IC Graphite Thermal Pad Available for Test and Review

interesting. since when is LM made of water?

 

.2mm Soft PGS didnt get a good mount as I had the little ones hover over me during reassembly

 

The hydrophobia, yet oliophilic, nature may create a surface which makes it difficult for LM to attach to the pad. The precise reasons related to this may be up for study and debate. Likely, it is the size of molecules at play that may be the explanation on what is being observed. But, considering the tests show no degradation in LM performance, it may be worth exploring in lapping situations where tolerances are not as tight as hoped or for less than optimal surface contact, such as a concave IHS without a perfect convex heatsink to match, such as a flat heat sink. Need to think on this some more, if being honest....

 

skimming through some research articles it indeed looks like that a hydrophobic surface can hinder the wetting ability of liquid metals. who knew!

 

@Falkentyne should be today

 

Um....the pad didn't arrive?
Really need more results.
Mr Fox had the non Soft PGS version (aka the crappy version), and @Sentential keeps claming his pad is on par with LM on *laptops* when no one else can back this claim up.
And I tested this Soft PGS pad on a DESKTOP with good mounting pressure and it performs on par with MX-4 and Phobya Nanogreae Extreme, so there's no way in living hell it performs like LM.

A control test has to be done, meaning a regular LM application on the silicon (high end control test), a "high end thermal grease" application (e.g. Kryonaut, Phobya Nanogrease, Gel Maker Nano), and the Soft PGS Pad bare test, as a direct control test, but it seems no one wants to do the work scientifically and properly. If I weren't disabled, I'd be the guinea pig and do it (I do have These Soft PGS pads, 30 grams of LM (probably 20g left), some Kryonaut, Nanogrease, Gel Maker Nano and NT-H1), but I'm not going to put myself in the hospital and possibly die, for science. It's not worth it.

 

apply LM on both sides of the pad then sandwich it?

 

I see I made a mistake, a big one :/

LM will not adhere to these pads. It just runs all around in drops. Think of like trying to spread LM on a surface with residue Isopropyl that isn't dried up fully.
And I see my typing fail.

I meant, a LM application on the GPU/CPU by itself.

Anyway please post your results when you get them.

 

I equally lack time between work and the little ones. The results I got with the 13r3 that the Dell tech mauled today were absolutely within a 5c* margin of error that I got using conductonaut. Memory serves I was seeing about 68c* and with this 71c* on a bad mount with a damaged fin array so I'd say that's awfully close.

As for the 15r3 it was assembled hastily as my oldest wanted to help and I couldn't explain why ESD is bad since he's 2. Would have kept it if not for the fact it was a gsync unit that I explicitly ask it not be but hey at least everyone here will see results of PGS behavior on all 3 AW units courtesy of Dell.

Which leads me to my final trick I just got approved for a fully loaded 17r4 as my 4th exchange so we will see how this one does assuming it isn't beat to hell like the 13" was. This one I'll assemble at the office after work as it has both the tools and lack of children I need to get a perfect mount. Will post results when I get them and compare them to iunlocks.

Also for those wondering if the PGS deforms to fill surface imperfections it absolutely does. I have a photo of the heatsink pulled from the previous working but now dead 13r3 and there's a perfect indent from the CPU so its definitely not reusable as IC claims.

Lastly @Falkentyne you can disagree with me if you like but all the screenshots I've posted are legit. I'll also take one on a 13r3 with the standard Dell Shinetsu/MicroSI paste for comparision.

 

Photo of PGS compression for those interested. Also no that residue was not there previously. That is pump-out from the pad itself and as a result of the cycling it is actually stuck like tape to the heatsink surface.

 

13r3 with Dell ShinEtsu/MicroSi paste. Same unit, same conditions and drastically worse temps by over 10C* compared to PGS

 

Shin Etsu is rather mediocre paste. There was a massive thread over on NBR years ago about that.
Need to compare a decent reputable paste like Kryonaut, Nanogrease Extreme, Gel Maker Nano, etc...

