[Arm-netbook] a10 eoma68 cpu card bring-up

luke.leighton luke.leighton at gmail.com
Wed May 15 15:51:55 BST 2013


On Wed, May 15, 2013 at 3:23 PM, mike.valk at gmail.com
<mike.valk at gmail.com> wrote:

>>  yes.  except that requires case rework, which will be an extra $10k.
>
> I thought to put it on the bottom side of the outside facing side.
> We're using two midmount connectors and a top mount micro-sd.
>
>       MicroSD
> -HDMI---------USB-
>       MicroSD

 yeahhh, space. (height)

> I know very simplistic though, I understand that the PCB is crowded
> with traces already.

 it's not that: there's not enough height.  low-cost 6-layer PCBs come
in at 1.2mm - if you go to 0.8mm it's quite a jump (relatively
speaking).  MicroSD card slots are 1.9 to 2.0mm, so that's 5.2mm in
total, and PCMCIA casework is 5.0mm meaning you only have 4.8mm to
play with and also need some clearance top and bottom for a plastic
end-plate.

 as it is we're looking at 4.8mm - 3.2mm = 1.6mm divided by 2 is 0.8mm
above and below (for plastic!) so it's real tight.

>>
>>> Some of the benfits.
>>> 1. Boot from removable storage. While keeping a swappable storage available
>>> 2. Less wear on the soldered on flash
>>> 3. The A10 can expose UART and JTAG via the sd interface
>>> http://linux-sunxi.org/MicroSD_Breakout
>>>
>>> Ehm why aren't you trying UART/JTAG over the SD-Card interface now anyway?
>>
>>  because they wired the wrong fucking SD card interface *without*
>> notifying me.  did i say "don't ask"? :)
>
> Sorry I did ;-). I assumed solder pads for JTAG/UART on the board.
>
>>
>>  it's actually for very very good reasons, namely that the board's
>> been designed extremely well - 2 GND planes, 1 Power meaning it stands
>> a high chance of passing FCC - but it's leaving only 3 signal planes.
>>
>>  so, only 3 signal planes for 4 DDR3 RAM ICs, processor, PMICs, NAND
>> Flash, MicroHDMI, MicroUSB, MicroSD and a 68-pin connector in a space
>> only 45 x 75 mm it's a goddamn miracle the engineer pulled that off -
>> he's extremely competent in that regard.
>
> Applause. Good job indeed.
>
>>
>>  so they couldn't route SD0 to the MicroSD because it was on the other
>> side of the CPU, likewise SD3 would have crossed over.  so the
>> engineer made a very sensible decision to swap them over... but didn't
>> _tell_ me.
>>
>>  and i'd already truncated the GPIO pins from 16 to 8, which cut SD3
>> in half.... actually now SD0.  if they'd asked i would have said "oi!"
>> and asked them to drop IR_TX and IR_RX and put the full SD0 on them
>> instead.
>>
>>  but yes.  we want a top-loading micro-sd... in the future, not not,
>
> Someting like this you mean?
> http://static.videomaker.com/sites/videomaker.com/files/styles/vm_image_token_lightbox/public/articles/15633/micro-sd-card-in-cell-phone.jpg?itok=jd9JlrmU

 yep that's it.

> That whould indeed make more sense for an SD card with the OS. Which
> you don't want to accidentally eject while running.

 correct. now you need a hole in the top casework, and that's
around.... $5k to $10k for the development and tooling because it's
custom tooling.

 all these things that need to be taken into account, it's a bloody
miracle we got this far on such a small budget!

l.



More information about the arm-netbook mailing list