Innovation Cooling's Diamond 7 TIM test results

[littleowl]

ok guys I have went out and played with my E6600 some today.
It is at stock speed and on stock air for 3 week. My
Load Temps Are:
Core 0: 29c
Core 1: 27c

Room temp was 14.4C

I got it all changed over to a temp Water loop with the 6002 on it.

Load Temps after about 5 min were:
Core 0: 23c
Core 1: 21c

Room Temp up to 17.2c

here are some pics.

[[XC] riptide]

Nice Macro pics there.

[tastymannatees]

Nice Macro pics there.

No kidding I can see his fingerprint on the bracket corner

[tastymannatees]

Here's a lapp and edge job, little disappointed he did not get before numbers

http://www.ocforums.com/showthread.php?t=557899

[123bob]

Farm-09 Q6600 on an IP35 Pro. "Edge lap" before and after. All with D7 paste.

Refer also to my posts #202 and #417 for more info on this same machine. I'll try to provide the full story here, but who knows......it's been a long week already!

I'll hold my temp data until the end of the post so it's all together.

Here was the original pressure test films from before, out of post #417. It was pretty clear that this quad had the "island, moat, high edge" that most quad IHS's show.



I blued up the CPU with "poor man's" Dykem blue, a blue sharpy marker. (Dykem blue is something gunsmiths use to check contacts. SiG and SS probably know about this...)



I took a few flat strokes on the 1/2 solid copper bar I use to lap. (This thing is dead flat as shown by using a high grade machinists rule, and holding it up to bright light.) I did this to see the intial profile better. The same rule on the CPU showed that it indeed had the "island, moat, edge" profile, as shown by one of Tasty's diagrams. Where the blue is gone is where the high spots are.



So, I went about figuring out how best to take down the edges since I had the high spot profile. I re-blued the thing to get another baseline and took the CPU by the edges with a slight tilt to the copper bar. I worked a few strokes at a time and rotated the CPU to the next edge. The tricky part was getting it even. I actually think it's easier to lap flat than to try this technique. But hey, this is science, right? Here is a pic after the rough edge lap. Note that it has the reverse of blue color left on it than the first profile pic.



Of course, no self respecting lap job can go without finer grits and polishing. A good lap job is one where the camera takes a picture of itself, out of it's own vanity... I had to try it, so I used D7 as the polish. (What the heck, I already tasted the stuff, it tasted like it would be good polish... ) It worked quite well...as you would expect out of diamond compound.



So, the next test was to see the difference in the pressure test. Comments after the pic.

I thought it was a curious result. I may break it down and try it again. I don't know what's up with the diagonal white line down the left of the middle, or the white spot in the lower center. Does the film perhaps have a dead spot in it? It definitely has a bias to the front of the CPU. I'm also curious now what a flat lap would show.

Looking with the machinist ruler, at a bright light, the TRUE is slightly convex, and this CPU is now a bit convex. I would have expected a solid patch in the middle. I'm doubting now that the TRUE spring screws put consistent pressure all the way around the mount. The order that I've put in the screws has been the same in all testing on this machine. Lower left first, then upper right, then lower right, then upper left. Maybe this result makes sense with that pattern.

OK, now to temps.

Before Lap - D7 paste.
Ambient 20.3
CT idle 24,26,24,24
CT load 40,44,38,40

After Lap - D7 paste.
Ambient 17.5
CT idle 21,22,19,19
CT load 37,40,35,36

Ambient delta of 2.8 degrees C. Picked up some improvement on idle, picked up only a little improvement at load. I'm not sure what to make of this other than to try another film on this setup, then lap flat, film it again and re-run temps. Tasty, let me know your thoughts. I would not be against lapping both sides on this to see what happens....

Another thought. I'm beginning to wonder how linear the CPU temp diodes, and/or Coretemp is performing. I will attempt to better control my ambient variation when I take test points.

Regards,
Bob

[littleowl]

very interesting!

