i9-9900k IHS

[larryccf]

does the 9900k share the same IHS with any other model cpu?

i was going to order one of those copper ones from Aliexpress but they don't show one listed for the 9900k

i don't want it to mount on my cpu, but to model or blueprint it to machine one from a different material (i've got an idea for a material that has extremely high rate of thermal conductivity)

if someone by chance has an IHS from a Cpu that they fubar'd, i'd be willing to pay for it

 

[EarthDog]

These are soldered ihs on 9900k. While you can get it off, I dont think people saw bur a few C improvement...and removing a soldered top is more risk.

That said, not sure if they are exactly the same.. I think there are light differences.

 

[larryccf]

yep - was aware of that. Der8auer has a video on youtube, he's showing 13-14C decrease in temps, but i'm not interested in pulling mine off for right now - just need one to blueprint or copy in a different material. For the sake of the test i want to do, i will attempt to delid mine, if i can't find someone that offers the service professionally, like silicone lottery did.

I retired from mfgring almost a year ago, and now i'm bored. I've always chased my interests / hobbies for new product ideas, and this idea i've got, if it flies, will be a significant improvement in transferring heat out of the die. But that "retired and bored" part, can be dangerous

 

[EarthDog]

Try it with a cheaper cpu.

What material are you planning to make this out of?

 

[larryccf]

EarthDog - i hope you will understand, but i'd rather keep the material close to my chest for the time being. I'll say this re it's ability to transfer or move heat, using a propane torch, flame temp @1500F, we were finally able to get a spot the size of a quarter (25 cent pc) glowing cherry red same as a sheet of metal would. I say finally, because if we moved the torch a hair, the glow would disappear. Got it glowing, pulled the torch away, instantly the glow disappeared and within a few seconds (2-3) i was able to touch the very spot where it was glowing with my bare finger - it was already barely above room temp. As high a co-efficient of thermal conductivity that brass or copper has, you couldn't do that without burning your finger.

 

[EarthDog]

When you say finally, that alludes to it taking a while to get hot, no? That's an issue. Though its curious something that takes long to heat up is fast to cool down...

Dish....what material?

 

[larryccf]

"When you say finally, that alludes to it taking a while to get hot, no? That's an issue." No, exactly the opposite - it was shedding the heat almost as fast as the torch was delivering it.

maybe i didn't explain clearly - a torch flame has a "sweet" spot where it's hitting it's max flame temp - if the flame is 1.5" long, usually just before the tip of the flame to the tip is the hottest part of the flame. This material sheds the heat so fast, if i didn't maintain steady contact between the sweet spot and the material it would cool fast enough to loose the glow. I ended up putting the torch in a small vise on top of a stool and let the torch do it's thing. Once i ot it glowing .... well as i said above, i pulled torch away, the glowing spot disappeared instantly and i was able to physically touch it with my bare finger within a few seconds of pulling the torch away

Heat any metal to a point of glowing, and watch how slowly the glow dies - usually a couple of seconds or so

 

[bignazpwns]

"When you say finally, that alludes to it taking a while to get hot, no? That's an issue." No, exactly the opposite - it was shedding the heat almost as fast as the torch was delivering it.

maybe i didn't explain clearly - a torch flame has a "sweet" spot where it's hitting it's max flame temp - if the flame is 1.5" long, usually just before the tip of the flame to the tip is the hottest part of the flame. This material sheds the heat so fast, if i didn't maintain steady contact between the sweet spot and the material it would cool fast enough to loose the glow. I ended up putting the torch in a small vise on top of a stool and let the torch do it's thing. Once i ot it glowing .... well as i said above, i pulled torch away, the glowing spot disappeared instantly and i was able to physically touch it with my bare finger within a few seconds of pulling the torch away

Heat any metal to a point of glowing, and watch how slowly the glow dies - usually a couple of seconds or so

At this point why not just go without a ihs. I don't it on a few cpus. Kinda sucks for lga but at that point it's better to just lose it and go die to waterblock. Really easy to screw your cousin up but if you can safely delid a soldered ihs then running lidless is nothing to complex.

 

[larryccf]

"At this point why not just go without a ihs. I don't it on a few cpus. Kinda sucks for lga but at that point it's better to just lose it and go die to waterblock. Really easy to screw your cousin up but if you can safely delid a soldered ihs then running lidless is nothing to complex. "

You've got a point, but i think a lot of users would prefer to not run their CPU nekid, plus going that route entails purchasing one of those der8auer kits - if this material proves for use as the IHS, then the next step would be for the waterblock or heatsink on an air cooler. In fact what i've got in mind is eventually the entire aircooler made of this material.

