[Skinnee Labs] The TIM Thread...

Quote:

---

Originally Posted by Vapor

PK-1 and MX-4 are already done, PK-1 was a knockout winner. :p:

Shin-Etsus are months away, I'm trying to alternate batch sizes of 2 and 3...Shin-Etsu will be a 3-TIM batch. If I'm on batch 0 right now, Shins are going to be batch 6, roughly 3ish months away. I'm curious to see if X23-7783D can keep up with PK-1 at moderate contact, my hunch is "no" based on how well PK-1 thinned out. I have no idea what to expect from X23-7762 and G751, but based on Shin-Etsu's own data, I'm not expecting much.

And thanks guys, each TIM takes 150 hours of CPU loading, so it's good to hear people are appreciating it :)

As for the IX issues, make sure the board is level, make sure your block isn't on too tightly (about the same pressure you'd get from a stock Intel heatsink or an Apogee XT or an EK Supreme HF Easy Mount, I suppose), and make sure your heatsink/block has a smooth base (no steps, no HDT, avoid superbows). I do two reflows every 7-10 days for my CPU block tests, and I don't do any tricks to get it to work. I'd say 5 out of 6 reflows are completely successful on blocks I've never done before....and basically a 100% success rate with ones I have done prior.

Miah, the one in your pictures is almost certainly from not being level. :(

Utnorris, yours does sound like too much mounting pressure. I make sure I tighten down the same amount on all four corners, then the non-bulge side I'll loosen slightly if there's a bow and tighten it when the PCMA (solder) is liquid. Considering the mounting pressure and bow I used for Great Contact setting, I wasn't expecting it to work, but it did (a look at the used ETIs looks bad, but temps were right on).

Yeah, thinking too much pressure.

I don't really think there's any trick to it....but I got it right the first time and since then I've just tried to stay the course. Occasionally I'll try to do something a little differently, but I never find any benefit when I do. I can imagine if I didn't do it right the first time I'd be trying to overcorrect and be generally uncomfortable with it. Fortunately the Apogee GTZ I did it with first was very amenable to the process (right amount of pressure, copper base is a little more reflow friendly, bow wasn't too extreme). It's definitely the hardest TIM application out there and if you know you have really good contact, I don't know if I would say it's worth it over a PK-1 (or X23-7783D from my other tests).

I'm surprised no one else is very alarmed by the MX-3 batch differences. Assuming IX is .5c away from 'perfect,' new batch MX-3 (both tubes purchased when MX-4 released) is 125% worse than original batch MX-3 (purchased at MX-3 launch) with Poor Contact. I've used (with full-on monitoring) multiple batches of Shin-Etsus, MX-2, Ceramique, PK-1, IX, and AS Matrix and there hasn't even been a hint of a difference. Original batch MX-3 was the best of any paste with Poor Contact, new batch MX-3 was one of the worst. Even more alarmingly, the TIM prints made me think pump out was only a few thermal cycles away from happening.

---

If there was such a difference on the batch of MX3, then I am wondering if there will be the same issue with MX4. If I read your review correctly, you only had one tube of MX4, so it might be worth picking up a second tube from a different source to see if the batches are consistent. I have been happy with mine, but i was using AR Ceramique before.

LOL I bought some "generic" X23-7783D off of ebay I hope it's the real thing - the guy's feedback was excellent though.
http://cgi.ebay.com/5g-Thermal-X23-7...19601823935139
Planning on remounting where IC-7 was on my 4.8ghz 2600k.
If this doesn't work out I'll pick up some PK-1 =)

Quote:

---

Originally Posted by Utnorris

If there was such a difference on the batch of MX3, then I am wondering if there will be the same issue with MX4. If I read your review correctly, you only had one tube of MX4, so it might be worth picking up a second tube from a different source to see if the batches are consistent. I have been happy with mine, but i was using AR Ceramique before.

---

I had 3 tubes of MX-4, surely all from the same batch though (bought at the same time, right after launch). Used one and a third tubes or so (11/4 split for mounts, IIRC) for these TIM tests....there's no noticeable discrepancy in MX-4 performance so far.

I kind of wonder if this was why MX-3 was EOL'd. Original batch MX-3 was performing and behaving Heat Buster-esque with Poor Contact, extrapolate that out and you have very little scaling with contact (much better than MX-2 at poor contact with a disappearing advantage as contact improved). Heat Buster (a.k.a., Original Batch MX-3 Extrapolated) is slightly better than MX-4, so it's conceivable that they knew they had production problems with MX-3 and brought out MX-4 not to improve on MX-3, but just to correct the problems.

