The Official Thermal Interface Material thread

[Brotato]

Posted this in the news thread but figure this might be the better thread to do so...
______

Bought that thermalright Helios pad for my 7800x3d. As it’s on an AIO I figure it would be there for the long haul. And if the AIO or motherboard eats it?…I have bigger problems than a 7 dollar thermal pad to replace.

didn’t do the chill or freeze method but just went ahead and tried the install. You only yolo once as kids would say.

effortless install. So if you suck like I do and want to ptm or close enough to it, this would be the one to get.

Fwiw the texture seemed less clay like than the supposed ptm7950 I had purchased from Amazon. That Amazon ptm was mx6 levels of good so we will see how this turns out.

I picked up some a while ago from amazon...some random ass one and it was good. performed well at any rate. A good 3C better than MX-6.

As much as I find the dude annoying, that Linus tech tip guy is also selling PTM

PTM7950 Phase Change Thermal Pad

At long last, the legendary Honeywell PTM7950 Phase Change Thermal Pads from this video are available on lttstore! If you're looking for a thermal interface material that offers higher performance than standard paste, but you don't want to go through the trouble of something like liquid metal...

www.lttstore.com

Probably a bit more reliable than aliexpress or ebuy...would probably be significant nerd rage if it turns out he's selling fake shit too.
He's got a wide audience and probably how a lot of people were introduced to PTM from one of his videos.

I can't seem to edit things. Mixing up pastes. It was the ceramique vs mx6 that was slightly over 3 or so from ceramique..still trying to wrap my head around that one.
Amazon's generic 3rd party seller's PTM was on par with MX-6

 

[jayjr1105]

Here's a trick to easy application on a direct die. Cut a piece a little bit bigger than the die of the chip, remove one side of the film and lay on the chip, use something like a pen or plastic spudger to score the edges of the die so it cuts into the PTM sheet, tap on the middle of the sheet to get it to bond better to the flat part of the die then just pull up and it should come up without issue.

 

[b0uncyfr0]

Where can i buy legit PTM 7950 in the EU? Honeywell is the best brand right - i dont want any of the fakes.

Id be applying it to a 7800xt - i dont know which size i get.

 

[jayjr1105]

Where can i buy legit PTM 7950 in the EU? Honeywell is the best brand right - i dont want any of the fakes.

Id be applying it to a 7800xt - i dont know which size i get.

Aliexpress Upsiren PCM-1. Not a fake, it's been proven to perform identically to 7950

 

[Taraquin]

Where can i buy legit PTM 7950 in the EU? Honeywell is the best brand right - i dont want any of the fakes.

Id be applying it to a 7800xt - i dont know which size i get.

Thermalright heilos is also the same.

 

[FreedomEclipse]

Posted this in the news thread but figure this might be the better thread to do so...
______

Bought that thermalright Helios pad for my 7800x3d. As it’s on an AIO I figure it would be there for the long haul. And if the AIO or motherboard eats it?…I have bigger problems than a 7 dollar thermal pad to replace.

didn’t do the chill or freeze method but just went ahead and tried the install. You only yolo once as kids would say.

effortless install. So if you suck like I do and want to ptm or close enough to it, this would be the one to get.

Fwiw the texture seemed less clay like than the supposed ptm7950 I had purchased from Amazon. That Amazon ptm was mx6 levels of good so we will see how this turns out.


I can't seem to edit things. Mixing up pastes. It was the ceramique vs mx6 that was slightly over 3 or so from ceramique..still trying to wrap my head around that one.
Amazon's generic 3rd party seller's PTM was on par with MX-6

If youre applying directly to the bottom of a CPU cooler or to an IHS. You shouldnt need to chill it. However with direct die application, its a little finnicky and fragile as you are working with a much more smaller surface area so chilling it helps.

 

[b0uncyfr0]

Thermalright heilos is also the same.