A couple of years ago i fabricated a ribbed plate of this stuff, basically what you see sold as heat sinks for use on NVMe drives or on the VRMs. His oil pan had a fairly decent sized flat bottom so it was ideal for a heatsink, the ribs were 1/8" thick X 1/2" tall, and spaced about 3/8" apart. He reported back his oil temps dropped by 14-15F on the same track on a similiar low 90F day. Not exactly the most documented scientific method but he was happy and i was returning a bunch of favors he had done us. THe stuff just seems to suck the heat out of whatever it comes in contact with and transfer it to the surrounding air.

 

[bignazpwns]

"At this point why not just go without a ihs. I don't it on a few cpus. Kinda sucks for lga but at that point it's better to just lose it and go die to waterblock. Really easy to screw your cousin up but if you can safely delid a soldered ihs then running lidless is nothing to complex. "

You've got a point, but i think a lot of users would prefer to not run their CPU nekid, plus going that route entails purchasing one of those der8auer kits - if this material proves for use as the IHS, then the next step would be for the waterblock or heatsink on an air cooler. In fact what i've got in mind is eventually the entire aircooler made of this material.

A couple of years ago i fabricated a ribbed plate of this stuff, basically what you see sold as heat sinks for use on NVMe drives or on the VRMs. His oil pan had a fairly decent sized flat bottom so it was ideal for a heatsink, the ribs were 1/8" thick X 1/2" tall, and spaced about 3/8" apart. He reported back his oil temps dropped by 14-15F on the same track on a similiar low 90F day. Not exactly the most documented scientific method but he was happy and i was returning a bunch of favors he had done us. THe stuff just seems to suck the heat out of whatever it comes in contact with and transfer it to the surrounding air.

Yeah but anyone who's willing to delid a soldered ihs won't have an issue running the bare die if you are worried about a 2-5c drop. From what I seen anything higher then that is because the ihs is a soup bowl and you would get a good drop just lapping the ihs.

Personally I would look into carbon nanotubes for it.
If I remember correctly it's about 15 times more thermaly conductive then copper and thermal expansion properties are great. I don't deal with materials but I have talked to our meterial engineer at work about using it as an interface for a set up.

 

[larryccf]

the carbon nanotubes wouldn't be a bad choice for material either but i haven't worked with it, i have with the material i'm planning to use. But like i said above, this stuff seems to literally suck the heat out of whatever material it's in contact with - i've got other examples i've experienced, but the oil pan one serves the point. I want to see what it does in contact with the die, and 2nd gen prototype will have "wings" on 2 sides or extensions out that are basically small heatsinks of their own, wings that are maybe 1/2 in width, and run the full length of the side of the IHS. Then go with a waterblock fabricated from it - to go the waterblock first, i'd still be left with the nickel plated copper IHS serving to "bottleneck" the heat to some extent. I'd like to be able to hookup with silicone lottery, or an outfilt like them, to do the delidding so that svc could be offered to a customer. Where i want to take this like i described above is a complete package from die to cooler, whether waterblock or air cooler

"...because the ihs is a soup bowl..." Thank you, i thought i was the only one to notice a lot of IHSs are concave - i mentioned that in another forum and damn if i didn't get looked at like i was from Mars

 

[EarthDog]

Dude,,, what material already??? Seriously... I am a betting man and I would say its been tried and for one reason or another isn't being used. Out with it!

The 'problem' with a heatsink in that area is a lack of airflow. To be worthwhile, airflow must be pointed at those wings.

Yeah, it was pretty common knowledge, since the Core days, that some IHS were concave.

 

[larryccf]

"Dude,,, what material already??? Seriously... I am a betting man and I would say its been tried and for one reason or another isn't being used. Out with it!

The 'problem' with a heatsink in that area is a lack of airflow. To be worthwhile, airflow must be pointed at those wings."

EarthDog - i'd like to identify the material but experience has taught me to hold it close to my chest, for the time being. Reason i suspect it hasn't been tried before is it's way overkill and expensive - but i have good access to the material and if it proves, then i'll invest in an injection mold which will bring the cost way down.

And it wouldn't take that much airflow to show benefit - right now my 9900k is aircooled (Noctua D15S) and i've got the fans as low as they'll go so they're moving some air down on the motherboard. Obviously the "wings" wouldn't show full benefit that they would with fans blowing directly on them, but any additional decrease in temp is a good thing.

 

[bignazpwns]

"Dude,,, what material already??? Seriously... I am a betting man and I would say its been tried and for one reason or another isn't being used. Out with it!

The 'problem' with a heatsink in that area is a lack of airflow. To be worthwhile, airflow must be pointed at those wings."

