ICD 7 Test Results

[ICD7]

Originally posted by: Fullmetal Chocobo
Excellent. I'm definitely ordering bolt-thru mounts for any HSFs that have push-pins. My two examples in the "bad pressure" section are proof enough for me. I also think it's about time to redo the TIM on my primary rig.

The IC diamond pressure results were an eye opener

My experience on my own copper synthetic 1 in. die which is flat // to .0005 and I have easy access top, side and back with no flexing MB or socket issues. This is a definite advantage when seeking perfect contact, even so I still have to wiggle the last couple of degrees alternately securing and loosening screws while monitoring temps to get the lowest number possible it usually takes me about 20 min before I am satisfied.

So at least from my observations it is a pretty challenging pursuit to get an optimal mount under the best of conditions. Many get there through multiple remounts taking the best temp they can get by "feel" and reworking when necessary which is what I do with my set up although compounded by complexity with an in system mount.

I have an old T bird sink that I use as a ref. with a single bent spring clip which only makes contact with on/over the exact center of the die with consistent/constant pressure. 4 years ago, today and tomorrow it will give me the same result. I had overlooked hardware as problem issue because my experience was lacking in the new stuff so the beta testing here has been valuable, again thanks to all participants.

I have learned alot on the hardware side as I have not really looked hard at hardware mounting and will be very useful in some sink development I am doing. An odd point here, a manufacturer could not have the best performing sink in a range of sinks but with a good hardware mount design could run the web site review circuit and average out better with good contact and balanced pressure than better performing sinks. I would imagine a best hardware mount should be a priority for the competitive sink market.

As noted the spread on these results run about 9C the difference between a great mount and a marginal one is the difference between a high end sink and an average one.

Taking a repetitive holistic approach. Temp results can range 9C factoring out the heat sink and just considering compound/contact/pressure on it's own.....Non trivial with the whole picture in mind.

 

[Fullmetal Chocobo]

I definitely feel spoiled by ICD7. I've been hunting around the house trying to find my tube of ICD7 for this install. I found all of my other TIM (all 6 different kinds), but I want the ICD7.

 

[Assimilator1]

Heheh, I know exactly where my collection is .

ICD7
I too learnt alot about the mounting issues, I was surprised by the range of issues that can occur.

Btw have you had all test results in? If not PM me the MIAs & I'll PM them to poke them .

 

[petrusbroder]

I can not resist a post in this old thread:

I have passed week done maintenance on my "old" P4 which has been running at stock 3.2GHz for almost 3 years 24/7.
When removing the heatsink-fan to remove all the dust from the fins I decided to use the knowledge from this thread: use ICD7 and to increase the preasure of the HSF on the CPU. The result:

Before: @ 3.2GHz, after one day 100% utilisation of the CPU: 58ºC
After: _ @ 3.2GHz, after one day 100% utilization of the CPU: 44ºC

Before: Max OC: 3.39GHz @ 60ºC
After: ___ _ OC: 3.84GHz @ 51ºC rock solid (no errors after 8 h of Prime and 16 hours of BOINC)!

Max OC not reached yet ... I am working on - each step (5MHz of the FSB now) takes some 24 hours.

These numbers are reached with the Intel Stock cooler!!
I have used some washers on the backside of the MoBo to increase the pressure.
I was also much more careful in applying the thermal interface compound.

 

[Rudy Toody]

Nice numbers, Peter!

I made a request to NewEgg to stock ICD7 a few weeks ago. So far, they haven't stocked it. Maybe if a few more ask for it, it will happen.

 

[petrusbroder]

Well, now I know the maximal OC for this chip: 4GHz. It booted well, everything runs well, but I got an error in Prime after approx 7 hours of crunching.

I have now backed it to 3.92GHz, which is a FSB-frequency of 245MHz. The temp of the CPU after 7 hours of Prime was 53ºC. I am somewhat surprised, considering that this is a Pentium 4, using the stock-HSF, but with ICD7 and an increased pressure on the CPU. The Mobo is a Asrock, quite new (hmmm, it is less than three years old). The RAM is the ValueRam from Kingston, and I run the memory bus @ 200 x 4 = 800MHz.