Thanks! The Heilos is readily available on EU amazons. Ordered two 40x40.

Now lets hope i can get that hotspot down 10 degrees!

One last question though - when cutting it spec on the 7800xt core - should i be carefult not to overdue it? If the card gets hot enough and any excess PTM makes contact with the PCB board - is that an issue?

 

[Noci]

Thanks! The Heilos is readily available on EU amazons. Ordered two 40x40.

Now lets hope i can get that hotspot down 10 degrees!

One last question though - when cutting it spec on the 7800xt core - should i be carefult not to overdue it? If the card gets hot enough and any excess PTM makes contact with the PCB board - is that an issue?

PTM should be non electrically conductive, but better check the product data sheet. That should be available on the Thermalright website.

And it never harms to prevent from overlap of the die, just make sure that the die itself is fully covered.

Edit:
just checked the webpage and it is not electrically conductive

 

[Borc]

Igorslab tested DOWSIL TC-5550: https://www.igorslab.de/en/dowsil-tc-5550-gelid-gc-4-and-snowman-mtgs-in-the-thermal-paste-test/

I have this paste and yes it's great, my best paste.

 

[ShrimpBrime]

Igorslab tested DOWSIL TC-5550: https://www.igorslab.de/en/dowsil-tc-5550-gelid-gc-4-and-snowman-mtgs-in-the-thermal-paste-test/

I have this paste and yes it's great, my best paste.

Zink and aluminum oxides. See this a lot in all the average run of the mill pastes. Not very abrasive, good enough to fills paces, but be ok conductive. Seen a lot of mixed thoughts about GC4.

Not a lot of talk about diamond pastes though. I've often wondered why people wouldn't be more into them aside from scuffing the Lazer etching and the copper a bit. Scratching some would say. I've always assumed this would create surface area.. like the inards of a waterblock.

I like Antec Formula 7. I've found it outstanding. Have used it many times on naked dies too.

The issue is that there are like no reviews of the paste. At least nothing recent I've seen. But through the years, I've stuck with it.

Are diamonds worthy of thermal transfer in a cpu thermal paste application??

 

[68Olds]

Zink and aluminum oxides. See this a lot in all the average run of the mill pastes. Not very abrasive, good enough to fills paces, but be ok conductive. Seen a lot of mixed thoughts about GC4.

Not a lot of talk about diamond pastes though. I've often wondered why people wouldn't be more into them aside from scuffing the Lazer etching and the copper a bit. Scratching some would say. I've always assumed this would create surface area.. like the inards of a waterblock.

I like Antec Formula 7. I've found it outstanding. Have used it many times on naked dies too.

The issue is that there are like no reviews of the paste. At least nothing recent I've seen. But through the years, I've stuck with it.

Are diamonds worthy of thermal transfer in a cpu thermal paste application??

I got a tube of the diamond dust paste years ago in the free test sample offering days. I had (& still have) a water block that sets down on the CPU with spring pressure. The diamond paste was too thick to compress & spread well using just the springs on my block mount. Performance was really bad thanks to the lack of pressure with my CPU mount. Probably works great with a harder mount, though.

 

[lexluthermiester]

Not a lot of talk about diamond pastes though.

This is because diamond has qualities that make it difficult to use as a TIM without damaging the surfaces they're applied to.

I like Antec Formula 7. I've found it outstanding. Have used it many times on naked dies too.

Interesting. Why have we never heard of this?

 

[Noci]

Are diamonds worthy of thermal transfer in a cpu thermal paste application??

If it is as a thermal conducting solid component in a thermal paste, it must be ultra fine diamond dust.
I have no idea if this has impact on the thermal conductivity of 2,200 W/(m·K) as in pure crystals. This must be quite complex or impossible to measure, so debatable. Here some info: here it seems it has more functionallity in optronics and stuff.