EarthDog - i'd like to identify the material but experience has taught me to hold it close to my chest, for the time being. Reason i suspect it hasn't been tried before is it's way overkill and expensive - but i have good access to the material and if it proves, then i'll invest in an injection mold which will bring the cost way down.

And it wouldn't take that much airflow to show benefit - right now my 9900k is aircooled (Noctua D15S) and i've got the fans as low as they'll go so they're moving some air down on the motherboard. Obviously the "wings" wouldn't show full benefit that they would with fans blowing directly on them, but any additional decrease in temp is a good thing.

We use it in aerospace. Problem is the cost of it then add in the machining. Much easier to stamp a copper slug then make it out of carbon nanotubes. It always boils down to cost.

 

[larryccf]

We use it in aerospace. Problem is the cost of it then add in the machining. Much easier to stamp a copper slug then make it out of carbon nanotubes. It always boils down to cost.

can the carbon nanotubes be injection molded? - where i'm going with this, and one reason for fabbing the IHS first to prove it, is to make the waterblock & IHS one unit, eliminating the TIM and that interface which has to be a loss of heat transfer efficiency

the idea came to me when i was looking at my XSPC waterblock in pure copper with no nickel plating, and then learning intel's IHSs were nickel plated copper - first thought was "why not just solder the waterblock to the IHS (after milling the nickel plating off) - but then it hit me, why not a one pc unit

this material is fairly light, much lighter than copper. The XSPC waterblock is actually a 2 pc assembly, with the bottom pc simply a plate, that by my eye looks to be about .090" thick - with that plate fabricated as part of the IHS, the intel locating / locking / clamping unit could be retained, and the user simply installs the cpu with that "combo bottom waterblock plate/ IHS" installed as the replacement IHS, then that plate would be threaded for screws to attach it to the IHS / bottom plate part (backwards from the XSPC which has the screws coming up thru the bottom plate into the body of the waterblock.

 

[EarthDog]

"Dude,,, what material already??? Seriously... I am a betting man and I would say its been tried and for one reason or another isn't being used. Out with it!

The 'problem' with a heatsink in that area is a lack of airflow. To be worthwhile, airflow must be pointed at those wings."

EarthDog - i'd like to identify the material but experience has taught me to hold it close to my chest, for the time being. Reason i suspect it hasn't been tried before is it's way overkill and expensive - but i have good access to the material and if it proves, then i'll invest in an injection mold which will bring the cost way down.

And it wouldn't take that much airflow to show benefit - right now my 9900k is aircooled (Noctua D15S) and i've got the fans as low as they'll go so they're moving some air down on the motherboard. Obviously the "wings" wouldn't show full benefit that they would with fans blowing directly on them, but any additional decrease in temp is a good thing.

We need to teach you how to quote properly...

A fan on low through a heatsink is focused air over the fins (and a lot more of it than non direct case airflow)..of course it works!

The problem with ALL watercooling is it takes airflow away from around the socket in most cases. So without direct airflow, unless those wings have some square inches on them, they wont do much. I recall a block with heatsink find on it years ago...if it was all that (showed worthwhile improvements) one would imagine it would have caught on...but we dont see it.

Regardless, I want to see what you do, but the veil of secrecy over the material is really for no reason.

Edit: how do you plan in doing this without a TIM between the block and die???????

 

[larryccf]

We need to teach you how to quote properly...

A fan on low through a heatsink is focused air over the fins (and a lot more of it than non direct case airflow)..of course it works!

The problem with ALL watercooling is it takes airflow away from around the socket in most cases. So without direct airflow, unless those wings have some square inches on them, they wont do much. I recall a block with heatsink find on it years ago...if it was all that (showed worthwhile improvements) one would imagine it would have caught on...but we dont see it.

Regardless, I want to see what you do, but the veil of secrecy over the material is really for no reason.

Edit: how do you plan in doing this without a TIM between the block and die???????

Sorry, i left something out in my earlier explanation - the wings would be for the air cooled unit. and the air unit would not be a one piece unit with the IHS - and by fans down low, i meant the cooler's fans mounted low on the cooler so a portion of the fan is protruding below the fins. But it would still require TIM between the heatsink and IHS. The ribbed / heatsink wings would be on the two edges of the IHS that are parallel to the case's roof & floor.

It's the waterblock, being one pc with the IHS, that won't require TIM - as it's one pc with the IHS, there'll be no place to apply it. If you meant under the IHS, TIM would still be required there.

 

[prime81]

Silicon lottery is the key to the 9900K, the soldered IHS is good to go. Power delivery is key as well as getting the heat dissipated from that power. I'd spend my money elsewhere when it comes to delidding the 9900k.