I hope what I'm calling "new batch MX-3" has very little reach in the market.

A conspiracy theorist approach is that they sandbagged new batch MX-3 to make MX-4 look good, but that would be basically illegal (misrepresenting goods).

@LagunaX, X23-7783D has lots of shapes and sizes. It's used in nearly every Mac and probably a lot of HP/Dell/etc systems and with a large enough order they all get to pick their packaging. I wouldn't be surprised if that's legit X23-7783D.

I wonder if these batches are mixed in big industrial mixer bowls of sorts and someone was sleeping on the job and forgot to scrape the bowl that day to keep the mixture consistent...

I thought I only did that when I "Attempt" to try and cook something..

At one time I used to have a job where I had to climb down into a glue mixer at a plywood mill. Not exactly tools of precision I can say that much and not a job I wanted to keep for long either..

Quote:

---

Originally Posted by Martinm210

I wonder if these batches are mixed in big industrial mixer bowls of sorts and someone was sleeping on the job and forgot to scrape the bowl that day to keep the mixture consistent...

I thought I only did that when I "Attempt" to try and cook something..:D

At one time I used to have a job where I had to climb down into a glue mixer at a plywood mill. Not exactly tools of precision I can say that much and not a job I wanted to keep for long either.

---

I doubt it's manually managed, though it probably is made in big vats, I'd imagine they use electrical mixers for consistency and it was a formulation issue...that's just a guess though.

I just remember vividly cleaning the glue vats and spreaders of the plywood mill. We actually used to use tubs of grease to coat parts to prevent the glue sticking for the next time we had to do duty and clean that nasty mess. Who knows what sort of contamination could actually occur at the end of dispensing a batch where you're getting the less than perfect mixture.

Not many users wouldn't know any better unless they specifically were testing TIM materials.

Quote:

---

Originally Posted by Martinm210

I just remember vividly cleaning the glue vats and spreaders of the plywood mill.

---

Yea, you never forget something like that. I was on an assembly line last summer and I hated every second of it, but it's SO vivid in my memory.

Quote:

---

Originally Posted by Vapor

I'm surprised no one else is very alarmed by the MX-3 batch differences.......



.....MX-3 was EOL'd.

---

If it had not been EOL'd, then it'd be a cause for concern IMO ;)

It might be interesting to add another "TIM" to the list: nothing.

For that test, it might be best to skip the "poor contact" test, but moderate and great contact tests should be safe/okay - it couldn't be any worse than the process of reflowing indigo xtreme.

The reason? I'm starting to think that it might be better to use no TIM at all when compared to some of the compounds I've encountered. Seriously, some of these so-called "thermal interface" materials might very well be acting as more of an insulator!

Take care
gary

Quote:

---

Originally Posted by garyd9

It might be interesting to add another "TIM" to the list: nothing.

For that test, it might be best to skip the "poor contact" test, but moderate and great contact tests should be safe/okay - it couldn't be any worse than the process of reflowing indigo xtreme.

The reason? I'm starting to think that it might be better to use no TIM at all when compared to some of the compounds I've encountered. Seriously, some of these so-called "thermal interface" materials might very well be acting as more of an insulator!

Take care
gary

---

Hmmmm, interesting. I do doubt that any of the TIMs on my list have higher thermal restriction than air though.

FWIW, when I do reflow for IX, I back down to like 2600MHz at 1.1v....a far cry from 3800MHz at 1.4v (rough math says testing load is greater than 2.25x the load of reflow load). I've accidentally tried to get into Windows at full OC with a fresh ETI and it just crashes (this is with the pumps on, of course). But the ETI also lifts the base off the IHS somewhat....so not a full indication of TIM-less performance.

Because of the concern for damage I have, TIM-less testing will have to be put aside for the very end when I do CLP/CLU/ICDiamond (i.e., the group of TIMs that will permanently alter the IHS and/or the blocks).

i was told i couldnt ever use indigo extreme...

my system cant put the required load @ cpu end unless i turn off my pumps. lol..

and i really really dont like having my system on without the pumps.

It be like me purposely turning off my backups for my backups..

I would like to try indigo extreme, but i would propably toast my gtx480 sli setup that runs from same pump as cpu XD

Hey Vapor,
Gabriel Torres @ Hardware Secrets seems to agree with your about the PK-1:
http://www.hardwaresecrets.com/artic...ay-2011/1249/5
Dunno about the MX-4 in his testing though - hard to believe AS-5 would beat it...