[Emerica]

Farm-09 Q6600 on an IP35 Pro. "Edge lap" before and after. All with D7 paste.

Refer also to my posts #202 and #417 for more info on this same machine. I'll try to provide the full story here, but who knows......it's been a long week already!

I'll hold my temp data until the end of the post so it's all together.

Here was the original pressure test films from before, out of post #417. It was pretty clear that this quad had the "island, moat, high edge" that most quad IHS's show.



I blued up the CPU with "poor man's" Dykem blue, a blue sharpy marker. (Dykem blue is something gunsmiths use to check contacts. SiG and SS probably know about this...)



I took a few flat strokes on the 1/2 solid copper bar I use to lap. (This thing is dead flat as shown by using a high grade machinists rule, and holding it up to bright light.) I did this to see the intial profile better. The same rule on the CPU showed that it indeed had the "island, moat, edge" profile, as shown by one of Tasty's diagrams. Where the blue is gone is where the high spots are.



So, I went about figuring out how best to take down the edges since I had the high spot profile. I re-blued the thing to get another baseline and took the CPU by the edges with a slight tilt to the copper bar. I worked a few strokes at a time and rotated the CPU to the next edge. The tricky part was getting it even. I actually think it's easier to lap flat than to try this technique. But hey, this is science, right? Here is a pic after the rough edge lap. Note that it has the reverse of blue color left on it than the first profile pic.



Of course, no self respecting lap job can go without finer grits and polishing. A good lap job is one where the camera takes a picture of itself, out of it's own vanity... I had to try it, so I used D7 as the polish. (What the heck, I already tasted the stuff, it tasted like it would be good polish... ) It worked quite well...as you would expect out of diamond compound.



So, the next test was to see the difference in the pressure test. Comments after the pic.

I thought it was a curious result. I may break it down and try it again. I don't know what's up with the diagonal white line down the left of the middle, or the white spot in the lower center. Does the film perhaps have a dead spot in it? It definitely has a bias to the front of the CPU. I'm also curious now what a flat lap would show.

Looking with the machinist ruler, at a bright light, the TRUE is slightly convex, and this CPU is now a bit convex. I would have expected a solid patch in the middle. I'm doubting now that the TRUE spring screws put consistent pressure all the way around the mount. The order that I've put in the screws has been the same in all testing on this machine. Lower left first, then upper right, then lower right, then upper left. Maybe this result makes sense with that pattern.

OK, now to temps.

Before Lap - D7 paste.
Ambient 20.3
CT idle 24,26,24,24
CT load 40,44,38,40

After Lap - D7 paste.
Ambient 17.5
CT idle 21,22,19,19
CT load 37,40,35,36

Ambient delta of 2.8 degrees C. Picked up some improvement on idle, picked up only a little improvement at load. I'm not sure what to make of this other than to try another film on this setup, then lap flat, film it again and re-run temps. Tasty, let me know your thoughts. I would not be against lapping both sides on this to see what happens....

Another thought. I'm beginning to wonder how linear the CPU temp diodes, and/or Coretemp is performing. I will attempt to better control my ambient variation when I take test points.

Regards,
Bob

Wow Bob, real nice results with that lap. Some people never notice a difference in lapping, but you definitely did.

[tastymannatees]

Another thought. I'm beginning to wonder how linear the CPU temp diodes, and/or Coretemp is performing. I will attempt to better control my ambient variation when I take test points.

Great job you put a lot of effort into this test much appreciated.

The film impressions are great and making me think. The initial print had the loads moat wise center then edge. The second print has the loads redistributed inboard from the edge but less pressure on center so may have a kind of averaging effect.

Sooooo......... If you have good contact at the start on center maybe on this moat kind of a profile you just take a flat bastard file to to the edges to increase pressure in the center region only and leave the center region untouched. Maybe relieving as much to leave the only contact area over the die as the only possible contact area with the sink to even further concentrate pressure on center

Another option is you may be at be at an optimum contact with the die from the start so any changes could be would be merely minimal.

