PDA

View Full Version : Cooling units on Pentium chips



Raven
September 22nd, 2010, 04:36 PM
I'm confused as to what systems need and do not need cooling units in the Pentium line - I've seen systems run without them, but I'm not sure if it's intentional or if they just broke or fell off at some point before I got my hands on them.

Specifically, I have a 100Mhz chip in a Socket 5 board and the cooling unit won't stay on, I was wondering if you need it on a system like this?

wolfie
September 22nd, 2010, 04:46 PM
i had a 75Mhz chip that i don't remember having a cooling fan on th heat sync. i think it is probably better to have a fan there and not need it rather than need it and not have it.

Raven
September 22nd, 2010, 04:57 PM
Does one definitely need a heatsink, then? When I say "cooling unit" I meant the whole heatsink and/or fan combo - this box has none at the moment but it's just been stored for a few months, so hasn't been operating since it lost it.

Unknown_K
September 22nd, 2010, 05:02 PM
I think they pretty much all need them (and some should even have it written on the top like AMD does). Some OEMs didnt use a fan but had some super large heatsinks and maybe a case fan blowing over it (like the early Packard Bell P66 machines). There could be special embedded versions that got away with small heatsinks and no fans (I have a TI 486/100 that didnt have anything on the chip and it was meant for industrial designs).

Chuck(G)
September 22nd, 2010, 05:03 PM
Every P1 system that I've ever seen with a CPU from 75MHz on up has had a heatsink of some kind on the CPU.

Raven
September 22nd, 2010, 05:04 PM
Alrighty then, that answers that - thanks.

Ole Juul
September 22nd, 2010, 06:37 PM
I've got a P5 goldtop (60MHz) which has no heat sink but a small fan which I added when it overheated.

I also have an Intel Pentium 133MHz which came with a fairly large heat sink and attached fan. I unplugged the fan and it still works like a charm on hot 35 C (95 F) days. It is an open frame so there is no case fan. It has been on 24/7 for a couple of years now.

Chuck(G)
September 22nd, 2010, 07:01 PM
I've always liked HP's way of handling P1/P2/P3 cooling--use a heatsink, but leave off the fan--instead use a duct from the main (large) case/PSU fan to direct air over the heatsink. The headsink fins don't plug up with dust and the machine is deathly silent.

Raven
September 22nd, 2010, 07:04 PM
It's just not easy to apply that theory to any given case, which is unfortunate - you have to find a computer with that sort of system in place.

glitch
September 22nd, 2010, 08:11 PM
Another trick I've used to get Pentium machines to run fanless is to use a P3 or Athlon XP heatsink and remove the fan. They're designed for a much higher thermal load, and usually stay close to room temperature unless your case just has no air movement. Copper-core aluminum Athlon XP heatsinks (the generic ones that came with the XPs and were more or less completely insufficient) do a really nice job.

EDIT: and if you're having trouble with heatsink clamps, grab some Arctic Silver thermal /adhesive/. Use four dots (sparingly!) on the corners of the processor and fill the space in between with regular thermal paste, and you'll be able to separate the heatsink from the processor if you need to. You can use the adhesive across the whole processor, just like regular thermal paste, but you seriously risk damaging the processor if you ever need to pop the heatsink off. Also, thermal tape or "frag tape" works, but not as nice of a finish.

Chuck(G)
September 22nd, 2010, 08:20 PM
EDIT: and if you're having trouble with heatsink clamps, grab some Arctic Silver thermal /adhesive/. Use four dots (sparingly!) on the corners of the processor and fill the space in between with regular thermal paste, and you'll be able to separate the heatsink from the processor if you need to. You can use the adhesive across the whole processor, just like regular thermal paste, but you seriously risk damaging the processor if you ever need to pop the heatsink off. Also, thermal tape or "frag tape" works, but not as nice of a finish.

