[GTA04] When is the next and more powerful openmoko releasing

Al Johnson openmoko at mazikeen.demon.co.uk
Mon Oct 25 20:33:07 CEST 2010

On Saturday 23 October 2010, Dr. H. Nikolaus Schaller wrote:
> Am 23.10.2010 um 15:40 schrieb Al Johnson:
> > On Friday 22 October 2010, Dr. H. Nikolaus Schaller wrote:
> >> Am 22.10.2010 um 02:17 schrieb Al Johnson:
> >>> On Friday 22 October 2010, EdorFaus wrote:
> >>>> On 10/21/2010 07:59 PM, Alfa21 wrote:
> >>>>> you should try 3d print like this:
> >>>> <snip>
> >>>> 
> >>>>> made in ABS which is a good plastic and up to 0.01 inches resolution
> >>>> 
> >>>> Unfortunately, that's not quite good enough (different units).
> >>>> 0.01 inches = 0.254 mm, which is 2.54 times the required size:
> >>>> 
> >>>> On 10/21/2010 07:32 PM, Dr. H. Nikolaus Schaller wrote:
> >>>>> the required precision (0.1 mm, 0.5 mm wall thickness, etc.)
> >>>> 
> >>>> 0.1 mm is approximately 0.0039 inches.
> >>>> 
> >>>> I have a question, though - is this precision required for making
> >>>> *any* well-working case design (due to the electronics etc.), or is
> >>>> it just for making more of the current case design?
> >>>> 
> >>>> If it's the latter, it would be possible to create a new case design
> >>>> that would be easier to get manufactured...
> >>> 
> >>> The electronics don't need much precision in the case design. Cases
> >>> that clip together firmly and securely do need precision though, and
> >>> that's true of the current design. An alternative design needing less
> >>> precision is possible, especially if we accept things being screwed
> >>> together.
> >> 
> >> Generally yes. If we sacrifice quality, ergonomics, and size, we can
> >> find simpler and less expensive case constructions.
> > 
> > I'm sure we can find alternative case construction methods that don't
> > sacrifice much in any of those departments. They just won't be methods
> > used in the mass market because they don't scale up to those production
> > volumes, just as their methods don't scale down to ours. The pcb might
> > need some design
> Well, we have anlysed approx. 5 or 6 different methods and none was
> inexpensive for low volumes. The only one that came down below 50 EUR per
> case is injection moulding.
> But I would be happy if you can guide the GTA04 to a different approach.

It would be good if you could share the results of your research so other open 
hardware projects can get an idea of what a custom case might cost, and see 
how different manufacturing methods compare. It might also stop us suggesting 
things you've already investigated an rejected! Also exactly how low is low 
volume? Given the differences in setup costs for different manufacturing 
methods it could make a huge difference to the unit cost.

> > elements suited to a different method of mounting than the current 'clip
> > it into the injection moulding' variety, but this is for future
> > generations not the current pcb.
> Well, the current board is really mounted by 2 torx screws and the
> connectors which fit into holes on the sides. The hooks and clips are not
> really necessary, which allows for a lot of simplifications. This is where
> some of us are currently working on to get the plastics produceable by a
> RepRap.

I'll have to open my FR again to refresh my memory. I remember the GSM antenna 
connection being somewhat integral with the case assembly, but it would be 
nice to be mistaken! 

> >> One aspect to keep in mind for precision are the push-buttons (AUX and
> >> Power). The buttons [1] are specified for a "Travel 0.3 mm +0.1 –0.2
> >> mm". I.e. if we don't want that the buttons get stuck we come to this
> >> 0.1mm precision. We have experienced the same with the PCB design where
> >> the position of the buttons must be within these 0.1mm...
> > 
> > That doesn't mean the case needs that tolerance though. It just means the
> > end of the actuator needs to stick over the edge of the pcb by the right
> > amount to be operable. The external button can be lightly spring to take
> > up the tolerance, or rest slightly clear of the button actuator.
> > Alternative switch
> If I understand correctly, then the button would not fit smoothly into the
> case. It may stand out (or in) so it either gets trapped in the gauze of
> your trouser pockets, or you need long fingernails to press it.