I am quite happy: the OC-increase is 22%, the Whetstones and Dhrystomes of BOINC have increased correspondingly.

 

[Assimilator1]

Petrus
Wow that's a massive improvement!:Q

I assume that's a Prescott based core then?

 

[petrusbroder]

Yes, it is ... Now it has run @ 245MHz x 16 = 3.92GHz for more than 20 hours - no problems at all. The CPU temp (as meaured by SpeedFan) is 52ºC. It is crunching MilkyWay@Home and all WUs have been validated OK.

Not too shabby for a CPU this old: with a very good thermal interface compound, some extra washers to increase the pressure between HSF and CPU.

I really ask myself if the difference between a good result and a bad one is much more a result of careful mounting and the right pressure than of the design of the HSF. You can do whatever fancy technological design and it will not cool better if it is not mounted optimally and if the heat stays with the CPU..
I would like someone to check all the different HFS (air cooled) and correlate their weight and/or mounting mechanism with the cooling result. I myself have too little technical know how to do such a evaluation ...

 

[ICD7]

I assume the ICD held up reliably for you you guys?

petrusbroder -"""I really ask myself if the difference between a good result and a bad one is much more a result of careful mounting and the right pressure than of the design of the HSF. You can do whatever fancy technological design and it will not cool better if it is not mounted optimally and if the heat stays with the CPU..
I would like someone to check all the different HFS (air cooled) and correlate their weight and/or mounting mechanism with the cooling result. I myself have too little technical know how to do such a evaluation ..."""

No matter the paste from the contact testing the key idea that emerged as you correctly noted was the contact and pressure. About 70%-80% saw an improvement out of the gate and of those that did not most were contact or pressure issues. In the contact testing we had a Non Trivial range of 9C between good and poor contact

Most of you guys know this instinctively when troubleshooting they typical forum post "I just mounted my heat sink and my temps are POOP" First response is usually " how is your contact?" You have to be something of a machinist/mechanic 30% of the time on problem mounts.

I am doing a design on a thermosyphon and keyed in on this point. A fail safe mount seems to me to be a priority after reviewing the data. Some mounting systems were consistently bad across all forums and petrusbroder's statement ""You can do whatever fancy technological design and it will not cool better if it is not mounted optimally and if the heat stays with the CPU.."" is an important point for me as there are probably a number good sinks on the market that failed to gain sales traction due to poor mounting designs

I added The content from our testing to our web site and most is data supplied from you guys Thanks again for all the effort.

Be a little forgiving on the design front as I am at the bottom of the learning curve when it come to HTML and web design and once I get the content and links squared away I will turn it over to a professional to smooth it out.

I will be doing some more forum beta testing and am mulling over some different tests I need done to fully round things out. In any event thanks again.

http://innovationcooling.com/index.html

 

[Fullmetal Chocobo]

My ICD7 has help up wonderfully thus far. And I haven't had any issues in removing it (replaced CPU) or anything like that. There was some staining of the HSF, but I took that as it was working correctly and getting into the small grooves.

 

[ICD7]

Not sure what the staining is about. There is carbon black in the mix and if you have ever played with copier toner you know what I am talking about, it gets every where. The only other ingredient are oils which are the same as what most others use, I would not want to get on your clothes but otherwise it is pretty inert stuff as far as metals go

The only other staining I have seen is where on a fresh copper surface on the contact area when mounted is airtight and stays fresh and the exposed area oxidizes so when you remove the sink the non oxidized copper contrasting with the oxidized area?

 

[ICD7]

For those that have an interest we are sponsoring another giveaway at OCAU
and results are starting to come in.
http://forums.overclockers.com...howthread.php?t=718724

 

[petrusbroder]

Thanks, that is an interesting read!