Call me old-fashioned, but I use good old Permatex anti-seize compound (the silver kind). There are several tests using this showing that it's no worse than most silver pastes--and it's non-conductive. To hold a heatsink onto the ZIF socket, I simply fashion a clip out of a 15 ga. stainless steel bicycle spoke.

glitch
September 22nd, 2010, 08:32 PM
Call me old-fashioned, but I use good old Permatex anti-seize compound (the silver kind). There are several tests using this showing that it's no worse than most silver pastes--and it's non-conductive. To hold a heatsink onto the ZIF socket, I simply fashion a clip out of a 15 ga. stainless steel bicycle spoke.

Indeed, I use white beryllium oxide thermal paste on most of the "higher performance" stuff I play with (large MOSFETs and such), simply because we get it in gallon cans at work. Apparently contact with the eye causes you to go blind in a horrible, painful way though. We also use it on heatsinks with computer processors at work, and it seems to perform as well as Arctic Silver 5 (hardware monitors report around the same temps...we've not actually done tests on the two).

I do like Arctic Silver's thermal adhesive though, for cases where you can't really attach a heatsink with clamps -- the best example of that being an overheating PC/104 board I've got, which uses one of those 486-compatible System-On-Chip modules. I'm pretty sure its original enclosure involved part of the chassis contacting the top of the SoC for heatsinking, but there's nowhere to bolt a heatsink to. So, I stuck an old southbridge heatsink to the top!

Unknown_K
September 22nd, 2010, 08:39 PM
Cheap radioshack heatsink compound is what I use (and very little of it). Quite a few Pentium1-3 heatsink/fans use a thermal strip so you don't need any heatsink compound. Luckily I snagged a dozen or more new P1 and P3 heatsink/fan combos a few years back when they were dumped so I have stock (wish I did the same for 486's).

The P60/66 were 5V and ran very hot, if the fan that pulled cold air over the large heatsinks (glued onto the chip most of the time) failed so did the CPU.

saundby
September 22nd, 2010, 08:41 PM
For P1s I usually use passive cooling unless I'm overclocking. A 486 aluminum heatsink will usually fit a Pentium, and the wire clippy will usually work, too. If I'm overclocking I put on a fan, or if the original system included a heatsink/fan combo then I'll use that. One of my 166s has a fan, the other doesn't.

Monitor your temps, run a demanding program. Add a fan if necessary.

Ole Juul
September 22nd, 2010, 08:51 PM
For P1s I usually use passive cooling unless I'm overclocking. A 486 aluminum heatsink will usually fit a Pentium, and the wire clippy will usually work, too. If I'm overclocking I put on a fan, or if the original system included a heatsink/fan combo then I'll use that. One of my 166s has a fan, the other doesn't.

Monitor your temps, run a demanding program. Add a fan if necessary.

I wasn't sure if a 166 would be OK w/o a fan. That's good to know.

I think you hit the nail on the head when you mention load. People are obviously going to have different ideas of what is "normal". I would never in a million years run a game so that wouldn't enter my mind, whereas a gamer would take it for granted and create a lot of CPU load.

Is there a way to measure temperature without some sort of special, and probably expensive, probe placed under the CPU?

Chuck(G)
September 22nd, 2010, 09:08 PM
Is there a way to measure temperature without some sort of special, and probably expensive, probe placed under the CPU?

Many DMMs come with at least the capability to read type K thermocouples and many come with one. In any case a type K bead-style 'couple will run less than $10.

Ole Juul
September 22nd, 2010, 10:07 PM
Many DMMs come with at least the capability to read type K thermocouples and many come with one. In any case a type K bead-style 'couple will run less than $10.
Gotcha. I was thinking one would have to use a special solid riser or something, but of courser there's a little space underneath. I wonder how close that reading would be to what the chip manufacturer quotes. Does the 'couple need to be glued or pushed against the CPU and at what place does one get a legitimate reading?

Chuck(G)
September 22nd, 2010, 10:29 PM
Here's a bit from Intel on the subject. (http://www.intel.com/cd/channel/reseller/asmo-na/eng/165924.htm).

Personally, I'd just wedge the "bead" between the heat sink and the chip top. You might want to use a copper plate with a machined-out portion to fit the thermocouple.

