anelok: first light

Werner Almesberger werner at almesberger.net
Sat Sep 21 11:22:52 EDT 2013


I put together a simple test board for display and wheel. It's called
"DUI", not for Developing Under the Influence but for "Development
board, User Interface".

Schematics:
http://downloads.qi-hardware.com/people/werner/anelok/tmp/dui.pdf

Files live here:
https://gitorious.org/anelok/anelok/source/dui

The board connects via a short cable and UBB to a Ben. The Ben
provides power and all control. The board only has the display, the
capacitors and resistors required for operating the display, the jog
wheel, and a few diodes for multiplexing display and wheel. There is
no MCU, RF, or any of the other complicated stuff.

This is what the electronics look like:
http://downloads.qi-hardware.com/people/werner/anelok/tmp/dui-wheel.jpg

The board has the size I'm currently aiming for for the final
device. (Note that this is likely to change. I've already grown it
by a few mm over the last days, to better accommodate the display.)

First impressions of the assembly:

- the display's FPC ("Flexible Printed Circuit", colloquially
  "Flexible Plastic Cable"), while small, was very easy to solder

- the data sheet doesn't make it clear, but the contacts are plated
  on both sides and seem to be connected (didn't test that). This
  means that one could also run the FPC under the PCB or (in a
  larger device) run the cable flat, without bending.

- that critter is tiny ! It's basically two thin glass plates with a
  film of magic in between. I was afraid of breaking it, especially
  when bringing adhesives into the game, but despite its delicate
  appearance it has resisted all my efforts at destruction so far.

- when checking the current setting resistors (R1 and R2, together
  760 kOhm), I found that they read far too low. I first suspected
  that I may have picked the wrong components. But it was probably
  just the flux. I've seen flux form conductive paths of some 100
  kOhms in the past, e.g., keeping chips in reset for a short
  while (long enough to make the other chip that depends on it fail
  to come up, yet short enough to destroy all evidence by the time
  one starts to measure), so this is something to take into account
  when building this sort of circuit.

- the wheel didn't pose any undue problems

Then I wrote a little test program on the Ben (using libubb) and
brought up the display.

The only problem there was that the Ben reset when turning on power
on the card slot. This is something that happens quite often when
connecting anything but a memory card. I did allow for a generous
"charge period" (driving all the data pins high to pre-charge any
caps through the pull-ups and only then turn on power), but this
wasn't enough in this case.

Not sure why the pre-charging trick didn't work this time. The board
has only about 11 uF in total and most of that is switched off on
reset. Fortunately, turning on uSD power on the Ben with the DUI
board present and then hot-inserting it did the trick.

The rest of the bringup went smoothly. I made my program display an
X bitmap file:

http://downloads.qi-hardware.com/people/werner/anelok/tmp/dui-hello.jpg

This picture was taken at regular lab hours, i.e., at night.
Readability is as good as one could hope for. Today's weather is
cloudy and the forecast says it'll stay that way for the next few
days. So I won't have a good daylight test for a while.

Contrast stays constant at any angle, very e-paperish and unlike
LCDs. The display does reflect bright light.

I then added Ron's nightmare, some tiny text (5x3 pixels) about 1.1
mm (3.3 pt) in height:

http://downloads.qi-hardware.com/people/werner/anelok/tmp/dui-arrow.jpg

Even that is barely readable. I then wrote a bit of code that reads
the wheel and tries to make sense of its inputs. That code is very
primitive (e.g., no debouncing) and it sometimes gets the direction
wrong, but it's good enough for demonstrating the concept:

http://downloads.qi-hardware.com/people/werner/anelok/tmp/anelok-dui.ogv

Xiangfu has uploaded the video to YouTube. I can't get it to load,
but this is where it should be:

http://www.youtube.com/watch?v=BYlb4---BU8

Don't read too much into what's happening there. That's not a real
GUI. All the program knows is how to display a bitmap and to draw a
little triangle. It has no concept of characters, fonts, etc.

Conclusions:

- I like the display. Works without a fuss and performs well.

- the wheel also seems to be fine. Debouncing needs some looking
  into. Since one can't grip the board very well with a Ben
  dangling off it, it's hard to tell whether the wheel will "feel
  right", but I'm optimistic about that.

- the PCB (0.8 mm) doesn't flex too much, even without a case
  supporting it.

- the wheel may be a bit too close to the display. I'll probably
  either move the display by a few mm to the left or enlarge the
  board by a few mm to the right. Depends on how the layout goes.

- I initially thought to have a some rigid structure for the spacer
  between display and PCB but given the flimsy nature of the
  display, I think just a bit of foam with a weak adhesive (the kind
  you find on Post-it(r) notes) will do.

- the display is susceptible to smudging and scratching. May need a
  silicone seal when put into a case.

Still not tested/checked:

- power consumption,

- voltage levels, especially the level of the high voltage (the OLED
  has a DC-DC converter to generate ~12 V),

- behaviour when turning on more than 50% of the pixels. This OLED
  has the known issue of its DC-DC converter being too weak for
  driving all the pixels at the same time. This shouldn't be a
  problem for normal white-on-black use, but we'd still want to know
  what happens when exceeding that limit.

- shutdown behaviour. The controller data sheet describes bringing
  down the rails one by one, but that doesn't seem to be a
  requirement. Need to check whether anything odd happens if I just
  do the proper software shutdown, assert reset, then cut all the
  rails in parallel. I don't expect any issues with that.

The next big item will be getting the MCU to work.

- Werner



More information about the discussion mailing list


interactive