AlphaServer DS10 mainboard in a DS10L?
Marc Schlensog
I planned on swapping the mainboards from an EV6 DS10 and an EV67 DS10L
to eventually get an EV67 DS10 and an EV6 DS10L. One thing that struck
me were two heatsinks in the DS10 which weren't present in the DS10L
(one in the front near the fan and front panel connectors, one in the
back near the CPU).
Now I am wondering if I could safely remove the heatsinks from the DS10
board. I would guess that the airflow in a DS10 is much better due to
larger fans.
Will I run into thermal issues with the heatsinks removed?
Thanks and regards,
Marc
H Vlems
Marc, first of all, I'm not familiar with either machine. That said,
removing heatsinks is probably not a good idea.
The DS10L is a smaller box. Removing heat is an art in itself; e.g.
the Multia thread now current in comp.os.vms.
A heat sink increases the surface of a hot device that is in contact
with cooler air. A heat sink is made of a material that
conducts heat wel AND has a low heat transfer resistance.
That's why a stone floor is colder to the touch than a wooden floor
even when both have the same temperature.
A device without a heatsink needs a lot of air. The cooling surface
isn't much to begin with and heat transfer is limited.
Lots of cool air are neede and the smaller DS10L probably has more
fans than the DS10. The DS10 is more silent than a DS10L.
The latter was designed purely for use in datacenters (before blades
existed) so the noise level wasn't an issue, rack space was.
The DS10 was a departmental server, could be placed in an office so
less noise was a design goal.
If the EV67 emits more heat than the EV6 (and it very likely does even
when they run at the same frequency) then a swap is
not recommended unless you can improve the heat transfer. Adding more
fans may do that. But it makes no sense to increase air flow
if the air can't spend enough time in the box to heat up. If the
temperature difference between incouming and outgowing air is very
small than either there's no heat source inside or the air doesn't
absorb the heat.
Hans
Marc Schlensog
H Vlems <hvl...@freenet.de> wrote:
[...]
Hi Hans,
thanks for your reply. I know that the idea to remove the heatsinks
isn't the greatest I've had so far. That's why I asked before doing
stupid things :)
The thing that struck me odd was an article I found on the development
of the DS10l[1]. One paragraph in it goes like this:
To make room for a full-length PCI card, engineers removed the heat
sink from the CPU regulator and placed a black label on it. This
label primarily provides PCI-to-regulator "short circuit" protection.
Through "fan fail" condition (limited airflow) testing, Sullivan
found that the label also increased the surface emissivity of the CPU
regulator, lowering the surface temperature an additional 4C via
natural convection.
Now keep in mind that they were taking the exact same board. And the
heatsink of the DS10's CPU regulator is pretty big. I'm surprised that
that even worked out. The other heatsink, that's different in the DS10
I have, is connected to a power transistor near the CPU. The DS10l has
just some itty bitty heatsink connected to it.
I think I'll have to check the part numbers of the mainboards (I have
some 466MHz DS10l laying around, which from the looks are more similar
to the 617MHz DS10l than to the 466MHz DS10).
Actually I think that the EV67 should consume a bit less power at the
same speed than an EV6, since they went through a process shrink from
0.35um to 0.25um. Anyway, the CPU isn't what I'm worried about but the
CPU regulator and the power transistor.
b.r.,
Marc
[1] http://www.contentconvergence.com/editorial/compaq/compaq.htm
H Vlems
>
> - Tekst uit oorspronkelijk bericht weergeven -
Marc, when the EV6 and EV67 run at the same frequency it's safe to
assume that the EV67 runs at less power. It has to because the heat
output would damage the chip.
The regulator and the power transistor may run at a higher temperature
than the cpu, the circuit density is probably a lot higher on the
Alpha processor.
If I understand your post correctly both systems have the same power
transistor; there's a heatsink in both but the one in the DS10L is
smaller, right?
I guess this means that: either the smaller heatsink is more eficient
than the bigger, or it is not in which case the power transistor just
runs at a higher temperature. A temperature which won't damage the
component but higher nevertheless. Higher temperatures speed up the
aging process of semiconducters so lower tempartures improve
lifespans. Even if the temperature is within specs, higher means
shorter life. The same applies for the regulator.
More fans, increased air volume may keep the temperature down at the
expense of more noise.
So if you decide to swap the boards, increase the number of fans or
replace them with higher capacity ones (while you're at it, buy ear
protection ;-)
Another option is to check out heatsinks that might fit and have a
better heat transfer rate.
Hans
LenW
Hi Marc:
You run a VERY HIGH risk of burning up the original DS10 mother board
by removing the heatsink. I would highly recommend that you NOT do
this.
If you want some help on this call me as this is what we do.
Len
-Len Whitwer
Puget Sound Data Systems, Inc.
19501 144th Ave. NE Suite D-100
Woodinville, WA 98072
e-mail mailto:l...@psds.com
Internet: http://www.psds.com
Toll Free: (866)857-0710
Tel: (425) 488-0710
Fax: (425) 488-6414
Marc Schlensog
LenW <l...@psds.com> wrote:
[...]
>
> Hi Marc:
>
> You run a VERY HIGH risk of burning up the original DS10 mother board
> by removing the heatsink. I would highly recommend that you NOT do
> this.
> If you want some help on this call me as this is what we do.
>
> Len
Hi Len,
thanks for the offer. I withdrew from this endeavor because of the risk
involved :) Now I still need to make up my mind what to do next.
Oh BTW: I checked the part numbers and the DS10l board is of a newer
revision.
b.r.,
Marc
LenW
Hi Marc:
Give me a call if you want as we have piles of Alphas here and can fix
you up
really CHEAP with a solution that makes sense to you.