|
|
| Author |
Message |
n8nxf
|
|
Heatsink fans?
| 
Nov 17 03:08 UTC 1996 |
I'm wondering. Are those tiny little CPU fans really necessary? Do we
really need one more mechanical, motor driven, device to go bad in these
elcheap-o clone systems? From my observation these fans only add noise,
and more noise when their journal bearings wear out. They pack dust into the
fins of the undersized heatsinks they are mounted on. Then, to top it
off, many don't use a thermal coupling material/grease between the CPU and
heatsink/fan assembly! One system I looked at had the stores ID sticker
still adhered to the CPU, reducing the thermal coupling coefficient between
CPU and heatsink even further.
On a 486Dx/33 system I have I put a finned heatsink on the CPU that's
about the size of a heatsink/CPU combination. It runs warm, but not hot,
to the touch and I don't have the additional noise. Any other forced
cooled CPU observations out there?
|
| 12 responses total. |
arthurp
|
|
response 1 of 12:
|
Nov 17 06:08 UTC 1996 |
The Spec for a DX 33 never even mentioned needing a heatsink. a DX 50 is
required to have sink and fan. Many things beyond that use 3.3 volt instead
of 5 volt, so they dissipate less energy at the same clock speed. That is
how a Pentium 60 or 66 can run with a huge heat sink and no fan. Faster than
that and you need a fan. The new ones I work with at 166 with generous
coating of heat sink compound have a warm to the touch sink even while the
fan is running (not just like new, but *really* new ;) ). (I use a sink
and fan on my DX2 66, and I did the same on my DX 33. That way I don't feel
guilty about running my system in 80 degree ambient. And I also have a finned
sink on the S3 chip on my video accelerator since it was too hot to touch
before the addition.)
|
kentn
|
|
response 2 of 12:
|
Nov 17 07:09 UTC 1996 |
My 486DX2-80 runs fine (and relatively cool--you can touch it) without
the fan. It does have a rather monstrous heat sink, though. I suppose
some of the heat problems relate to the type of air flow through case
in which the CPU is mounted, the CPU's location in relation to other
forced air flow through the case, air flow blockages near the CPU, etc.
I've got a Pentium 150 at work, and I believe it has a CPU fan, though
that is expected.
|
ajax
|
|
response 3 of 12:
|
Nov 17 09:09 UTC 1996 |
I trust that fans really are a necessity for reliable operation of some
of the faster chips. Maybe I'm being gullible, but the last several 586
motherboards I've bought have had fans on them anyway, so I'd just as use
them.
|
arthurp
|
|
response 4 of 12:
|
Nov 17 22:44 UTC 1996 |
The DX2 80 would be a 3.3 volt chip. The analysis of power consuption gets
pretty conplex when you cahnge a chip that much. Assuming nothing but voltage
changes then you get 2/3 as much power used in a 3.3 volt. But with lower
voltage total charge transport goes way down, so current also drops off. They
use a lot less power. Airflow in the case makes a big difference, though.
There is a dual Pentium Pro 200 that uses no CPU fans. It has a duct that
directs flow from a case fan directly over the chips.
|
davel
|
|
response 5 of 12:
|
Nov 18 11:40 UTC 1996 |
For what it's worth, I know of a case where someone was careless and let a
cable get in the way of a fan's turning. The system died fairly quickly.
(This wasn't me, and was a while back, so I'm not sure if it was a 486 or
a pentium. It's also certainly possible, for all I know, that the non-
turning fan (and the cable, maybe) blocked enough airflow from the power
supply fan to make a difference.)
|
n8nxf
|
|
response 6 of 12:
|
Nov 18 15:10 UTC 1996 |
The 486DX/33 capable mother board I have has a plug right next to the CPU
for a fan. Since it's a 5v board with a 33MHz max clock speed, I assumed
a fan was at least recommended. This HP Vectra 486DX33N has a small fan
blowing on the CPU but no heatsink. You can't put put your finger on the
center of the CPU and count to ten without getting an urge to pull it off.
I also have a used P60 board and I found that the heatsink was loaded with
dust, making it pretty ineffective. I used a compressor to blow out dust
in such situations but a clean paintbrush also woks pretty well. (If you
blow the dust out of a PC that's been around for a while, do it outside
or you might be sorry.)
|
kentn
|
|
response 7 of 12:
|
Nov 18 17:52 UTC 1996 |
AFAIK, my 486DX2-80 chip is 5V since that's what my MB is (and there is
no jumper to select voltage that I can find).
I try to be careful about cable placement when putting the cover back
on the case. I've even tied them out of the way just to insure good
air flow. Even allowing for the fact I have a full-size AT case, it's
easy to drop a ribbon cable over the CPU chip if you aren't careful.
Seems to me I've seen cases advertised that had dual fans. I suppose
you load a full tower up with several hard disks and a Pentium chip, and
you need all the air flow you can get. Not personally at that point,
yet, though.
|
scott
|
|
response 8 of 12:
|
Nov 19 01:26 UTC 1996 |
You might want to be careful using a compressor to clean out a PC, since the
pressure can be high enough to blow chips out of their sockets.
|
arthurp
|
|
response 9 of 12:
|
Nov 19 04:22 UTC 1996 |
There were a few assorted faster 486 chips that were 5 volt. Pretty uncommon.
Chips tend to derate ate 150 C. If the outside of the package is more than
slightly warm I would increase cooling somehow. Make it last longer.
|
scg
|
|
response 10 of 12:
|
Nov 19 06:01 UTC 1996 |
My 386-33 is often somewhat cool to the touch. It doesn't have a heat sinnk,
and fairly obviously doesn't need one. My 486DX2-66 has a heat sink that gets
somewhat warm, but probably not quite to the point of needing a fan. It works
fine. Pentiums I've seen where the fan has stopped have had problems.
|
n8nxf
|
|
response 11 of 12:
|
Nov 19 15:17 UTC 1996 |
When I design electronic equipment that requires a fan, I'm always sure to
put a finger guard on both sides of the fan. One to keep fingers out and
one on the other side to keep wires out.
I use the rule of thumb a Willowrun Labs engineer told me about. If you
can't put your finger on a solid state component and cout to ten, it's
too hot!
|
arthurp
|
|
response 12 of 12:
|
Nov 20 04:52 UTC 1996 |
I would agree with that rule. I've used it before. Some of the faster
pentiums get hot so quickly that with your finger on it you turn it on and
you can't wait till 10.
|