If designed to rest slightly clear of the pcb button it would sit where 
designed. If lightly sprung it would sit slightly further in. A tapered case 
edge, as we have with the current power button, makes this less disruptive as 
well as easier to find by touch. Even with a 0.5mm recess, which I doubt we 
would need, a button the size of the current aux button wouldn't need 
fingernails to press.

> > types are an option for future designs, including a change to capacitive
> > sensors as provided for by the v2 nav board.
> Not completely. The power button must be a mechanical button to wake up the
> processor from deep sleep. Or we drain the battery for a always-powered
> sensor chip. And, I am not sure if a sensor contact is safe enough for a
> smartphone in a pocket.

The MPR121 on the nav board is supposed to draw 29uA for a 16ms sample period, 
less if we can stretch it out further. That gives a nominal battery life 
greater than 4 years, so battery drain shouldn't be a problem. It has an 
interrupt out pin, so wake from deep sleep shouldn't be a problem. I guess 
we'll need some experience to see if it's good enough for life in a pocket, 
but other phones with capacitive sensors seem to manage.

> >> The same is with mounting the display. If that is not precise enough,
> >> touch operation will be deteriorated.
> > 
> > Can you elaborate? There may be ways around this one too.
> The main thing is that the display bezel must be plane. Otherwise it may
> press on the display (touch) at one end and increase the risk of breaking
> the display glass. Or you get dust inside the device.

The flatness requirement is fair for rigid(ish) materials. A thin compliant 
strip could help seal it and even out the pressure distribution, but is 
another part to cost. This brings the (probably crazy) idea of an entirely 
silicone case...anyone know anyone who deals with silicone?

The only mass-produced phone I've had that didn't get dust inside was 
waterproof. The FR has come a very close second on that front, but based on 
experience of other phones being dustproof is not necessary.

> On the other hand we know that the bezel of the Freerunner is already too
> thick. I.e. for better operation it should be much thinner. But still
> stable.

Or at least with a gentler slope towards the screen edge so it's easier to 
touch there. 

> Well, all this can be improved by changing the construction.
> >> So if we reduce precision of the case production process, the buttons
> >> become larger and finally, the phone is more a brick than a handheld
> >> phone
> >> 
> >> :)
> >> 
> >> Personally, I am in favour of high-end high-quality design (like the
> >> iPhone) even if it costs money. But it should never cost freedom and
> >> independence...
> > 
> > high quality design doesn't have to be expensive to manufacture. I
> > suspect something attractive could be made in low volume at a reasonable
> > price using laser cut plastics. I'll try to ask the local cutters about
> > their precision and cut thickness in different materials.
> I agree that just suspecting that it exists is not enough .
> We have already spent several months discussing and asking for quotes and
> everyone can make such things. Technically. But if we ask for cost we are
> either beyond 100 EUR per piece in low quantities. Or there is need for
> some upfront investment that only pays off if we make more than around
> 1000 units.

If you've already investigated laser cutting I'll stop now. I suggested it 
because the setup cost is very low relative to other methods, and I've seen a 
couple of one-off cases made that way. It also lends itself to manufacture at 
someone's local workshop, including laser-etched personalisation. I'm assuming 
people will be up for a bit of self-assembly here.

> I think we all will be very happy if we finally find another method that
> does not need big VC money, i.e. can be done by joining all forces of our
> community.
> Well, an alternative approach could be to open a funds for a new injection
> mould. I.e. a "Openmoko Case Production Foundation" (or German e.V.).
> Everyone who gives some money for the funds can buy cases at a reduced
> rate. And everyone can buy a case. An injection mould costs somewhere
> around 20-100 k EUR. So if we get 300 community members to donate 100 EUR
> into the funds...

That's certainly worth a look, especially if with a case design that would 
last for several generations of board.

More information about the community mailing list