 

This looks very promising. Re. the pump-out = very surprising. Do you think it is silicone???

 

Binding resin with low thermal stability perhaps?

 

I'm not sure but it is similar in consistency to the standard chewing gum style thermal pads. It's almost like the graphite is woven like steel wool and then soaked in that resin.

So this confirmed a couple of things to me. First there is a burn in period to melt the wax binder and 2 you need to cut a pad substantially larger than the target as it seems to shrink somewhat from it's original size as shown from my previous screenshots and finally this is NOT reusable at all

 

The goal: using 1950W/mk XY to our advantage.

Since first application of IC7 diamond I believe carbon stuff is the future for cooling and seeing 1950 W/mk for XY + 10 W/mk for Z has already some potential, important is the way of using strong characteristics of PGS (XY transfer) to substitute low (Z transfer) and combining them to our advantage. I think I am going to concentrate more on transferring heat off dyes to heat-sink as QUICK and far AWAY as possible.

To me it seems that the dye size is in general too small. Copper is unable to transfer heat away fast enough=bottleneck. See picture 1. We need to eliminate the red spot which makes thermal sensors mental.

I thought sticking one-sided sticky PGS with thinnest possible adhesive and with highest possible W/mk, covering area as on picture 2 is a faster way to transfer heat to larger area, and mainly directly super-fast AWAY from dyes (fully utilising good XY PGS heat-transfer to copper). This SHOULD lower maximum temperatures IF the temperature sensors are within the dyes and definitely take care of temperature (e.g. one core thermal 5 GHz boost Intel 8th generation) spikes caused by slow heat-intake of copper.

Combination of 0.020mm 1950W/mk PGS (0.010mm carbon + 0.010mm adhesive) and 0.2 mm soft PGS ( @Sentential) or a dot of IC7 should work. What is lost on low Z transfer will be replaced by 'PGS virtual huge dye hardware PGS.img' equallin size of PGS sticker on heat-sink LOL

Now I go _\|/_ sleep because I am too deep in this wannabefokineinsteinPGSdream lol

 

I too have pondered using an aerosol spray adhesive to stick non-sticky TIMs to surfaces where it would prove beneficial.

 

https://imgur.com/ejNQhUa

Hmmmmmm.............


Been a while since I posted anything here. But I've been tinkering with Panasonic PGS. The PGS-Sheets, PGS Graphite-PAD and Soft-PGS...and the IC pad. Testing using a Test Bench and Desktop products. ALL TESTS: IHS to WATERBLOCK. No Direct Die.
https://na.industrial.panasonic.com...hermal-management/thermal-management-products

Test conditions for PGS Graphite Sheets, SSM was a 7700K running at: 4.7GHz @ 1.23v Prime95 26.6 8K/8K FFT and a 5.0GHz 1.350GHz Prime95 26.6 8K/8K FFT. 4.7GHz load is approximately 65-67W. At 5.0GHz it's approximately 100W load. I used 25um, 40um, 70um and 100um RAW sheets. No adhesive or anything. Straight Graphite. Links below to the products.

(25um Sheet) https://www.digikey.com/product-detail/en/EYG-S091203/P13690-ND/1630955?utm_medium=email&utm_source=oce&utm_campaign=1069_OCE18RT&utm_content=productdetail_US&utm_cid=498264&so=54722800&mkt_tok=eyJpIjoiT0Rjd05tSTVaVFpsTVdVMiIsInQiOiJtUThkaTVUbUdaSmtmSWpDVlhodGFMZk5xMUI4RVI1aFRhZSs4ajNwQWVCS1J2QUpHZEVXS3ozbkl5VFpha3F2K2Zaa2M0TG90cG5cL1pTcHc2d3FtTitvQjI0YmpYRFdheUFtSXZDdTVjWHVpZTNENDFZMmRJb0xaeHVXZkVUa1gifQ==

