Intel p4 retail heatsink

[Max Coppin]

I'm about to re-attach my heatsink to my CPU and have just started to clean
them up. On the bottom of the heatsink is a metal pad that did have a thick
black substance on it that I've cleaned off (and off the CPU). Should I
remove the metal sticker-type thing on the heatsink or leave it on there? I
plan on using a silver compound.


Thanks for your help!

 

[Wayne Youngman]

I'm about to re-attach my heatsink to my CPU and have just started to clean
them up. On the bottom of the heatsink is a metal pad that did have a thick
black substance on it that I've cleaned off (and off the CPU). Should I
remove the metal sticker-type thing on the heatsink or leave it on there? I
plan on using a silver compound.

Hi,
that black thing is the original TIM. Remove it and give the base of the
heatsink a good clean. You could also lap it a bit. Then apply some
quality thermal compound to the surface of the CPU/heatspreader and
carefully install the whole lot.

The TIM is meant to be *one-shot* as far as I know, and if you remove the
heatsink once installed INTEL recommend you use a new TIM.

 

[MC]

Thanks for the rapid reply, what does TIM stand for and where can I get a
new one? Should I just use paste instead?

 

[David Schwartz]

Thanks for the rapid reply, what does TIM stand for and where can I get a
new one? Should I just use paste instead?

Thermal Interface Material. If you paste and the CPU surface and heat
sink surface already fit together nicely, you should not use anything but
the paste. However, if there are any gaps at all you will need some TIM
other than paste! Paste only works between two flat surfaces that mate
smoothly over their full surface area.

DS

 

[sbb78247]

Thanks for the rapid reply, what does TIM stand for and where can I
get a new one? Should I just use paste instead?

Yes, I think it is a one shot use too. It works so well that when I pulled
the heatsink off, it yanked the cpu out of the socket! Luckily, I put it
back in with the new heatsink and all is well.

Shannon

 

[JB]

I'm about to re-attach my heatsink to my CPU and have just started to clean
them up. On the bottom of the heatsink is a metal pad that did have a thick
black substance on it that I've cleaned off (and off the CPU). Should I
remove the metal sticker-type thing on the heatsink or leave it on there? I
plan on using a silver compound.


Thanks for your help!<

The Intel retail solution works fine. Intel says the unit is not resuable
if you remove the material provided.

I am running a 1.8 at 2.75 with their solution and the cpu reaches a max of
47-48 C under prolonged full load with a room temp of 24 C.

BTW, a later model Intel heatsink, which I am using, has a copper core and a
different fin structure than came with the 1.8.

 

[Nate Edel]

The Intel retail solution works fine. Intel says the unit is not resuable
if you remove the material provided.

I've had no trouble reusing the retail heat sink, as long as the thermal pad
was replaced with grease (which I ended up doing the first time I installed
it as the pad came with a giant scratch.)

I am running a 1.8 at 2.75 with their solution and the cpu reaches a max of
47-48 C under prolonged full load with a room temp of 24 C.

BTW, a later model Intel heatsink, which I am using, has a copper core and a
different fin structure than came with the 1.8.

I'm using the original from a P4/1.8, but then my overclocking is a lot more
conservative; I'm running the 1.8 at ~2ghz.

 

[Slug]

I've had no trouble reusing the retail heat sink, as long as the thermal pad
was replaced with grease (which I ended up doing the first time I installed
it as the pad came with a giant scratch.)

Yea, I always remove the pad anyway and use thermal paste instead.

 

[Dorothy Bradbury]

Heatsink to processor is an interesting area:
o There is only about 1% contact between the two
o 99% is interstitial air gaps - a poor conductor of heat

Two main approaches:
o Thermal compound/grease:
---- Best thermal interface + flow well + allows dissassembly
---- Disadvantage is pump-out, causing voids & hot-spots over time
o Thermal Interface Pad
---- Doesn't suffer pump-out + fitted by manufacturer + no mess
---- Disadvantage is one-use, and degraded performance if re-used
------ there is also a higher attachment pressure re mechanical stress on BGA

New pads come close to thermal compound, so the difference that was
once 4-5oC is now commonly around 2-3oC - not a lot for non-o/clockers.
A dry-joint is ~2.9oC/W, grease 0.9oC/W, compound around 0.9oC/W.

