I spent a reasonably significant amount of time figuring out the Psion series 5 LCD and Keyboard interface. I built my own controller for the keyboard, and interfaced the display to a Gumstix. That was many years ago...
I've also done some prototyping in Shenzhen. CNC and 3D printed parts are cheap. If you've modeled it, everything apart from the membrane itself should be straightforward (and cheap in small quantities, <100USD). DirtySLA and itead provided cheap SLA printing options.
The membrane itself is likely more problematic. But you can make FPCBs, flexible circuit boards. In small quantities and at relatively low cost. Because the membrane is quite large I'd guess maybe 500USD? Getting domes in it probably is harder. I wonder if you can deform Polymide in some kind of jig. Alternatively could add a thin, 3d printed spacer perhaps?
If you really wanted to do it properly, I don't think that would be massively expensive either. Having a prototype might help. There are a bunch of companies that /do/ agree to do small runs. And I know many people how have done this. Often you end up having to visit the factory a bunch to push them to do what you want, and have occasional failures. But it's often still far easier than it might be elsewhere.
Wow, you figured the LCD out. I'm very curious how the LCD actually works - how is it updated? Through some kind of serial protocol? I get the impression it's pretty fast - I made a program in OPL a while ago that created a borderless window a metric stupid number of pixels high and EPOC was able to scroll it up the screen remarkably quickly (I was trying to see whether it could do haptic scrolling, it definitely could). It also seemed that I could drag windows around the screen faster than the liquid crystals could keep up - there was no choppiness here, it was rather an organic problem, so the LCD controller bandwidth/implementation seemed really good.
Flexible circuit boards are such a cool technology. To clarify, though, you're saying the Psion's keyboard is large? I presume the $500 is for initial tooling and that individual membranes will be a couple of dollars to run.
Another keyboard approach I think is really cool is that found in the original BlackBerry 850/950. I've admittedly only seen pictures of these but it looks quite usable.
> If you really wanted to do it properly, I don't think that would be massively expensive either.
That's incredibly encouraging, thanks.
> Having a prototype might help. There are a bunch of companies that /do/ agree to do small runs.
Oh okay, cool to know.
> And I know many people how have done this.
:D
> Often you end up having to visit the factory a bunch to push them to do what you want, and have occasional failures.
Hmm, plane tickets...
> But it's often still far easier than it might be elsewhere.
Older LCDs have a lower level interface. From memory, there's a pixel clock, line clock and frame clock. Then a parallel bus with the pixel value. The cirrus logic ARM SoC in the Psion 5mx interfaces with the LCD directly. You also need to supply it with some weird voltages (+/- 15V?). When I was doing that I found that part more difficult, I think I'd find it easier now...
The interface is pretty standard, so the gumstix (TI OMAP3something) can also drive it.
The keyboard is pretty big for a FPCB. I'm just guess 500USD, for maybe 10 pieces. Because that's the prototype level pricing I've seen before (check itead.cc or Seeed, they both do FPCB now I believe).
The process they actual use in membrane keyboards is different, I don't know what that process is called.
Hmm. When you say "talks to the LCD directly", there's still a tiny low-level controller attached to the LCD, right?
While highly nonstandard, but I want to see how feasible it is to truly drive an LCD directly. To me that sounds like a) huge pin count, b) "fun" prototyping costs, and c) difficult conversations with manufacturers who don't want to explain how their controllers work (and I guess d) a lot of work on my end to support all the different LCDs I'll use). My motivation is to see whether it's possible to control an LCD and touch panel or keyboard from a single chip.
> The interface is pretty standard, so the gumstix (TI OMAP3something) can also drive it.
Good to know.
> The keyboard is pretty big for a FPCB. I'm just guess 500USD, for maybe 10 pieces. Because that's the prototype level pricing I've seen before (check itead.cc or Seeed, they both do FPCB now I believe).
Oh okay, so $50ea for prototyping. Cool.
> The process they actual use in membrane keyboards is different, I don't know what that process is called.
I was curious about integrating LCD control into a single processor because the security implications are really interesting. Just an idea at this point though.
I remember looking inside cheap electronic organizers and seeing a lone blob on the PCB. So I know managing simple LCDs is doable.
I've also done some prototyping in Shenzhen. CNC and 3D printed parts are cheap. If you've modeled it, everything apart from the membrane itself should be straightforward (and cheap in small quantities, <100USD). DirtySLA and itead provided cheap SLA printing options.
The membrane itself is likely more problematic. But you can make FPCBs, flexible circuit boards. In small quantities and at relatively low cost. Because the membrane is quite large I'd guess maybe 500USD? Getting domes in it probably is harder. I wonder if you can deform Polymide in some kind of jig. Alternatively could add a thin, 3d printed spacer perhaps?
If you really wanted to do it properly, I don't think that would be massively expensive either. Having a prototype might help. There are a bunch of companies that /do/ agree to do small runs. And I know many people how have done this. Often you end up having to visit the factory a bunch to push them to do what you want, and have occasional failures. But it's often still far easier than it might be elsewhere.