In my experience when I make large changes as you did with the lapping and see no or minimal changes there is usually a third dominant factor somewhere that is masking result.

It's good you ran the test with ICD7 before and after. The Idle vs load readings in coretemp are totally at odds with with physics.

ICD7 contains 92% diamond some carbon black and a couple % synthetic oil. All the materials are stable through the range of heat you guys use with a quadruple margin. It will not change in thermal resistance across the range of temps. The same holds true for any compound I ever tested, C/W is C/W it does not change it is a constant we can now use

So you have a 2C margin improvement@ idle, say your are running 50W in idle (I do not know your idle Watts were just approximating) with a TIM resistance on the Die of .01-.02 C/W At 100W the margin can only increase (C/W is C/W) the spread on temps would now be in the range of 2-4C @ 150W 3-6C @ 200W 4-8C.

Resolution on your test gets better the higher you go. Your original interface resistance before lapping it dropped 1-2C and was probably in the 02-.04 C/W range so on idle you have approx. a 50% improvement on the thermal interface which would hold through the range

Your numbers are actually going in the opposite direction which is contrary to anything in my experience as how a thermal grease performs. I've seen this on a number of those in the marginal improvement category.

I would suspect that it would be more likely subtle variations in the diodes as I have both great and marginal results on all software programs which is a great call on your part.

Overall my bet would be that if you had a thermocouple on the set up you would have seen a 4-5C drop overall. Testing is hard work, thanks

Edit: Compare load and Idle tests on this test before and after. Idle dropped by a significant amount and load remains unchanged The delta temp can only go up with increased watts
http://img528.imageshack.us/img528/9...ltsimagwv4.png

[123bob]

What really puzzles me is that I DID leave the center high spot unchanged. I took off only the edge profile. Seeing that the CPU and TRUE are both convex, I would have expected to see a spot in the middle with the film....I will use another film on this, before I do another step. That will leave me with one film left for other testing.

What do you think my next step should be? Should I lap both CPU and TRUE as flat as I can get? (hehe, I still have a shot at winning the "max contact contest" here... Right now, it looks like Bail is still in the lead...)

I do find is curious that the temp change compression is greater at the load values than idle. It does make me wonder about the diode/software linearity.

Trying to get a handle on the variables and attempting to control them has proven to be a daunting task in this exercise....

Regards,
Bob

[tastymannatees]

What really puzzles me is that I DID leave the center high spot unchanged. I took off only the edge profile. Seeing that the CPU and TRUE are both convex, I would have expected to see a spot in the middle with the film....I will use another film on this, before I do another step. That will leave me with one film left for other testing.

What do you think my next step should be? Should I lap both CPU and TRUE as flat as I can get? (hehe, I still have a shot at winning the "max contact contest" here... Right now, it looks like Bail is still in the lead...)

I do find is curious that the temp change compression is greater at the load values than idle. It does make me wonder about the diode/software linearity.

Trying to get a handle on the variables and attempting to control them has proven to be a daunting task in this exercise....

Regards,
Bob

As regards to lapping, It occurs to me that the film only shows where the two parts make contact so the void area could be .0005 or a mile. One idea is that it might be pretty close to the particle size. I will have to talk to the pressure film rep.

Paste impression may reveal void thickness

The next thought might be removing the excess material now allows for a little "Flex" under pressure. This is sheet metal, put 50 psi on it and and you might get a little ripple across the surface, variation might only be .001 -.002 but still appear as a void on the film. I missed that part about leaving the center intact so maybe the bubble relaxed is a ripple under pressure?

Another point- I have not looked it up but Intel has some tolerance spec somewhere on IHS flatness that probably would correlate with a TIM thickness requirement. You are never going to mate to parts closer together than the particle size of the compound. I could envision having a void that when filled with compound the particle size would be higher then the non compounded contact. That's a thought to chew on The contact areas become voids? Maybe we should be sandblasting the IHS Instead?

