Plug computers

[D. Hugh Redelmeier]

| From: Lennart Sorensen <lsorense-1wCw9BSqJbv44Nm34jS7GywD8/FfD2ys at public.gmane.org>

| It takes a lot more power to operate at 1Gbps than at 100Mbps.
| So actually it does make sense.  A fairly common estimate seems to be
| about 100mW for 100Mbit, 1W for gigabit and 10W for 10Gbit.  Or at least
| in that order of magnitude.

Interesting.  Thanks for the information.

1w extra isn't a lot to dissipate but this is a very passive design.

This document is about the central chip: 
<http://www.marvell.com/products/processors/embedded/kirkwood/HW_88F6281_OpenSource.pdf>
The PHY for ethernet is off-chip.  I have no idea what PHY thermal 
characteristics might be.

Section 8.3 deals with thermal power dissipation.  Eyeballing, it
looks like the chip only dissipates two or three Watts and that the
non-PHY ethernet power isn't broken out or if it is, it isn't much.

Looks like a pretty remarkable chip.  Too bad the CPU core isn't as
well supported as newer ARMs.

Looking at
<http://www.globalscaletechnologies.com/t-guruplugdetails.aspx> I
would guess that the main thermal problem is that the power supply is
in the same package.  Convenient, but it adds a fair bit of heat to
dissipate.  Especially if a couple of USB devices are drawing a lot of
power (max is 500mA * 5V * 2, close to 5W).
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists