Getting seriously weirded out here... (thermal interface related?) - Page 2

bigben2k · 08-18-2003, 01:36 PM

I gotta go with the fluke and/or burn-in theory proposed by mad mikee, or a probe failure/fault.

You ought to be able to re-mount the copper block, for a comparison: similar results will point to a burn-in, different results will point to a temp probe fault.

Cathar · 08-18-2003, 05:44 PM

Quote:

Originally posted by bigben2k
I gotta go with the fluke and/or burn-in theory proposed by mad mikee, or a probe failure/fault.

You ought to be able to re-mount the copper block, for a comparison: similar results will point to a burn-in, different results will point to a temp probe fault.

I intend to remount. I have had absolutely no time of late to spend on this. Less time even when considering the vagaries of thermal paste settling.

I'm pretty much of the opinion that it was just a freak case of very rapid thermal paste settling, as the final results from initial mount to the low point does concur with "2+ week old" AS3 mounts that I've had in the past, it's just that I've never had such a rapid occurrence of it before. I guess I was just surprised.

Ketchak · 08-18-2003, 08:11 PM

Cathar
If it serves him as something my experience I will tell him that I take enough time silver-plating my blocks. In the first place because I like the color of the silver, in second place because in my tests I obtain 1ºC less than with the copper block.
Why that difference?
I do not know it, but as always I obtain that grade less than difference I continue silver-plating them.
Never detect changes to better past the days, I don't concentrate a lot on it.
Certain it is that my model is much less effective that his and any added extra makes improve its performance a lot.

Possible symbiosis with AS3 or that with the silvering obtains a cleaner surface of contact.??????

Ketchak.
www.devilmaster.cjb.net

Since87 · 08-18-2003, 08:28 PM

Quote:

Originally posted by Ketchak

Possible symbiosis with AS3 or that with the silvering obtains a cleaner surface of contact.??????

I've wondered about that as well. Silver can be cold pressure welded, and I wonder whether some of the particles in the AS3 are getting welded to the baseplate. Probably take an electron microscope to tell...

BTW Cathar, can you tell me what the spring constant is for the springs you shipped with the original White Water.

SN# 17 is actually in service now. Finally.

Unfortunately I've got one of the P4's that gives absolute BS temp readings.

2.4C @3.24 1.675V
Load 42
Idle 39
Ambient 23

Cathar · 08-18-2003, 09:26 PM

Quote:

Originally posted by Since87
BTW Cathar, can you tell me what the spring constant is for the springs you shipped with the original White Water.

I asked for a rate of 5kg/cm. The original springs didn't have quite enough turns though, so unfortunately they do reach critical deformation at just before full compression. They are 20mm long and have a travel length of 1.4cm. Some of the very early Cascades shipped with the old White Water springs.

1 full turn of M4 thread has a travel of 0.8mm, so you can work out the pressure being applied fairly easily.

With the newer Cascade springs the rate is 6kg/cm. The wire used is thicker and with more turns so that deformation is not reached even at full compression. They are 24mm long and have a travel length of 1.3cm.

If you want I can ship you some of the newer springs. I may have to ship you slightly longer bolts too because the P4 blocks with the old 20mm springs didn't have a lot of thread left over, and the 24mm long new springs will probably mean that you won't be able to fit the wingnut on easily.

Since87 · 08-18-2003, 10:55 PM

Quote:

Originally posted by Cathar
If you want I can ship you some of the newer springs. I may have to ship you slightly longer bolts too because the P4 blocks with the old 20mm springs didn't have a lot of thread left over, and the 24mm long new springs will probably mean that you won't be able to fit the wingnut on easily.

No thanks. I took the IHS off. I don't need that much pressure.

Is there any reason to think, that the 'ideal' pressure for a P4 without IHS, is much different than for an AMD processor?

Althornin · 08-19-2003, 02:44 PM

Quote:

Originally posted by Since87
No thanks. I took the IHS off. I don't need that much pressure.

Is there any reason to think, that the 'ideal' pressure for a P4 without IHS, is much different than for an AMD processor?

"ideal"?
Higher pressure is always better for heat transfer.

Since87 · 08-19-2003, 04:10 PM

Quote:

Originally posted by Althornin
"ideal"?
Higher pressure is always better for heat transfer.

Bad choice of words on my part. Perhaps 'in the right ballbark' would be better.

Actually, what I want to know is; exactly how much pressure can I apply before my particular P4 die cracks? Although, I'd happily settle for Cathar's best guess at maximum pressure with a "good enough" safety margin.

bigben2k · 08-19-2003, 04:58 PM

Someone applied up to 100 lbs of clamping force, with no problems. It really comes down to how level you can keep that block.

If you go over the AMD specs, you ought to be able to approximate "something", by calculating the proportion of the sizes of the dies. AMD didn't change their specs with the latest cores, so there's obviously a hefty margin above it, just like the rest of the specs.

Cathar · 08-19-2003, 05:32 PM

Both AMD and Intel recommend at most 100lbs of static force, and 200lbs of dynamic force applied to the CPU for bare-die CPUs.

200lbs, of course, being the result of movement such as banging the case into a wall/door when carrying to to a LAN or something.

I think that's pretty much your answer. None of the springs I provide will give you as much as 100lbs when you compress all four to the maximum, so basically you're safe so long as you don't keep on screwing past maximum compression.

Most motherboards will flex dangerously long before there's enough pressure to crush the die. As always though, it's really about keeping the force that's applied very "flat".

Scila · 08-19-2003, 05:47 PM

Well, this is my first post here, and i´m about to add more wood to the fire!!!!!! but what the heck

I recently took my rig completly apart for cleaning and some basic modding.
So I took my vantec aeroflow out, and guess what I found:
The cpu left markings on it, I could see the lines of the cpu codes, impossible to read though.
I´m currently using a good old thunderbird 1200. I guess the high mounting pressure of the aeroflow, plus the CPU markings on the core, and the high temps of the setup all came together for this.
The mobo (yea, yea, crappy sensor) reads high 40´s to low 50´s , and the hs has been mounted for at least 3 months.

My point is, the original hypothesis is still valid, though more testing is needed to validate it, or any other.
To finish this already long post,
Does anybody now if silver is softer than copper at a certain temp?, What about when copper is 10-20C hotter?

have fun!

myv65 · 08-22-2003, 10:04 PM

Scila,

TBirds had laser engraving on the die that resulted in a surface resembling a mountain range in comparison to the flatness of the current XPs. Anyone that mounted a heatsink or block made from aluminum or copper AND that didn't use too much thermal compound would see results much like yours. I have a few old heatsinks carrying the lettering from various TBirds and a Duron.

The present XPs are extremely flat as are Cathar's blocks. While that's not to say that zero deformation takes place, it isn't nearly so likely nor obvious as it was with the TBirds.

Hey, and welcome to the forums.

jaydee · 08-22-2003, 10:31 PM

As to what myv65 was saying. They did the print ontop of the die on the earlier XP's aswell but stopped that and started using the black tag on the body instead. Here is my older XP1600+ and my XP1700+ Tbred. Note there is still an AMD logo on top of the die o nthe newer CPU's aswell but it is not nearly as defined as it used to be.

CheeseBall · 08-25-2003, 02:06 PM

I am running my copper block springless on my 1.4ghz tbird. So yes I have the stepping and corners of the die engraved in my block. W/ pure silver I'm curious to know how much of an endent the die would leave. I would also like to say that I have never notices a difference in temp from when AS3 was applied to a week later (w/ socket sensor 45C load 2.0vcore).

Sometime this week I'll be adding a 2100+, and a LCD temp diplay so it will be intersting to see what kind of readings I get.

jaydee · 08-25-2003, 03:13 PM

Quote:

Originally posted by CheeseBall
I would also like to say that I have never notices a difference in temp from when AS3 was applied to a week later (w/ socket sensor 45C load 2.0vcore).

I think it has more to do with how it is applied to the set in time. I never get a measurable difference either anymore. However I have at times noticed it when I apply it differently than I do now. Also I never use spings either but I don't install the blocks while the mobo is in the case either. Makes installation a hell of a lot easier.

