OLPC XO 1.5 overheating problems

[John Watlington]

On Oct 3, 2009, at 12:16 PM, Tiago Marques wrote:

> On Sat, Oct 3, 2009 at 4:19 PM, Ed McNierney <ed at laptop.org> wrote:
>> Tiago -
>>
>> Well, everyone gets to contribute something to thermal problems :-)

Actually, the average heat sources are (in descending order): the  
companion
chip (graphics engine, memory interface, and peripherals), the CPU,  
the DDR2,
the WLAN (!), and the clock generator.  At peak operation, the  
processor wins.

We are working to reduce the WLAN dissipation, and I'd love to be  
able to
afford a lower power clock generator as well.

>>  But the CPU's the main source and also provides internal  
>> throttling mechanisms to
>> manage it (along with the external heatspreader mechanisms, of  
>> course).
>
> I hope that the heatspreaders might fit in your budget for the
> machine. I've seen all kinds of bad happen to machines where
> throttling was necessary.

There is definitely a heat spreader.   We have encountered problems
in the past with manufacturing problems that contributed to a less
than stellar contact with the chips being cooled, and that appears to
be contributing to the current concerns.

The throttling is a measure of last resort.  It means that if you DO  
operate
an XO in a room that is a few degrees too hot, it won't crash.  It  
will just
slow down.

Cheers,
wad