions absolutely must
be fully backwards-compatible.  a surprise discussion a few months
back brought up a couple of great ideas: if you want higher
resolutions than 1366x768 the Housings *must* use line-buffer or
frame-buffer "upscaling" display chips which can still accept 1366x768
(or below) and yet will perform the necessary hardware-level scaling
to the Housing's output resolution (presumably but not necessarily
fixed).

so there are some things that are definitely final.

> I’m also not across which
> manufacturing skills, tools, and facilities are available within or to the
> group.

 there's been a lot of expertise that has crossed this list over the
past six years.

> With the above said, the best option to my mind would be to 3D print and
> finish high-quality moulds (in titanium or steel) for use in stamping out
> aluminium case components from sheet metal.

 the laptop's casework parts quite literally took eighteen months to
develop, six months of which was in the PCBs which had to be also be
developed and hand-populated during that time, to make absolutely sure
that the PCBs and casework would actually fit together.

 with only 150 pledges there's absolutely no way - at this immediate
time - that any kind of redesign or additional effort beyond
fulfilling those 150 pledges can be justified, especially given (as
has been mentioned a few times) that the budget is already
insufficient and a plan has been activated to bring in extra funds.
stressing that plan by adding to it is not a good idea, *but*, if
there are alternative workable ideas that come with resources attached
i'd love to hear them.

>  I see modular design to be
> important primarily for the internal component selection and layout inside
> the laptop body rather than for the body itself, which—to my mind—should be
> designed for longevity, strength, rigidity, and good looks (important for
> attracting more people to buy the product over time).

 i've looked at a number of designs of this type (pi-top being the
most recent) - i posted about many of them during the campaign.  i've
done the best that i can, and kept the weight down to 1.1kg (for a
15.6in laptop that's incredibly good) which means that the usual kinds
of structural stresses that require a metal chassis or special
engineering analysis design tools have not been necessary.  the
prototype i have found that it can be lifted up with one hand at its
corner and it *does not* bend out of shape, due to the light weight.

 that, and insisting on using faberdashery's high quality PLA.


> Lost-wax metal casting has also been effectively demonstrated with
> 3D-printed thermoplastic plastics but to me that seems like it might be a
> less environmentally friendly option (as compared to the above) despite
> possible reuse of the lost wax (the plastic). However, it might be cheaper
> and a more-viable option for such small manufacturing runs.

 there's quite a bit of detail involved (0.8mm thickness in places) so
i would be very very surprised if metal casting was successful.

> However, if the decision is made to go with plastic, and assuming a redesign
> is a possibility, one option would be to 3D-print the injection mould in
> metal (such as titanium) using a specialist 3D-printing service.  The mould
> could then be used in a standard injection-mould machine.

 do you have any idea on how much that would cost?  if i recall
correctly the PLA parts weigh 200g in total, and there are 35 of them.

> It can be tricky getting truely environmentally friendly biological
> materials (such as bamboo) from non-western countries.

 yeah birchwood turns out to be the most common in europe: 1.5mm 3-ply
which is perfect is very commonly used in making doll's houses.

> My proposed ordered list above discounts the carbon-intensity of materials
> manufacturing, as well as the damage caused from the mining process, but
> working out such things is extremely complex and requires detailed audits of
> source material, material suppliers, and processing manufacturers.  It also
> changes from country to country and from region to region within certain
> countries.  For this project, I think it would be necessary to look at
> long-term material use and reuse.  Aluminium, although requiring lots of
> (possibly dirty) energy to produce, is also an exceptional store of energy
> due to its capacity for perpetual recycling.  It should also meet the fit,
> finish, and strength attributes required for the laptop housing.

 surprisingly, faberdashery's PLA combined with the light-weight
nature of the laptop does a really good job.

 once through this first phase and this first 150 is out there, other
options can be considered.


> P.S. Without getting into the complicated world of finite element analysis
> and minimising points of high strain in the design, I would recommend
> well-rounded internal edges wherever possible if/when designing a chassis
> case.  This will reduce high stress concentration in those areas.  The
> sharper an edge or corner (e.g., on a rectangular hole) the higher the
> stress concentration in the material will be at that location.

 yeah it's very common for the longer parts to pull off the printbed
as the layers go up.  one of the worst offenders is the base back
(left/right) which is 160mm long, 11mm wide and *35* mm high.  i had
to add a lot of side-bolstering and also a wide brim to get it to
stick, but someone with a much higher-quality 3D printer
(all-aluminium, several thousands of dollars of time and materials)
was able to get away with just one "pinning" square at the front and
the back.

 however i now have two pieces of printbite which i will be trying out
soon, we will see how that goes.

 but, essentially, unless someone is prepared to help tackle the
casework (using libre software services, not e.g. github), due to the
amount of time it took (18 months) i will go with what i have for this
first revision.

 thanks bluey.

l.