Don't put too much heatshink compound on if you use it, it is to fill in the
microscopic voids in the metal's surface - it isn't like making toast

 

[Kevin Miller]

Yes, I think it is a one shot use too. It works so well that when I pulled
the heatsink off, it yanked the cpu out of the socket! Luckily, I put it
back in with the new heatsink and all is well.

Shannon

Heh -- that happened to me a few weeks ago with an Intel 2.8 HT, but I
wasn't as lucky as you were. Several pins on the cpu were bent and
broken. It really ticked me off to have to order a new one that
night...

Kevin Miller

"Either way, it is bad for Zathras."

 

[Sept1967]

Ditto. I have never used the black gummi "stuff" that comes with the retail
Boxed heatsink. Artic Silver all the time.

I believe Intel means the stock black pad, is one-time use. You wouldnt be
able to pull off a used stock heatsink/pad and use it on another install, as
the pad would already be melted, and probably stuck to the original cpu.

 

[Triffid]

The Intel retail solution works fine. Intel says the unit is not resuable
if you remove the material provided.

Unfortunately the 'unit' that is not reusable all too often turns out to
be the CPU, which tends to suffer bent and/or broken pins when the user
attempts to remove a heatsink which has been glued to the processor by
the 'Intel retail solution'.

Mind you, I've developed some expertise in straightening and replacing
CPU pins - and demand for my services is increasing, so Intel's
'solution' is not entirely a bad thing

 

[Dorothy Bradbury]

Mind you, I've developed some expertise in straightening and replacing

CPU pins - and demand for my services is increasing, so Intel's
'solution' is not entirely a bad thing

You do know about the future CPU/motherboard interface?
o Pins on chips are replaced by pins on the board
o Chips then have solder pads to make contact with the m/b pins

So far, many report that the motherboard-pin based solution have
a low duty cycle, with relatively ease of bending pins & wrecking
the CPU &/or motherboard. The pins I guess aren't mechanically
braced like they are when they enter a Ziff socket.

The reason for the change is two-fold:
o Higher pin density due to higher current & pin count of future CPUs
o Eliminating the relatively high cost of the Ziff socket itself

So you may find your market grows somewhat

 

[Slug]

Unfortunately the 'unit' that is not reusable all too often turns out to
be the CPU, which tends to suffer bent and/or broken pins when the user
attempts to remove a heatsink which has been glued to the processor by
the 'Intel retail solution'.

Mind you, I've developed some expertise in straightening and replacing
CPU pins - and demand for my services is increasing, so Intel's
'solution' is not entirely a bad thing

I had an experience with a bent cpu pin recently and it was all
Intel's fault. Their vacuum sealed packaging of the cpu is a pain
to to take apart. I cut all around the edges of the seal with scissors
but it was still stuck together somewhat so I had to reaf on it a bit.
That caused the cpu to fly out of the package and land on the counter.
I inspected the cpu and all seemed fine, should have used a magnifying
glass to check the pins because one pin was bent and I didn't see it,
installed the cpu into my mb and it shorted the mb out. Had to send
the mb back to Asus and bought a new Abit mb. Found the bent pin and
straightened it out and all worked fine in the new mb. When I got the
Asus mb back (free of charge except shipping) I used it to build a
second PC so now I have two PC's and it's all thanks to a bent cpu
pin.

 

[sbb78247]

Heh -- that happened to me a few weeks ago with an Intel 2.8 HT, but I
wasn't as lucky as you were. Several pins on the cpu were bent and
broken. It really ticked me off to have to order a new one that
night...

Kevin Miller

"Either way, it is bad for Zathras."

Totally screwed isn't it. I was happy the damn thing worked again since it
would have been a really bad time to be down. Makes you wonder why they use
that crap. If you remove it, it voids the warranty, since you are not using
the "original" equipment. The stock heat sink is nice for just that - stock
use. But pure copper seems to work nice, and water even better. Kind of
like the old muscle cars, no one left them alone because there was more
hidden power to be found.