Diamond is basically just Carbon and there are several different types of diamonds (crystalline structures). So talking about Carbon, more usable for us and proven performers are the Carbon pads and the recent Graphene pads.
Some TIM manufacterers (Innovation Cooling, Thermaltake, Evercool for example) already claimed they had diamond in their pastes, but as far as I know there is no magic TIM out there containing Diamond and smoking the competition.

No matter how good the TIM conducts the heat, it is just a tiny part (barrier) of a cooling solution, the limitations most likely are a sum of the TIM and many other variables in your cooling solution.
Just go with one of the best (check reviews) and easy to apply ones and look at other possibillities to lower the temps a couple of degrees, as this is the maximum achievable with TIM.

 

[ShrimpBrime]

If it is as a thermal conducting solid component in a thermal paste, it must be ultra fine diamond dust.
I have no idea if this has impact on the thermal conductivity of 2,200 W/(m·K) as in pure crystals. This must be quite complex or impossible to measure, so debatable. Here some info: here it seems it has more functionallity in optronics and stuff.

Diamond is basically just Carbon and there are several different types of diamonds (crystalline structures). So talking about Carbon, more usable for us and proven performers are the Carbon pads and the recent Graphene pads.
Some TIM manufacterers (Innovation Cooling, Thermaltake, Evercool for example) already claimed they had diamond in their pastes, but as far as I know there is no magic TIM out there containing Diamond and smoking the competition.

No matter how good the TIM conducts the heat, it is just a tiny part (barrier) of a cooling solution, the limitations most likely are a sum of the TIM and many other variables in your cooling solution.
Just go with one of the best (check reviews) and easy to apply ones and look at other possibillities to lower the temps a couple of degrees, as this is the maximum achievable with TIM.

A lot of huge good points there!!

Totally a barrier. This is the start of the Need of a good thermal transfer between 2 surfaces.

Copper is good. Better than Aluminum.

I've used copper Antisieze before. Actually worked pretty well. Beyond my expectations really.

It's not a claim. Those IC and EverCool prove it by scuffing of the ihs plate and cooler.

Not really trying to measure diamond "dust or nano particles" but rather the paste it's in. As far as I know, none of the diamond nano pastes claim some magical thermal transfer number or anything.

But I've found them to be pretty good and noticeable differences from most run of the mill pastes.

Or why not copper instead of aluminum oxides?

 

[KLiKzg]

Igorslab tested DOWSIL TC-5550: https://www.igorslab.de/en/dowsil-tc-5550-gelid-gc-4-and-snowman-mtgs-in-the-thermal-paste-test/

I have this paste and yes it's great, my best paste.

This is what I was telling about: real testing equipment, which possibility to exert the same amount of force same time & with testing sensors.

Looking forward to more tests: https://www.igorslab.de/en/category/cooling/watercooling/

 

[Lew Zealand]

A note about Arctic MX-5. Yeah we all know it separated on a lot of people and was replaced pretty quickly in the market. But that happened after I bought it and mine never separated so I've been using it.

Well, until now.

Kid's Ryzen 5 1600AF was running quite hot but this case is in a dusty area and there are constantly loads of things running in the background and it's got a Stealth cooler and I hadn't cleaned it out in a year, blahblah. OK I open it to replace the CPU as it's my higher voltage 1600AF (other one is notably better) but none of these real things was the problem. Instead:

The MX-5 completely dried out in less than 3 years.

WTF I have never had that happen. OK I don't have any other paste RN so I repaste the same CPU again with MX-5, adding a small Cryorig M9A tower cooler and it's idling (still with all the crap running in the background) at 38C instead of 63C in a warm room. Yes, some of it is the Cryorig but the majority is that the paste is still pasty.

Haven't ordered anything yet, MX-6 OK or go for MX-4 for CPUs?