I am getting to the point I believe the issue is one of tolerances, mechanical and electrical. No matter what you manufacture it requires some kind tolerance spec, a range in which it functions. No manufacturing process is perfect it is always a compromise.

When Intel produces a batch of processors the are not all the same some are 3.2's others are 2.8's etc. and are sorted into separate bins according to performance, none are perfect. Diodes are the same, they function in a engineered range of use, some read higher some read lower and some may or may not be perfectly linear across a range but they perform the intended job. I believe that picturing them as a drop dead temp indicator might be like leaning on a reed and maybe the design intent is just to keep users from burning up their CPU?

123bob, it would be interesting to go for the better contact. You own a Temp meter? I could cut a groove for a thermocouple. It would be interesting to take a no result CPU and compare it to one that has a optimum result vs diode measurement.

[SiGfever]

Farm-09 Q6600 on an IP35 Pro. "Edge lap" before and after. All with D7 paste.

Refer also to my posts #202 and #417 for more info on this same machine. I'll try to provide the full story here, but who knows......it's been a long week already!

I'll hold my temp data until the end of the post so it's all together.

Here was the original pressure test films from before, out of post #417. It was pretty clear that this quad had the "island, moat, high edge" that most quad IHS's show.


I blued up the CPU with "poor man's" Dykem blue, a blue sharpy marker. (Dykem blue is something gunsmiths use to check contacts. SiG and SS probably know about this...)


I took a few flat strokes on the 1/2 solid copper bar I use to lap. (This thing is dead flat as shown by using a high grade machinists rule, and holding it up to bright light.) I did this to see the intial profile better. The same rule on the CPU showed that it indeed had the "island, moat, edge" profile, as shown by one of Tasty's diagrams. Where the blue is gone is where the high spots are.


So, I went about figuring out how best to take down the edges since I had the high spot profile. I re-blued the thing to get another baseline and took the CPU by the edges with a slight tilt to the copper bar. I worked a few strokes at a time and rotated the CPU to the next edge. The tricky part was getting it even. I actually think it's easier to lap flat than to try this technique. But hey, this is science, right? Here is a pic after the rough edge lap. Note that it has the reverse of blue color left on it than the first profile pic.


Of course, no self respecting lap job can go without finer grits and polishing. A good lap job is one where the camera takes a picture of itself, out of it's own vanity... I had to try it, so I used D7 as the polish. (What the heck, I already tasted the stuff, it tasted like it would be good polish... ) It worked quite well...as you would expect out of diamond compound.



So, the next test was to see the difference in the pressure test. Comments after the pic.

I thought it was a curious result. I may break it down and try it again. I don't know what's up with the diagonal white line down the left of the middle, or the white spot in the lower center. Does the film perhaps have a dead spot in it? It definitely has a bias to the front of the CPU. I'm also curious now what a flat lap would show.

Looking with the machinist ruler, at a bright light, the TRUE is slightly convex, and this CPU is now a bit convex. I would have expected a solid patch in the middle. I'm doubting now that the TRUE spring screws put consistent pressure all the way around the mount. The order that I've put in the screws has been the same in all testing on this machine. Lower left first, then upper right, then lower right, then upper left. Maybe this result makes sense with that pattern.

OK, now to temps.

Before Lap - D7 paste.
Ambient 20.3
CT idle 24,26,24,24
CT load 40,44,38,40

After Lap - D7 paste.
Ambient 17.5
CT idle 21,22,19,19
CT load 37,40,35,36

Ambient delta of 2.8 degrees C. Picked up some improvement on idle, picked up only a little improvement at load. I'm not sure what to make of this other than to try another film on this setup, then lap flat, film it again and re-run temps. Tasty, let me know your thoughts. I would not be against lapping both sides on this to see what happens....

Another thought. I'm beginning to wonder how linear the CPU temp diodes, and/or Coretemp is performing. I will attempt to better control my ambient variation when I take test points.

Regards,
Bob

So you kind of partial to the little stickers on that camera?