How is your designing a thermosyphon doing? Any progress? Any pics?

 

[ICD7]

How is your designing a thermosyphon doing? Any progress? Any pics?

It's more or less ready to go but I am a little skittish with the economy right now and being cautious with big project start ups. I have been all cash since august with all my money out of the US. Hopefully things will settle out in the next couple of months and we can have some confidence to move forward.

 

[ICD7]

OCAU test Results Averaging 3.71 Improvement. Application refinement with you people have helped bring up the averages


Updated results OCAU

 

[Rudy Toody]

I assume these exist: use a tiny torque wrench to tighten down the heatsinks. It would need to be done twice--cold, and after the TIM has spread during heating.

We could have an Over Torque forum. How much torque can mobo X handle vs mobo Y. With or without backing plates. Hot vs cold.

We will be trendsetters!

EDIT: our toolsets would now include a 10-second buzzer and a torque wrench. (Apply ICD7, buzz to spread, torque, heat to spread some more, torque.)

 

[ICD7]

That is one really great Idea! - Takes the guesswork out of it

Google and ye shall find -lots of different types, prices

http://www.toolsource.com/driv...e-wrench-p-103839.html

I will try it when I get some time - Intel specs from memory top end 75lbs?

 

[Assimilator1]

Bloody hell that one's expensive!:Q, why don't you go for just a manual one?, don't need anything flash
Do you think my work's torque wrench will be any good? it goes upto 140lb/ft

Btw the torque wrench idea is a good one for HSs with solid 'direct' screws but for those of use that have HSs held down by spring loaded screws it won't make any difference.

 

[ICD7]

torque wrench's even for the spring loaded sinks would be good cause you could at least see if it was in an optimal range.

I feel pretty stupid not to have tried that one already I work with torque wrench's all the time....

It would maybe make the case to do a mod to increase sink force

Also a handy tool for anybody doing a heat sink comparison/review, fixing that variable would make for a more 1 to 1 comparison if all sinks could be tested at the same pressure loads

There are literally hundreds of different kinds out there - Load em all up to 75 lbs probably drop a few degrees

 

[Rudy Toody]

I am building 2 water-cooled rigs and want to do this for the cpu, nb, sb, and mosfet on each. I already have to mod the mosfet and nb heatsinks. If I do the torque, I will need to create backing plates for the nb, sb and mosfet to prevent the mobo from bowing under the pressure.

It's been 48 years since I did anything with torque, so I don't remember the correspondence between the torque and the resulting psi.

They have preset torque wrenches and torque screwdrivers that could be used if we determine a good number for all four heatsinks plus video cards. That would take the guesswork out of it.

--Fred

 

[ICD7]

PSI = Force/area

so for 62.3 lbs and IHS area of 1.29? inches 62.3/1.29 = 48.3 psi

 

[Insidious]

The twisting force on the screw (torque) is directly proportional to the clamping force, but certainly not equal !!!!

You can't simplify the equation down to that. Different thread pitches, different clamping designs and materials.... etc...

-Sid

 

[ICD7]

Understood. I would recommend for simplicity and safety's sake to incrementally tighten to/for a suitable comfort level and generating performance curves. Getting a good balance between screws just as important as pressure. I could throw in some pressure film as guide to help troubleshoot any issues and to help in setting the maximum mount.

Intel I believe (but will recheck) has a max pressure of 75 psi with some margin. 60-65 psi should probably be the target

 

[Rudy Toody]

Originally posted by: Insidious
The twisting force on the screw (torque) is directly proportional to the clamping force, but certainly not equal !!!!

You can't simplify the equation down to that. Different thread pitches, different clamping designs and materials.... etc...

-Sid

I knew it had something to do with Mother Nature's best tool--the inclined plane.

This looks to be an interesting problem.

 

[ICD7]

The early AM crowd, we must be working round the clock.

we can calculate it and then check with the pressure film to validate torque calculated numbers. I will put together a test kit for you