I sent a comment to Tim via Kickstarter, but thought I'd copy it here
I'm sure there is a reason you went with an STM8 co-processor, but using a second reprogrammable ATmega328 would make a lot more sense, and allow the end-user more customisation options.
It would be great if you incorporated solder jumpers to allow the 4 independant switches to also work through the coprocessor, so they function like other keys rather than needing additional I/O, debouncing, etc.
Maybe an optional soft power controller via the top right button (using e.g. LTC2950/2951) - even if you just included the pads, and not the components if there is a cost issue...
Finally, you might have more interest if you add a couple of extra pledge levels - e.g. a single unit and a 3-pack. And make the t-shirt an optional extra - I for one already have too many (according to my wife) and international sizing is always a bit non-standard even with t-shirts.
Now I think about it, they're probably using a "bare glass" LCD which needs specific drive signals, and the STM8 variant they've used probably has an LCD driver. Not sure if there is an Arduino compatible micro that has an LCD driver built in, but a small text or graphical LCD would be much more flexible (although also more expensive). My other comments are specific to my typical small quantity project build requirements, which I imagine arent' unique.
You are correct, we needed an LCD driver for the bare glass LCD and the STM8 is much cheaper than the Atmega169. It is flash programmable and someone could reflash it if desired (the pins will be on the board.)
That's a good idea regarding the 4 extra keys: layout the board so they could be added to the matrix. That would take only 1 more pin and the co-processor could handle it.
We plan to have some prototype room available on the board, you could add the LTC2950 there.
Thanks for the feedback.
Tim
I'd forego the decicated BACKLIGHT button...
Leave it as an unlabelled or 'FUNC' button.
Backlight should time out after nn secs of inactivity (configurable in SW to stay on or delayed fade)
First keypress from 'idle' state turns on the backlight until another timeout or other event.
I've done this s a lot, and it looks and feels natural.
If you put this keypad / display functionality in a pre-packaged library - you might also put a 'dim' PWM level in there - so the backlight is not completely 'off' in the dark! Then user can find the unit on his wall or desk - and it wakes up when they press any button!
Cheers - nice idea.
P.S - to take it a step further. (...also done in the past)
Mylar keypads are too expensive for small-runs, and the multi-layer (seven- if I recall) with switches and overlay / window masks etc ... all require artwork and fabrication...
Provide an option for low-profile mechanical SMT buttons.
The proto and low-volume market will be appreciative.
Make the button board with as many positions as possible in the available space, and the plastic case half-way punched for all of them. Then a simple 2-layer mylar label defines the actual legends needed by the end-user application - and you rarely run out of buttons! (navigation diamonds etc)
Good luck! It can be done!
Cost of a graphical LCD was the first concern. Second, even a 128x64 lcd has 8K pixels. That can put a load on memory and the processor.
For many applications, the old fashioned 7 segment display is the easiest to read and by far the simplest to write code for.
Regarding the backlight for the LCD: it's controlled by a pwm under software control. The "backlight" button you see in the example photos are on a custom keypad we did for a specific project.
The production version won't have a dedicated backlight key; see attached pdf.
The production version won't have a dedicated backlight key; see attached pdf.
So who thought it was a good idea to make the big first image on the page be of NOT the product you are making? This campaign is a bit confused I think.
It seems too specialised to me - for general purpose prototyping which is what you are trying to aim at - especially given the LCD segment display.
Really would be much better with either a pixel display, or a good old character LCD, a small-size one you can probably get 4x16 in there, then you can dispense with the co-pro, just have an I2C bus expander.
The pre-printed button labels would annoy me, I don't want my "start" and "stop" buttons labelled "F1" and "F2" etc..
Many moons ago there was a games "console" called a Tunix (in NZ, in the US it's the Emerson Arcadia), it had controllers with a membrane numeric keypad built into it and the games came with printed plastic overlays that clipped into the controller over the keyboard. You might want to consider including some sort of keypad overlay feature into your enclosure - you've gone part the way there with your label insert.