[tastymannatees]

I did get one more today from VA so whatever more come in tomorrow will go out tomorrow as the first batch hopefully the batch will be a little larger by the time it goes out.

They will send me the result as pics in an excel file by email which I will then forward off to the test recipient as soon as I get it.

[123bob]

Another point- I have not looked it up but Intel has some tolerance spec somewhere on IHS flatness that probably would correlate with a TIM thickness requirement. You are never going to mate to parts closer together than the particle size of the compound. I could envision having a void that when filled with compound the particle size would be higher then the non compounded contact. That's a thought to chew on The contact areas become voids? Maybe we should be sandblasting the IHS Instead?

An interesting thought....

When Intel produces a batch of processors the are not all the same some are 3.2's others are 2.8's etc. and are sorted into separate bins according to performance, none are perfect. Diodes are the same, they function in a engineered range of use, some read higher some read lower and some may or may not be perfectly linear across a range but they perform the intended job. I believe that picturing them as a drop dead temp indicator might be like leaning on a reed and maybe the design intent is just to keep users from burning up their CPU?

123bob, it would be interesting to go for the better contact. You own a Temp meter? I could cut a groove for a thermocouple. It would be interesting to take a no result CPU and compare it to one that has a optimum result vs diode measurement.

I would have to agree that Intel's interest is in not burning up the CPU rather than using the diodes for some kind of quantitative analysis. Otherwise they wouldn't go through such pains to tell folks how to measure temps on their products. None of which involves using the internal diodes...

I do not own a temp meter. The internal diodes are the best I can do here.

I think I'll tear the thing apart and re-film it to see how consistent the contact is. Then for the last film, perhaps use some TIM on it and see what it looks like? That will burn up the last of my film and then I'll send all 5 samples back to you. Sound like a plan?

Regards,
Bob

[tastymannatees]

An interesting thought....



I would have to agree that Intel's interest is in not burning up the CPU rather than using the diodes for some kind of quantitative analysis. Otherwise they wouldn't go through such pains to tell folks how to measure temps on their products. None of which involves using the internal diodes...

I do not own a temp meter. The internal diodes are the best I can do here.

I think I'll tear the thing apart and re-film it to see how consistent the contact is. Then for the last film, perhaps use some TIM on it and see what it looks like? That will burn up the last of my film and then I'll send all 5 samples back to you. Sound like a plan?

Regards,
Bob

I would not use the film with paste. It basically shows the pressure spread is uniform and not much else whether the contact is bad or good. Just shows it is doing it's job. Better to save for more comparative stuff. Your farm 7 Had a a good result? better to compare a good to a questionable one to pick up the advantages or disadvantages.

[123bob]

I would not use the film with paste. It basically shows the pressure spread is uniform and not much else whether the contact is bad or good. Just shows it is doing it's job. Better to save for more comparative stuff. Your farm 7 Had a a good result? better to compare a good to a questionable one to pick up the advantages or disadvantages.

OK, will do. (Just caught up with your PM too.)

I'll use film #4 to retest Farm-09, as it is, with no paste. This time, I'll put the screws on in reverse order than I normally do. If we see the film bias to the top, instead of the bottom, we will know it has something to do with the TRUE mounting method. I would also like to see if that wrinkle, or line, comes through again.

The last film #5, I'll throw on Farm-07 tonight. This one had the very favorable result against Shin Etsu. It would be worth seeing where that comes out.

OK, starting work now.

Regards,
Bob

[123bob]

So you kind of partial to the little stickers on that camera?

I just KNEW someone would pick that up...... I bought the camera about 7 years ago, and just never got around to taking the stickers off....The important part is did you notice the Carl Zeiss lens on that thing? Really sweet...

So, how am I doing Doc? I haven't bought a new machine in about 1.5 months now. I'm starting to get the itch.... The X3350 was nice, but I need those Nehalems to start showing up...