(40um Sheet) https://www.digikey.com/product-detail/en/EYG-S091204/P19076-ND/5844903?utm_medium=email&utm_source=oce&utm_campaign=1069_OCE18RT&utm_content=productdetail_US&utm_cid=498264&so=54722800&mkt_tok=eyJpIjoiT0Rjd05tSTVaVFpsTVdVMiIsInQiOiJtUThkaTVUbUdaSmtmSWpDVlhodGFMZk5xMUI4RVI1aFRhZSs4ajNwQWVCS1J2QUpHZEVXS3ozbkl5VFpha3F2K2Zaa2M0TG90cG5cL1pTcHc2d3FtTitvQjI0YmpYRFdheUFtSXZDdTVjWHVpZTNENDFZMmRJb0xaeHVXZkVUa1gifQ==

(70um Sheet) https://www.digikey.com/product-detail/en/EYG-S091207/P13688-ND/1630953?utm_medium=email&utm_source=oce&utm_campaign=1069_OCE18RT&utm_content=productdetail_US&utm_cid=498264&so=54722800&mkt_tok=eyJpIjoiT0Rjd05tSTVaVFpsTVdVMiIsInQiOiJtUThkaTVUbUdaSmtmSWpDVlhodGFMZk5xMUI4RVI1aFRhZSs4ajNwQWVCS1J2QUpHZEVXS3ozbkl5VFpha3F2K2Zaa2M0TG90cG5cL1pTcHc2d3FtTitvQjI0YmpYRFdheUFtSXZDdTVjWHVpZTNENDFZMmRJb0xaeHVXZkVUa1gifQ==

(100um Sheet) https://www.digikey.com/product-detail/en/EYG-S091210/P12726-ND/678309?utm_medium=email&utm_source=oce&utm_campaign=1069_OCE18RT&utm_content=productdetail_US&utm_cid=498264&so=54722800&mkt_tok=eyJpIjoiT0Rjd05tSTVaVFpsTVdVMiIsInQiOiJtUThkaTVUbUdaSmtmSWpDVlhodGFMZk5xMUI4RVI1aFRhZSs4ajNwQWVCS1J2QUpHZEVXS3ozbkl5VFpha3F2K2Zaa2M0TG90cG5cL1pTcHc2d3FtTitvQjI0YmpYRFdheUFtSXZDdTVjWHVpZTNENDFZMmRJb0xaeHVXZkVUa1gifQ==

Effective at up to 65-75W load. Same as MX-4 TIM used at that time. (approx 1 week ago). Past that point...looses ground quickly and becomes saturated/overwhelmed.

One interesting thing I did try was using it as a SHIM. That's where this material is QUITE interesting. It's thermally Transparent up to the 100W load tested. And worked transparently on 25, 50, 70, and 100um parts. I tried a crazy...6 TIM layers, 5 PGS layer sandwich...uh...that didn't work so well as the prep was terrible...just line, pgs, line, pgs, line,....etc. Then use the mounting pressure of an Supremacy EVO waterblock to press together...not good. Failed experiment. Although if it was properly prepared...may work.

Second round of testing...lets figure out what IC is using. Making their own? Or repackaging. It appears they're repackaging Panasonic Soft-PGS. I purchased P/N#: EYG-S0918ZLX2 (90mm x 180mm sheet)

https://www.arrow.com/en/products/eyg-s0918zlx2/panasonic

Test conditions: 6900K at 4.2GHz, 1.300 vCore Manual and using Prime95 26.6 8K/8K FFT for heat. Approx 160w load. Thermal Grizzly Kryonaut vs IC and Panasonic. Result between IC and Panonasonic: core temps are with-in 1C of each other. Effectively margin of error from a shutdown and reboot. It's approximately a 8-10C POSITIVE Temperature increase. aka...going from 55C to 65C vs Kryonaut. Not a good replacement at higher than a 65-75W load.