Quote:

---

Originally Posted by NaeKuh

i was told i couldnt ever use indigo extreme...

my system cant put the required load @ cpu end unless i turn off my pumps. lol..

and i really really dont like having my system on without the pumps.

It be like me purposely turning off my backups for my backups..

---

You would be amazed at how resilient Intel chips are. I ran my 875K for 11 minutes @ 99c (that was before I knew what to do with IX..) LOL and my chip is just as good today as it was about a year ago when I did that reflow. Same vcore to get 4.4GHz today as it took a year ago…

Any chance of testing the GELID line of TIM?

Does anyone else spread their TIM over entire surface with card/plastic bag+finger?

I did in last build. But wondering results in spreading vs drop in middle. Or the X.

Which is best way in your opinion?

Quote:

---

Originally Posted by barfastic

Any chance of testing the GELID line of TIM?

---

GELIDs are on the docket....but months away.

Quote:

---

Originally Posted by Gamb311

Does anyone else spread their TIM over entire surface with card/plastic bag+finger?

I did in last build. But wondering results in spreading vs drop in middle. Or the X.

Which is best way in your opinion?

---

For my own uses and in testing, I only do bead in the middle....anything else introduces air to the TIM joint too readily, IMO.

(n.b., if a TIM has instructions for an application other than a bead in the middle, I try my best to follow the instructions....the Deep Cool trio I'm wrapping up are such examples)

Quote:

---

Originally Posted by Vapor

For my own uses and in testing, I only do bead in the middle....anything else introduces air to the TIM joint too readily, IMO.

---

Thanks for that tidbit :)

I'm going to be re-doing my loop in the next couple of weeks, and I might give something new (for me) a go (using MX-4 at the moment).

Can't use Indigo Xtreme (tempted as I am) because I only have one big loop and I'm not confident with turning the pumps off and leaving the MB and the GPU without cooling...

However the Prolimatech looks like a feasible candidate :) Do we know if it's "safe" for GPU dies? (well, more like "safe" if a small amount spills on the SMCs around the core). Also, how easy is to spread? (I usually spread the TIM where there's no IHS).

I've used PK-1 on VRMs...worked fine. (to think it was before I knew it was any good....had a packet of it and needed some TIM on my motherboard VRMs, would have put it on the CPU if I had known better)

And finally the reason AS5 and Ceramique testing was delayed: a new TIM from Arctic Silver, Ceramique 2 (http://www.performance-pcs.com/catal...ducts_id=30781)

Quote:

---

Originally Posted by Vapor

I've used PK-1 on VRMs...worked fine. (to think it was before I knew it was any good....had a packet of it and needed some TIM on my motherboard VRMs, would have put it on the CPU if I had known better)

And finally the reason AS5 and Ceramique testing was delayed: a new TIM from Arctic Silver, Ceramique 2 (http://www.performance-pcs.com/catal...ducts_id=30781)

---

Vapor,
Have you done any similar tests with TIM's regarding GPU's?

No need, the results wouldn't be different.

The testing would be a lot more difficult though...sensors aren't as good, cards don't seem to be as robust (testbed durability and longevity might be an issue), GPU block mounting doesn't seem great on any product, flat loading looks to be more difficult.

If you know how TIM Dependent a block+GPU combo is, you can estimate gains/losses from switching TIMs. Ultimately, if a TIM is good on CPUs, it'll be good on GPUs.

Quote:

---

Originally Posted by Vapor

I've used PK-1 on VRMs...worked fine. (to think it was before I knew it was any good....had a packet of it and needed some TIM on my motherboard VRMs, would have put it on the CPU if I had known better) ... snip ...

---

Thanks for the reply :) I've just ordered some, and according to the comments in Skinee's test it's rather liquid so should be easy to spread :)

Updated with Deep Cool Z3, Z5, and Z9.

Also introduced a new chart, Overall Average Temperature (weighted):
http://skinneelabs.com/wp-content/up...verallTemp.png

Same data that goes into the "Relative Performance Score" chart, but in raw temperature format, not rescaled to a unitless score :)

Arctic Silver 5, Ceramique, and Ceramique 2 are being tested now :D

Do you have any Thermalright Chillfactor 3 to test? It came with my TR Venomous X and I don't feel like spending $10 for some new paste :D

Vapor: if manufacturer's suggested method of application is clearly inferior, why use it instead of your time proven bead in center method? Imho this roundup is more about TIM paste relative performance, not how bad one might perform applied by manufacturer instructions?
(imho one might do few test mounts of manufacturer suggested methods just to check if some particular paste really isn't better off with them, but as soon as it shows on average worth results - revert and perform all the tests with correct application)