From: Werner Almesberger
Subject: Anelok: Mk 3 running on battery (1/3)
Date: Thu, 20 Apr 2017 13:17:26 -0300

Good news ! The Anelok Mk 3 prototype is now running the user interface,
and it's all on battery power. In the following pictures, all the major
pieces are present, except for the display cover.

Here we have the logo shown when turning it on:

This is the account list:

And the indicator LED works, too:

Regarding the LED position, it's slightly off-center horizontally.
I'll explain in part three why.

Coming next: the Mk3 teardown.

- Werner

From: Werner Almesberger
Subject: Anelok: Mk 3 teardown (2/3)
Date: Thu, 20 Apr 2017 14:23:19 -0300

What's more fun than building something ? Right, taking it apart !
So here comes the Mk 3 prototype teardown:

Step 1: look at what's under the hood

If I had already made the display cover, which will be glued to the
case, one would have to detach it first. In this prototype, one can
just pop out the OLED.

The sMCU is in the upper right corner, the cMCU in the lower right
corner. On the left side of the bottom edge one can see the USB
detection and power selection logic.


Step 2: remove the two screws that hold the bottom plate

Step 3: remove the bottom plate

We get to see the battery and the middle plate. The design of the
latter isn't finished yet - I just picked one that more or less
fits off my junk pile.

The USB opening looks a little large. I'll explain why in the
third part.


Step 4: remove the memory card and pry out the middle plate

Next to the battery is the boost converter circuit with its large
inductor. "South" of the boost converteris the BTLE chip (nRF51822),
with balun, capacitors, and crystals. The load switch for card and
display power is between RF and the card holder.

Between memory card and USB connector, there are only a few ESD
protection components, which I've omitted in the prototype.


Coming next: errata.

- Werner

From: Werner Almesberger
Subject: Anelok: Mk 3 (20170314) errata
Date: Thu, 20 Apr 2017 15:24:35 -0300

Of course, not everything went quite as planned. I found a
number of issues, fortunately most of them small:

I'm not done yet with design testing, so this list may still
grow. There are also a number of configuration choices and
deviations from the schematics:

Most of the deviations were caused by me not having the right
part - usually some resistor - at hand, and should have
negligible effect on the overall performance of the prototypes.
I omitted the power consumption optimizations (configuration
choices) for now. It'll be easy to test them later.

So far, so good. However, ...


I found a rather major blunder: I placed the footprint for the
OLED connector 1.9 mm too far to the right, and didn't notice
until I had the two prototype boards already made. One can see
it in this picture:

The OLED sits on the top of the case instead of going into its
cavity, and while the FPC (flexible printed circuit) connecting
the OLED flexes nicely in the Y and Z direction, it will not
budge in the X direction.

Moving the OLED opening in the case would eliminate the "bridge"
between OLED and buttons, and the buttons would then be free to
fall out - something they already enjoy way too much with things
as they are.


I solved this by shifting the whole PCB, right into the left
side wall. The tweaked case is on the left, a normal one on
the right:

Fortunately, the wall is 2 mm thick, so the PCB only looks odd
there but doesn't stick out. Shifting the PCB caused the LED to
end up left off the center of the buttons.

The shift also means that the USB connector is too far from the
right wall. To still be able to connect to the board while in
the case, I grew the USB hole to accommodate also the overmold
of the plug.

A "normal" case is on the top, the tweaked one at the bottom:

I also had to trim the cylinders that hold the screws, and one
of them gets too close to the inductor, transferring mechanical
stress. You may have noticed the likely consequence in

Fortunately, the inductor also works without that corner. Last
but not least, I moved the "nipples" at the bottom of the button
caps such that they still touch the centers of the tactile
switches.


The proper solution will be to correct the placement of the OLED
connector in the layout for the next board revision.

- Werner