Have not tested DIE to IHS in a delid situation. Nor do I have the same 7700K on hand anymore. May work at lower loads. Will retest at a later time.

At this time unless proven wrong...the IC pad is nothing more than Panasonic Soft-PGS (compressible type)

https://na.industrial.panasonic.com...hermal-management/thermal-management-products

Link to a bunch of pics. I need to organize this whole bit more.

https://imgur.com/a/Hs4375c

 

Thank you. I am completely vindicated now.
There is absolutely NO way in hell that @Sentential comments about Soft PGS Being ANYWHERE NEAR close to liquid metal are accurate.
No.Way.In.Hell.

And I am not going to live and die as a coward. I'm going to back up those words and not be afraid of people on the internet.
It's funny how I'm proven right all along and yet so many people think I'm the Village Idiot.
Seems like only @heliada, @Kittys and @Vistar Shook ever believe anything I say.

+rep, thank you for taking the time to do these experiments. Glad that some people actually care about putting in true work instead of being lazy candyasses.

 

You're really reading too much into this, many of us have time commitments that prevent the kind of testing you're asking for. Frankly the fact that I'm willing to test the material on 3 AW BGA units ought to be plenty, it's not like there aren't other configurations that my or other results could be compared to.

I can only comment on my experiences and using PGS and Conductonaut on a 13r3 under similar conditions and what I got was close.

I also took apart a 15r3 and posted results as well. Did I say it was on par with LM? No because I've never used one on a 15" before so I posted the raw results and left others to comment.

When I get the 17" I will do the same and we'll see what it looks like compared to iUnlocks. If it has an upper wattage limit we're about to find out aren't we?

There's just no reason to get so worked up like this.

 

Interesting: IC graphite pad = gelid paste

https://www.innovationcooling.com/initial-desktop-tests-ic-graphite-vs-gelid-gc-extreme/

 

Funny how they refer to it being an i7-7700k throughout the article but the pictures show an i3-7350k, unless I'm misreading something?

 

It seems they use more than one setup...

Water Test Methodology

• Prime 95 26.6 1344K with 7600K
• Liquid Metal TIM on DIE of 7600K
• Custom water cooling 22.2c ambient temperature
• Score taken from the lowest temp.

Test Setup

• Liquid Metal on DIE of 7700K
• Cinebench R15 load test, run within a minute of booting
• 7700K @ 5200Mhz using 1.30v
• Maximum temperature average of all 4 cores after bench has finished.

And look on Cpu Package Power. + count the cores.

 

i think ure mixing up.some stuff there. they did use different setups but they didnt compare data across setups, only same setup data. and the screenshot ure showing is from the 7600k so four physical and logical cores is correct, isnt it?

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

I even posted pict in my post + I said... count cores How many physical cores have 7600K ?

 

So thread TL;DR: Liquid metal compounds still outperform any other TIM and the graphite thermal pads are only really good for stock loads, no overclocking/overvolting of any kind, with about MX-4 or Antec Formula 7 grade performance. Not bad since it's reusable but not the best performance either.

 

aaah i thought u meant the CPU0 to CPU3 part, gotcha

 

Hey guys. Before dismissing this completely, I thought I would try something different on my Clevo P870DM-G. I left Conductonaut on my CPU IHS and heat sink and added the IC Graphite pad in the middle instead of using the graphite pad dry and guess what... works great. Works better than Conductonaut all by itself, probably due to a combination of low contact pressure and the dry graphite pad being very porous. Since that worked well, I did the same on my GTX 1080 and it made that better as well.


So, I bumped the CPU up to 48x6 and still decent. Nowhere close to the insane temperatures I saw when testing with the graphite pad only.

@Innovation cooling

 

Well there you go, sometimes you have to think outside the box, Good job!

 

Very interesting. So basically using the pad as a shim. How many degrees improvement did you see compared to just LM without the pad?

