[Arm-netbook] EOMA68 pinout change may be needed
luke.leighton at gmail.com
Sun Feb 5 23:43:31 GMT 2012
On Sun, Feb 5, 2012 at 11:10 PM, Phill Rogers
<PhillRogers at jerseymail.co.uk> wrote:
> Hi, newbie alert (in this forum) so sorry if this has been covered.
allo newbie in this forum :)
> Why not just use PoE ?
> Even if the host device doesn't use Ethernet, or the user doesn't have a PoE
> switch, you can still use a 'power brick' to inject the power onto the Ethernet
> Probably still makes sense to have at least one real 5v pin in any case.
i'm inclined to say that if POE is needed, it should be on the
motherboard, even if that motherboard is about 56mm long and 25mm
the reason for that is the extra cost of the components in cases
where POE is *not* required (or used) would potentially make a
mass-volume product uncompetitive or unprofitable.
if it's actually in the _standard_ that POE is required (no,
standards *must* not have anything that is "optional" in them), then
that would potentially jeapordise the entire project.
EOMA is about covering as much as possible with "Lowest Common
Denominator" interfaces that have downwards / upwards compatibility
built in to those standards, already. so, yes, 10/100/1000 ethernet
is ok, because if you connect an 8-wire socket to a 4-wire plug (or
vice-versa), ethernet auto-negotiation kicks in. likewise for SATA
and USB: the standards auto-negotiate the fastest possible data
transfer rate (yes i've explicity said on the standards page that USB2
480mb/sec *only* CPU cards are banned).
surprisingly, the bar is extremely high (on the
Lowest-Common-Denominator) thanks to that auto-negotiation. but adding
things like POE? naah. can't do it.
More information about the arm-netbook