[Arm-netbook] EOMA-68 passthrough implementation

[lkcl luke]

On Wed, Jul 11, 2012 at 8:55 PM, Ryan Mullen <rmmullen at gmail.com> wrote:
> On Wed, Jul 11, 2012 at 2:47 PM, lkcl luke <luke.leighton at gmail.com> wrote:
>>  anyway: all of this is moot as far as a "simple" passthrough card is concerned.
>
> Good enough for me!

 :)

> Based on a quick skim of the datasheet from the TI HDMI->RGB converter
> IC, there appear to be no control pins that are expected to be tied to
> an MCU.

 no, i understand it to be a dumb DVI converter.  there are lots and
_lots_ of people who keep asking on TI's forums about "how to do full
HDMI including audio".  TI's staff recommend an additional 2 ICs
(eurghh....)

> Therefore, I don't see the need for adding extra smarts in
> that regard, since negotiation of all the extra "stuff" should happen
> in front of and behind the passthrough card.

 yes. basically.

 oh.  did i mention recommending dropping the RJ45 from the
passthrough card?  think about it: what's usually going to be on a
device's I/O Board?  um... an ethernet cable.  is there going to be an
ethernet transmitter or receiver IC on the I/O board? highly unlikely.
 if there is (such as the ADSL Model I/O Board idea on the
community_ideas wiki page), how can it be talked to? well, um, the I/O
Board is going to have an ethernet port, isn't it? :)  so in each case
all you'd be doing is adding an extra (redundant) RJ45 connector,
which in most cases would simply pass through 8 wires through to...
another connector, yaay!  utterly pointless: just unplug the goddamn
cable from the I/O Board for god's sake.  or just... leave it!

 but not so the SATA.  here there could well be many devices which
have internal SATA drives.  however i leave it up to you as to whether
you want to design this thing with a huuuge eSATA connector on it.
but, at least if it doesn't need a massive 12x10mm RJ45 connector on
the end that solves a lot of problems.


> We expect the I2C EEPROM to be onboard the EOMA68-compliant
> {tablet,LCD,keyboard} to be read by the standard PC-class device that
> connects to the USB/Ethernet/HDMI connectors.

 ... or more likely completely ignored :)

>The question, then,
> becomes: what connector would we use for the I2C lines that are passed
> through, so that they may be easily managed?

 oo, to be honest?  for most peoples' purposes?  a 0.4mm pitch single
line of 3 pins, right-angled to point out the end of the PCMCIA
connector would do it.  strictly speaking there's some sort of
standard for external I2C buses but i think you'll find the
connector's too big to use here.

> If we wanted to add a
> USB->I2C IC, (for example, FTDI FT232H), we could just slap on a
> mini/micro USB socket

 well there's a USB socket got to go on anyway, so why add 2 - why not
have a 2-port (or 4-port, waste a couple) USB Hub, they're typically
only about $1.50 (ok maybe that's in larger volumes).

> and the FTDI device should be able to act as a
> I2C master. We could then speak to the I2C bus with a ttyUSB or
> hyperterminal, or more native interaction if someone wants to write a
> program to do it. It's not the cheapest IC around, so maybe some
> shopping would be in order, but it's an example of one way to go.

 yes.

> Another option is to leave it up to the user to sort out and just make
> two I2C pins available. Users could use a Bus Pirate or Aardvark or
> something to talk to it.

 arrr :)


> What about this mechanical non-interoperability fascia plate business?

 yeahh, not going to worry about that.  decided to go for electrical
non-interoperability instead.

> Is there anyone out there who handles mechanicals? Where should the
> screw holes in the passthrough card's PCB go?

 que?  oh heck, don't worry about it for now.  unless you want to
actually make a case as well, in which case, the way i understand it
works is that the PCB is arranged to simply sit into the PCMCIA case
with no screws etc. it all just snaps into place.

> I suppose the placement
> and dimension of the holes should be standardized because AFAIK the
> intent is to use the same fascia plate for all EOMA-compliant modules.

 nono, they're all customised.  the hole of size 55 x 5mm is the only
limitation / requirement.

> Alternatively, is the fascia plate attached to the PCMCIA housing
> itself, without interfering with the PCB within?

 entirely up to each EOMA-68 implementor.  as long as it goes into a
PCMCIA Type I 5mm socket, that's good enough.

 l.