08-18-2003, 01:36 PM	#26
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	I gotta go with the fluke and/or burn-in theory proposed by mad mikee, or a probe failure/fault. You ought to be able to re-mount the copper block, for a comparison: similar results will point to a burn-in, different results will point to a temp probe fault. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

08-19-2003, 04:58 PM	#34
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Someone applied up to 100 lbs of clamping force, with no problems. It really comes down to how level you can keep that block. If you go over the AMD specs, you ought to be able to approximate "something", by calculating the proportion of the sizes of the dies. AMD didn't change their specs with the latest cores, so there's obviously a hefty margin above it, just like the rest of the specs. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

08-18-2003, 08:11 PM	#28
Ketchak Cooling Neophyte Join Date: May 2002 Location: Spain Posts: 30	Cathar If it serves him as something my experience I will tell him that I take enough time silver-plating my blocks. In the first place because I like the color of the silver, in second place because in my tests I obtain 1ºC less than with the copper block. Why that difference? I do not know it, but as always I obtain that grade less than difference I continue silver-plating them. Never detect changes to better past the days, I don't concentrate a lot on it. Certain it is that my model is much less effective that his and any added extra makes improve its performance a lot. Possible symbiosis with AS3 or that with the silvering obtains a cleaner surface of contact.?????? Ketchak. www.devilmaster.cjb.net

08-19-2003, 05:32 PM	#35
Cathar Thermophile Join Date: Sep 2002 Location: Melbourne, Australia Posts: 2,538	Both AMD and Intel recommend at most 100lbs of static force, and 200lbs of dynamic force applied to the CPU for bare-die CPUs. 200lbs, of course, being the result of movement such as banging the case into a wall/door when carrying to to a LAN or something. I think that's pretty much your answer. None of the springs I provide will give you as much as 100lbs when you compress all four to the maximum, so basically you're safe so long as you don't keep on screwing past maximum compression. Most motherboards will flex dangerously long before there's enough pressure to crush the die. As always though, it's really about keeping the force that's applied very "flat".

08-19-2003, 05:47 PM	#36
Scila Cooling Neophyte Join Date: May 2003 Location: Costa Rica Posts: 6	Well, this is my first post here, and i´m about to add more wood to the fire!!!!!! but what the heck I recently took my rig completly apart for cleaning and some basic modding. So I took my vantec aeroflow out, and guess what I found: The cpu left markings on it, I could see the lines of the cpu codes, impossible to read though. I´m currently using a good old thunderbird 1200. I guess the high mounting pressure of the aeroflow, plus the CPU markings on the core, and the high temps of the setup all came together for this. The mobo (yea, yea, crappy sensor) reads high 40´s to low 50´s , and the hs has been mounted for at least 3 months. My point is, the original hypothesis is still valid, though more testing is needed to validate it, or any other. To finish this already long post, Does anybody now if silver is softer than copper at a certain temp?, What about when copper is 10-20C hotter? have fun!

08-22-2003, 10:04 PM	#37
myv65 Cooling Savant Join Date: May 2002 Location: home Posts: 365	Scila, TBirds had laser engraving on the die that resulted in a surface resembling a mountain range in comparison to the flatness of the current XPs. Anyone that mounted a heatsink or block made from aluminum or copper AND that didn't use too much thermal compound would see results much like yours. I have a few old heatsinks carrying the lettering from various TBirds and a Duron. The present XPs are extremely flat as are Cathar's blocks. While that's not to say that zero deformation takes place, it isn't nearly so likely nor obvious as it was with the TBirds. Hey, and welcome to the forums.

08-22-2003, 10:31 PM	#38
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	As to what myv65 was saying. They did the print ontop of the die on the earlier XP's aswell but stopped that and started using the black tag on the body instead. Here is my older XP1600+ and my XP1700+ Tbred. Note there is still an AMD logo on top of the die o nthe newer CPU's aswell but it is not nearly as defined as it used to be.

08-25-2003, 02:06 PM	#39
CheeseBall Cooling Savant Join Date: Jan 2003 Location: Colorado Posts: 176	I am running my copper block springless on my 1.4ghz tbird. So yes I have the stepping and corners of the die engraved in my block. W/ pure silver I'm curious to know how much of an endent the die would leave. I would also like to say that I have never notices a difference in temp from when AS3 was applied to a week later (w/ socket sensor 45C load 2.0vcore). Sometime this week I'll be adding a 2100+, and a LCD temp diplay so it will be intersting to see what kind of readings I get.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)