 

About 3°-5°C lower temps versus just the Conductonaut.

But, this will be coming out of the oven in about 20 minutes. http://forum.notebookreview.com/thr...gtx-1080-upgrade.814711/page-43#post-10731555

 

Did you keep the same lower temps vs conductonaut even at your highest overclock?


I was considering something like this actually, wondering what would happen, because someone else had mentioned the idea, but I discarded it afterwards, and I haven't been able to do any work because of my back problems. Definitely experimenting soon.

 

I did not test it that much. Just a few quick before/after tests. I had to get the heat sink into the oven and did not want to spend more time with that yet.

 

This matches what IC found when doing the same thing with LM, namely that there was not a decrease by using this as the meat in an LM sandwich.

 

Would it be equal if you used a Goldsmith made thin silver 0.3/0.6mm shim? You haven't the nice thick 4.4mm Bitspower lid for Coffee chips anymore and the pressure on die is less than wanted..

 

Not bad. I feel like shimming with the graphite pad would help my unified HS and allow me to use LM on the IHS and GPU.

3-5C lower temps, vs. not a decrease, seem to contradict each other...

 

Ah! You found what I found Mr. Fox. This stuff works REALLY well as a shim. It's effectively transparent. Did not try it with LM...so thanks! Adds to the knowledge base.

That other test....128K FFT used....only generating 33-34W of heat. Something isn't quite right there. Also, at lower heat loads, the Panasonic PGS (since it's basically that, just repackaged) can keep up with TIM. There becomes a point of saturation where it can not move heat fast enough. Acts more like a thermal sponge.

I think PGS can have a good use case as a Shim used with TIM on low pressure mounts to increase the pressure and improve transfer.

 

So where is the best place to purchase this Panasonic PGS?

 

I have a post a page or 2 back with links. Arrow seems to have the best prices when shipping is factored in. $33.xx shipped. You can cut about 12-15 30mm x 30mm pads from one. Or approx 9-12 40mm x 40mm pads.

https://www.arrow.com/en/products/eyg-s0918zlx2/panasonic

The Soft-PGS line is what the IC pad basically is. The other links are for raw sheets that are really thin.

 

Is 0.2mm thick enough to be used as a shim? How compressible is the material?

 

By decrease I meant decrease in efficiency, not decrease in terms of lower temps. So, in other words, I was referencing that temps would not be higher.

"ICD was also used in one test the other without.

The pad with ICD performed the same as ICD/pad/graphite pad alone - This relates to the contact resistance stickiness I mentioned on the introduction, the Graphite pad w/ICD now equals ICD in thermal performance. might be useful in poor contact conditions but the graphite pad sucks up the oils and have no idea what effect this would have on reliability. So kind of defeats purpose and now subject to liquid paste failure

With LM's it's different, the pad is hydrophobic impossible to spread LM compound on the pad so I applied it to sink and heat source/IHS then applied the graphite pad sandwiched between the two.

Here the pad takes advantage of the LM contact resistance (stickiness) utilizing more of available material resistance and performs the same as the LM compound. The LM being the choke point for heat flow , so pad durability should be stable as the LM just beads up on the pad with zero absorption

Not sure whether any advantage to this other than perhaps filling gaps when mechanical tolerances stray to the extreme.

Pad tends to be more forgiving in poor contact situations. @ .005 thick




In the test above on a heat sink the base was bowed then lapped to a better contact for a 5 C improvement - the idea here is with the pad thickness you may get that 5C or some portion of it without lapping as observed in one of the laptop tests linked in the intro - pads are stackable but all this this needs more testing.

This stuff is flexible for alot of applications so it will be interesting over the next couple of years to see the ways that people think up in terms of application and use." http://forum.notebookreview.com/thr...r-test-and-review.815439/page-3#post-10709808

 

Thanks for the link. Yeah I'm liking the idea of a LM sandwich more and more given the poor contact situation on the typical Clevo unified HS.