Neon_WA
September 22nd, 2010, 11:58 PM
from Chapter 10 of the Pentium Development Manual


10.1. MEASURING THERMAL VALUES
To verify that the proper TC (case temperature) is maintained, it should be measured at the
center of the package top surface (opposite of the pins). The measurement is made in the same
way with or without a heat sink attached. When a heat sink is attached, a hole (smaller than
0.150” diameter) should be drilled through the heat sink to allow probing the center of the
package. See Figure 10-1 for an illustration of how to measure TC.
To minimize the measurement errors, it is recommended to use the following approach:
Use 36-gauge or finer diameter K, T, or J type thermocouples. Intel’s laboratory testing
was done using a thermocouple made by Omega (part number: 5TC-TTK-36-36).
Attach the thermocouple bead or junction to the center of the package top surface using
high thermal conductivity cements. The laboratory testing was done by using Omega
Bond (part number: OB-101).
The thermocouple should be attached at a 90-degrees angle as shown in Figure 10-1.
The hole size should be smaller than 0.150” in diameter.

as passive heat-sinks were the norm then.. the probe was vertical to the chip face.
Now active heat-sinks are the norm to test processors they make a groove in the chip heat-spreader to fit the thermocouple

MV75
September 25th, 2010, 05:29 AM
For socket 4 motherboards (pentium 60/66), I figured out the cooling solution that best fits and is actually available. (try finding a socket 4 heatsink fan, you won't).

That is, a pentium pro heatsink/fan. I had my poor little socket4 board for ages, but couldn't really use it due to lack of cooling, but one day I got a dual ppro and sized it up, prefect fit. So I bought a cheap atx board that came with a ppro200 just for the heatsink. The width is perfect, I just had to bend back the "long" side of the clip arms. Fit the socket clips perfectly and clamps well to the cpu. Bonus is the ppro heatsink is nice and large.

Chuck(G)
September 25th, 2010, 08:03 AM
One aspect of cooling in older systems that is often not addressed is airflow within the case itself. A heatsink and fan isn't going to do much good if there's no way to get the warm air displaced outside the case somehow. Because there was no standardized layout for older boards, this can sometimes be a problem.

MV75
September 25th, 2010, 02:40 PM
That's what the power supply and fan are for. :)

But luckily those back plate slot coolers with squirrel fans are easy enough to get too for cases with no rear fan mounts. :)

Chuck(G)
September 25th, 2010, 03:10 PM
That's what the power supply and fan are for. :)

Yeah, right. :rolleyes: Everyone knows that power supplies are engineered for maximum case ventillation. :sarcasm:

Unknown_K
September 25th, 2010, 03:21 PM
I tend to use towers and have one fan blowing air in at the bottom and another sucking air out under the PS (or above it in bigger AT towers) so there is air flow (as long as I clean the grill of dust every year or two). Even with bad airflow a fan will still remove heat better then just a heatsink, that extra removed heat will just heat up everything else in the box. You create airflows just from heat rising from the heatsink and cooler air dropping. If you want to talk stupidity why are video card heatsinks and fans facing down instead of up!

Tetrium
October 5th, 2010, 11:06 AM
I found this webpage to be very usefull as a start for finding out how much heat one particular processor may output.
http://mysite.verizon.net/pchardwarelinks/elec_pentium.htm
Note that underclocking a P1 166 (non-mmx) to 100Mhz may actually run cooler then the original P1 100Mhz as the 166 was made in a smaller process, thus producing less heat ;) (Well, in theory atleast :P )

Minerva10210
October 16th, 2010, 09:56 PM
Hmmm, I'm a bit late to the party on this thread, but I thought I'd add my 2c and a funny story as well! ;)