Regards,
Bob

[123bob]

Farm-07 Pressure test results

(Sue me, I'm triple posting. I'll quadruple post after the next one. It's the only way I can keep my data straight.... )

This machine is a Q6600 running on an EVGA 680i. It has a Zalman 9700 for cooling. Refer to post #167 for thermals. It had good results with D7 vs Shin Etsu G751.

Broke down machine for pressure test so I may as well add the D7 spread pics first. Here they are



Not too bad looking. This mount performed very well.

OK, here's the pressure test.....

I was off a bit on placement and missed the very top edge.

About all i can say is WHAT WHIMPY MOUNT PRESSURE!! But you know, it performed well against the Shin......

Regards,
Bob

[123bob]

Farm-09 2nd Post Edge Lap Pressure test.

Farm-09 is all over this thread. It's a Q6600 on an IP35 Pro with a TRUE. It was "edge lapped". Here's pressure test number two.

Note - the important difference in this pressure test is the order I put the screws in on the TRUE. I went Upper Left, Lower Right, Upper Right, Lower Left. This is reverse of the first test. The result proved interesting......

First, edge lapped D7 from breakdown...EDIT: Note that this spread pattern is from the original screw order of Lower Left, Upper Right, Lower Right, Upper Left.



We can see a bias toward the top, reasonable spread with heavy contact in the middle.

Here's the 2nd pressure test.

Wow, quite different from the first. More in-line with what I would expect based on the lap profile and the TRUE screw mount order. Note the bias is toward the top on this, which makes sense to me given the screw order. This has great contact in the middle, without the puzzling diagonal white line that test #1 had. Might have to run one more on this if I have spare film.

Well, it's late and I'll let the pics speak for themselves.

Regards,
Bob

EDIT: Ah cr*p. I mispelled "LAP" in the last pic. It's permanent sharpy, so it will have to remain.....

[tastymannatees]

You may win the best contact contest with that one

[littleowl]

You may win the best contact contest with that one

NNNNNNOOOOOOOOOOO I Will win this one LOL I never win anyways

[123bob]

You may win the best contact contest with that one

It's all in the name of science, LOL. It was a pretty good print though....

I'll be dropping the films in the mail today.

Sorry Littleowl....

Bob

[Bail_w]

NNNNNNOOOOOOOOOOO I Will win this one LOL I never win anyways

you can lap your IHS and heatsink then do your film test again

[tastymannatees]

10 samples went out for processing today.

[SiGfever]

I just KNEW someone would pick that up...... I bought the camera about 7 years ago, and just never got around to taking the stickers off....The important part is did you notice the Carl Zeiss lens on that thing? Really sweet...

So, how am I doing Doc? I haven't bought a new machine in about 1.5 months now. I'm starting to get the itch.... The X3350 was nice, but I need those Nehalems to start showing up...

Regards,
Bob

Well since we know that you are waiting on Nehalem to arrive and you have been doing such great work with all this testing you are right on track to either a nervous breakdown (quad jonesing) or will discover the true meaning of contact pressure versus the every changing Intel IHS.

Oh, I just pulled the stickers off my wife's little Nikon (purchased two years ago) and when she finds out that I did not remove the goo she will be pissed.

I really like the Carl Zeiss lenses.

[tastymannatees]

Well since we know that you are waiting on Nehalem to arrive and you have been doing such great work with all this testing you are right on track to either a nervous breakdown (quad jonesing) or will discover the true meaning of contact pressure versus the every changing Intel IHS.

Oh, I just pulled the stickers off my wife's little Nikon (purchased two years ago) and when she finds out that I did not remove the goo she will be pissed.

I really like the Carl Zeiss lenses.

It's kinda like ever diminishing concentric circles of Chaos theory. I may be signing up for therapy myself soon. The minutia is getting to me

123bob if nothing else you are the most consistent guy I have ever seen. Looking over the farm 9 results you have 9 tests. At load averaging all cores with three different pastes except for the one AS5 which on a remount along with 7 other pastes tested within about 1/2 a degree from each other with the IHS lapped and unlapped. I know you noted the compression issue but first time I had a chance to look close at he numbers. pretty remarkable