 

Hmmmmm... The posts must've been working their way into your subconscious

http://forum.notebookreview.com/thr...-test-and-review.815439/page-17#post-10720996

http://forum.notebookreview.com/thr...-test-and-review.815439/page-16#post-10720668

Glad to see someone give it a shot.

 

As I was putting my P370EM back together after reflowing a failed solder join on one of the heatpipes I tried the LM sandwich with the pad on both the CPU and slave GPU (not master GPU). Each already had shims and LM (and kapton/foam dam) in place.

I ran a simple test just sitting there in the Crysis 3 jungle level, all Ultra 1080p with 100fps cap, aircooling only, GPUs at 115W each max 1950/0.95V

With pad:
CPU thermal throttle @ 99C (maintaining 3.7-3.9ghz), slave GPU thermal throttle @ 92C (maintaining ~1600mhz), master GPU in the 60s
Throttlestop 1B: max 96C @ 4.2ghz (avg low 90s)

Removed pad: CPU 92C max (constant 4.2ghz), slave GPU 82C (maintaining ~1750mhz), master GPU 60s
Throttlestop 1B: max 88C @ 4.2ghz (avg mid 80s)

fps didn't drop much so the thermal throttles weren't that bad. I think due to the shims already there, no gain in gap filling or mounting pressure was there to be had

 

Wow the pad made it a lot worse.

 

@bennyg
Was that the Soft PGS pad, or the regular graphite pad? (seems to be proven that the IC pad is Soft PGS).

 

You said you were using shims. How thick were those shims. If more than 0.2mm, them you left a gap. The other alternative is it was light contact/low pressure. That decreases performance of the PGS, which effects everything else. You should mention those aspects because without them, you are missing key bits of information that allow drawing proper conclusions for this information.

Sent from my SM-G900P using Tapatalk

 

Mind you, this is compared with a setup I've spent a lot of time trialling & erroring as it's basically using 115W GPUs on 75W heatsinks without the water flowing.

Calipers read ~0.2mm thick (but tbh I don't really trust them at such small distances). Seems rather compressible by squeezing them on a small offcut I was able to get the calipers to read 0.0mm

0.5mm shims, it's a fell-apart-and-rebuilt fused CPU and slave GPU heatsink with a copper pipe for carrying liquid and "learning experience" level workmanship, it needed shims to have any contact in the first place... but once that contact issue was sorted the water cooling setup kept both GPUs at ~70C (~35C dT) under 200W stress test load.

The value of my simple comparison is that there is no automatic benefit to using them as a shim.

The one thing I didn't mention is that Core 1 was still the hottest on the CPU by 4-5C just as it is with liquid metal - the implication there is a concave CPU HS that would benefit from lapping - which I would do if I could find my ^!@#$)& pile of fine sandpapers going up to 5000grit




Out of interest heres a macro photo of what it looked like afterwards. After I took the GPU one off, it was interesting that some of the LM did choose to stick to the pad... I put it back in its plastic sheath which is what it's pressed up against (without any pressure applied)

Spoiler: because, large

 

Surprisingly good! lol.. literally took no time, well maybe cutting out a template for the CPU and GPU took me a little bit but after I slapped it on there, mounted it and I was back online. Max temps a little lower and consistent than the TGK I had on there. Overall satisfied.

Installed on a GT73VR:
6820HK @4.0GHz
GTX1070 @2GHz

Installation time: aside from cutting out a template, didn't take more than 15 minutes.

Only thing is waiting for it to go back on stock on Amazon as I'm not a tester and had to buy it and I live overseas so +a week to shipping.

 

Did you coat the GPU resistors with nail polish first? That was EXTREMELY dangerous! That pad would glide across my r9 290X out of place if I even so much as stared at it too long. Like NO traction at all. If that pad shifted and touched those resistors, your GPU would be gone.