Related: I recently repasted a cheap Chinese RTX 2060 with the MX-5 (before knowing about the dryout) fully knowing it wasn't ideal because of direct-die pump-out but more as a test to see if the stock cooler would work better with different fans. I'd love to try PTM 7950 on it but this cooler's contact patch is the least smooth thing I've ever encountered in anything resembling a heatsink. By a long shot, it's like sandpaper. Will PTM 7950 work under such crap conditions?

 

[Shrek]

The MX-5 completely dried out in less than 3 years.

Silicone oil should be good to 500°F, so I wonder what they use.

 

[freeagent]

I got a tube of the diamond dust paste years ago in the free test sample offering days. I had (& still have) a water block that sets down on the CPU with spring pressure. The diamond paste was too thick to compress & spread well using just the springs on my block mount. Performance was really bad thanks to the lack of pressure with my CPU mount. Probably works great with a harder mount, though.

My thoughts exact, I had the same tube, and I had a crazy high mount pressure.

 

[Taraquin]

A note about Arctic MX-5. Yeah we all know it separated on a lot of people and was replaced pretty quickly in the market. But that happened after I bought it and mine never separated so I've been using it.

Well, until now.

Kid's Ryzen 5 1600AF was running quite hot but this case is in a dusty area and there are constantly loads of things running in the background and it's got a Stealth cooler and I hadn't cleaned it out in a year, blahblah. OK I open it to replace the CPU as it's my higher voltage 1600AF (other one is notably better) but none of these real things was the problem. Instead:

The MX-5 completely dried out in less than 3 years.

WTF I have never had that happen. OK I don't have any other paste RN so I repaste the same CPU again with MX-5, adding a small Cryorig M9A tower cooler and it's idling (still with all the crap running in the background) at 38C instead of 63C in a warm room. Yes, some of it is the Cryorig but the majority is that the paste is still pasty.

Haven't ordered anything yet, MX-6 OK or go for MX-4 for CPUs?

Related: I recently repasted a cheap Chinese RTX 2060 with the MX-5 (before knowing about the dryout) fully knowing it wasn't ideal because of direct-die pump-out but more as a test to see if the stock cooler would work better with different fans. I'd love to try PTM 7950 on it but this cooler's contact patch is the least smooth thing I've ever encountered in anything resembling a heatsink. By a long shot, it's like sandpaper. Will PTM 7950 work under such crap conditions?

It melts after use so yes, bit you may have to tighten the heatsink a bit after a few hours use.

 

[lexluthermiester]

The MX-5 completely dried out in less than 3 years.

Why does that matter? Drying out doesn't matter if the TIM continues to be effective. MX-5 stays effective.

WTF I have never had that happen.

Noctua NT-H1 dries out in 6 months and after that stops working as well.

 

[KLiKzg]

Silicone oil should be good to 500°F, so I wonder what they use.
View attachment 358818

Doesn't need to be fried, as it can also get oxidized on elevated temperatures.

 

[lexluthermiester]

Doesn't need to be fried, as it can also get oxidized on elevated temperatures.

Few oils oxidize at the temps computer CPUs them expose to.

 

[Lew Zealand]

Why does that matter? Drying out doesn't matter if the TIM continues to be effective. MX-5 stays effective.

Did you miss this part:

so I repaste the same CPU again with MX-5, adding a small Cryorig M9A tower cooler and it's idling (still with all the crap running in the background) at 38C instead of 63C in a warm room.

Same paste, same CPU, slightly better cooler which could be a few degrees better at low load. Not 25C better. Paste connection went bad, I assume as it was dried out unless there's a better reason for this. I've replaced a lot of 10 year old dried out paste and some if it was still functioning well where I only got a 5C improvement after re-paste and others where I got a 15-20C improvement. This is the record improvement for me and after only 3 years, which does not speak well of MX-5.

 

[Shrek]

38C instead of 63C

You went from 38C to 63C idling?

 

[Lew Zealand]

You went from 38C to 63C idling?

It was idling (low load, not completely idle) at 63C before repaste. It idles at the same conditions at 38C now.