It must have been circa 1995/1996 when my old man man got a P60 mobo and chip, and at this stage is where I started fiddling with the innards of PC's...
I recall seeing a DIP switch or jumper pin that stated 60/66MHz. I set the jumper to 66MHz, not really knowing at that point that it was in fact my first attempt at overclocking, which worked fine for years after that. At a much later stage, I fiddled with the board for some reason, and I recall stripping the fan/cooler to clean the fan and sink out a bit. Obviously my overclocking created a bit of extra heat in the CPU, and had just bout burnt the casing of the fan where it fitted to the sink, made nice parallel shiny lines on the fan casing :D

So, my long winded story is just to enforce my personal opinion and experience - if it is a standard CPU, 486 @ 50MHz and up, I will put a sink and fan combo on it... I would rather overkill on the cooling than have too little. Even on my latest machines, if any temp goes > 50*C I start shopping for fans/coolers/etc.

Cheers! :)

Ole Juul
October 16th, 2010, 11:30 PM
I found this webpage to be very usefull as a start for finding out how much heat one particular processor may output.
http://mysite.verizon.net/pchardwarelinks/elec_pentium.htm
Note that underclocking a P1 166 (non-mmx) to 100Mhz may actually run cooler then the original P1 100Mhz as the 166 was made in a smaller process, thus producing less heat ;) (Well, in theory atleast :P )

Good find. Yes, I've always been a fan of underclocking.

The charts are also interesting in that they give the maximum case temperature. I note that for a Pentium Classic, they quote 70 C which would be hard to reach under normal circumstances. Where I live, the ambient temperature never gets above 40 C so a case fan wouldn't even serve any purpose. Personally, on older computers like that, I don't use the case anyway, so cooling would only be needed if the house was on fire. :) I note too that some chips can have a maximum case temperature of 95 C which would be unreachable in any environment inhabitable by humans. IOW, cooling is irrelevant in most "cases".

MV75
October 17th, 2010, 12:00 AM
I'm more a fan of under volting than under clocking. :)

Tetrium
October 18th, 2010, 02:41 PM
..... I simply fashion a clip out of a 15 ga. stainless steel bicycle spoke.

Could you please explain to me how you do this exactly? I find this info very interesting as those clips are excellent for holding heatsinks to ZIF sockets and those clips are quite tough to find!

Chuck(G)
October 18th, 2010, 03:43 PM
Could you please explain to me how you do this exactly? I find this info very interesting as those clips are excellent for holding heatsinks to ZIF sockets and those clips are quite tough to find!

Take a look at a typical Socket 7 ZIF--along the centerline of the socket, there are two rectangular bosses on either side that are used by those 2-piece flat spring clips. On either side of these, there are usually two more bosses--rectangular but with rounded bottoms. The trick to making a clip using a bicycle spoke and bending it in sort of a three-dimensional "Z" shape so that each end slips under the rounded bosses.

If that's not clear enough, I can take some photos.

http://upload.wikimedia.org/wikipedia/commons/thumb/1/16/Socket_7.jpg/544px-Socket_7.jpg

MV75
October 19th, 2010, 03:14 AM
I use one of these:

http://www.google.com.au/imgres?imgurl=http://images.tomshardware.com/2005/11/21/the_mother_of_all_cpu_charts_2005/pentiumpro_platform_big.jpg&imgrefurl=http://www.tomshardware.com/reviews/mother-cpu-charts-2005,1175-19.html&h=768&w=1024&sz=210&tbnid=S4NGvSx4N2sXLM:&tbnh=113&tbnw=150&prev=/images%3Fq%3Dpentium%2Bpro&zoom=1&q=pentium+pro&hl=en&usg=__A8OrVZdV6GZwh46ASOmT9fZ2HFA=&sa=X&ei=2ny9TL-GM5OEvAOT1ukm&ved=0CDUQ9QEwBg

on my socket 4 board. I mentioned it earlier, but since it has a similar off center clip like the above, I thought I'd throw this back on out there. The hsf I'm talking about is the one on the left with the large mc on it. Pretty common ppro hsf. Notice how the top assembly is all together? I just bent back the "long reach" (left in picture that's over the socket cam), clip arms to compensate for the shorter socket. Spacing width was pretty spot on.