Shannon

 

[sbb78247]

Ditto. I have never used the black gummi "stuff" that comes with the retail
Boxed heatsink. Artic Silver all the time.

I believe Intel means the stock black pad, is one-time use. You wouldnt be
able to pull off a used stock heatsink/pad and use it on another install, as
the pad would already be melted, and probably stuck to the original cpu.

No probably about it, it is. Good and bad since it works, but if you remove
it, you have a fair chance of ruining a CPU!

S

 

[KR Williams]

You do know about the future CPU/motherboard interface?
o Pins on chips are replaced by pins on the board
o Chips then have solder pads to make contact with the m/b pins

Are you talking about LGAs (Land Grid Arrays)? LGA ~ BGA^-1.
These have been around for a *long* time (in "web-years" terms
anyway). PeeCees haven't used them because they're not a great
fit with the PeeCee market (stick the high-value processor on the
junk board as, or after, it leaves the distributor).

 

[George Macdonald]

Heatsink to processor is an interesting area:
o There is only about 1% contact between the two
o 99% is interstitial air gaps - a poor conductor of heat

Two main approaches:
o Thermal compound/grease:
---- Best thermal interface + flow well + allows dissassembly
---- Disadvantage is pump-out, causing voids & hot-spots over time
o Thermal Interface Pad
---- Doesn't suffer pump-out + fitted by manufacturer + no mess
---- Disadvantage is one-use, and degraded performance if re-used
------ there is also a higher attachment pressure re mechanical stress on BGA

New pads come close to thermal compound, so the difference that was
once 4-5oC is now commonly around 2-3oC - not a lot for non-o/clockers.
A dry-joint is ~2.9oC/W, grease 0.9oC/W, compound around 0.9oC/W.

The recent retail CPU TIMs I've seen are not pads at all and don't need the
high pressure for efficiency that pads needed. Hmm, I'd say they are more
like a thickish paste. It's my understanding that the CPU may run very
slightly higher than optimal temps for the first few hours of use as the
paste becomes more liquid and spreads with the heat. There is then a
curing process where the stuff hardens into something which is like a hard
plastic at ambient temps.

BTW I've also seen regular (good quality) heatsink grease/paste which has
shown no evidence of thermal pump-out after ~4 years of daily use. I was
surprised - it was a K6-2/500 and when I released the spring clip, the CPU
slid around on the slight cushion of "grease" just like it did when it was
originally fitted. This was not an excess of grease.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Dorothy Bradbury]

Yes the new pads are a great improvement on the older ones,
and are a 1-shot cure-when-first heated interface material too.

Heatsink paste pumping is for real, but it is also quite minor.
I think it's mainly a problem for LT applications + vibration +
probably larger outputs than CPUs (eg, big Mosfet arrays).

Only time I've seen it is on old power supplies - where the
electrolytics have begun to go & I've upgraded/replaced them.
In getting to them I've removed mosfets and noticed it, later
PSUs (especially PCs) use a grey insulatory material so the
primary SMPS heatsinks aren't live (but don't assume that

More often than pump-out is dryout, but that too is readily
avoided with even the lowest quality of paste these days.

 

[Dorothy Bradbury]

Are you talking about LGAs (Land Grid Arrays)? LGA ~ BGA^-1.

These have been around for a *long* time (in "web-years" terms
anyway). PeeCees haven't used them because they're not a great
fit with the PeeCee market (stick the high-value processor on the
junk board as, or after, it leaves the distributor).

Yes, and yes they have been around a relatively long time.

However, they aren't so user-friendly:
o Most PCs never see the CPU removed
---- eg, branded PCs or many corporate PCs
o SOHO PCs do tend to see the CPU removed
---- and some segments see extensive CPU changes

Thus far, reports indicate that the proposed system is only good
for a very low number of few fit & removal cycles. Even then,
the cycles are subject to considerable risk during the fitting with
bent pins & damaged motherboard &/or CPU being a reality.

It may come down to the clamping solution design I guess.
The Intel P4s was nice re thro-board, easier lever than skt370,
but I disliked the amount of stress on the thickness reduced
motherboard re track stresses. Never heard of a failure tho.