PSA: The Rockit Cool Copper IHS for the i9-12900KS will make your CPU incompatible with most custom socket upgrades

[cleric_warlock]

I ordered Thermal Grizzly's 12th gen contact frame knowing full well that it might not work with my copper IHS upgrade, but wanting to take the risk to see if it could work anyway. Of all the aftermarket socket upgrades that I've seen, thermal grizzly's has the most forgiving clearance around the bulk of the stock IHS so I figured that it would have the best chance of fitting my modded IHS, but unfortunately even this socket upgrade doesn't fit. This also probably rules out other custom contact frames like Thermalright's for instance that more closely surround the main bulk of the IHS. 

 

Personally though, I'm probably not going to return the contact frame I bought. My ulterior motive in buying this frame was to have a good design reference even if it didn't fit. As a mechanical engineer by profession who's comfortable with CAD, I'm going to try to design a new frame that will fit my IHS for fun with this as a reference. I'm definitely not DrBauer and somebody else might end up beating me to the punch if they haven't already, but if I end up developing a model for a frame that actually works with the copper IHS i'll probably put it up on grabcad or something. This is definitely a pet project for me so not going to make any guarantees about when I'll get it done.

[person123456789]

On 8/9/2022 at 7:46 PM, cleric_warlock said:

I ordered Thermal Grizzly's 12th gen contact frame knowing full well that it might not work with my copper IHS upgrade, but wanting to take the risk to see if it could work anyway. Of all the aftermarket socket upgrades that I've seen, thermal grizzly's has the most forgiving clearance around the bulk of the stock IHS so I figured that it would have the best chance of fitting my modded IHS, but unfortunately even this socket upgrade doesn't fit. This also probably rules out other custom contact frames like Thermalright's for instance that more closely surround the main bulk of the IHS. 

 

Personally though, I'm probably not going to return the contact frame I bought. My ulterior motive in buying this frame was to have a good design reference even if it didn't fit. As a mechanical engineer by profession who's comfortable with CAD, I'm going to try to design a new frame that will fit my IHS for fun with this as a reference. I'm definitely not DrBauer and somebody else might end up beating me to the punch if they haven't already, but if I end up developing a model for a frame that actually works with the copper IHS i'll probably put it up on grabcad or something. This is definitely a pet project for me so not going to make any guarantees about when I'll get it done.

the copper ihs will perform better than any contact frame/ihs lapping

[LIGISTX]

On 8/9/2022 at 4:46 PM, cleric_warlock said:

I ordered Thermal Grizzly's 12th gen contact frame knowing full well that it might not work with my copper IHS upgrade, but wanting to take the risk to see if it could work anyway. Of all the aftermarket socket upgrades that I've seen, thermal grizzly's has the most forgiving clearance around the bulk of the stock IHS so I figured that it would have the best chance of fitting my modded IHS, but unfortunately even this socket upgrade doesn't fit. This also probably rules out other custom contact frames like Thermalright's for instance that more closely surround the main bulk of the IHS. 

 

Personally though, I'm probably not going to return the contact frame I bought. My ulterior motive in buying this frame was to have a good design reference even if it didn't fit. As a mechanical engineer by profession who's comfortable with CAD, I'm going to try to design a new frame that will fit my IHS for fun with this as a reference. I'm definitely not DrBauer and somebody else might end up beating me to the punch if they haven't already, but if I end up developing a model for a frame that actually works with the copper IHS i'll probably put it up on grabcad or something. This is definitely a pet project for me so not going to make any guarantees about when I'll get it done.

Should be super easy to cad that up and get it machined. Love it!

[cleric_warlock]

1 hour ago, LIGISTX said:

Should be super easy to cad that up and get it machined. Love it!

The geometry is definitely easy and I have a good pair of calipers to measure everything out, but getting the tolerances tight enough and getting the material selected and conditioned the right way so there is no conductivity or galvanic corrosion will probably end up being more tricky and expensive since costs increase exponentially as tolerances are tightened. This would be a lot easier if I had my own CNC machine, but sadly I'm not that loaded so i'll probably outsource to a good fab company that can do it for a reasonable price. 

[LIGISTX]

4 minutes ago, cleric_warlock said:

The geometry is definitely easy and I have a good pair of calipers to measure everything out, but getting the tolerances tight enough and getting the material selected and conditioned the right way so there is no conductivity or galvanic corrosion will probably end up being more tricky and expensive since costs increase exponentially as tolerances are tightened. This would be a lot easier if I had my own CNC machine, but sadly I'm not that loaded so i'll probably outsource to a good fab company that can do it for a reasonable price. 

Pretty sure there are mail order CNC shops just like 3d print shops. Tolerances shouldn’t be that hard to hold, although the galvanic corrosion part is a fair point. 

[RockitCool]

Just to let you know. We now make all copper IHS to fit the "contact frames".

Contact me if you would like one to test.

 

 

[cleric_warlock]

5 hours ago, RockitCool said:

Just to let you know. We now make all copper IHS to fit the "contact frames".

Contact me if you would like one to test.

 

 

Are you selling the frame compatible copper IHSs now? Because I don't see any that call out that compatibility your website - can you provide a link here please if you're ready to sell those?

 

Speaking for myself, the re-lidding you guys did on my CPU reduced my temps by 10C and I don't think re-lidding again just to fit a new socket on is worth the risks or expense involved, so in my particular situation designing and fabricating a new contact frame that fits my existing IHS is much more sensible.

 

In order to fit into existing frames it would be necessary to reduce the thermal contact surface area of the old copper IHS which may have an effect on thermals and would render the expanded surface area selling point of the original copper IHS potentially moot. I'd have to see data that showed that there is a significant thermal improvement on the updated copper IHS with the contact frame over the older copper IHS with the stock ILM that I currently have before I would even entertain the idea of another re-lidding.

[cleric_warlock]

It's been a while so I thought I'd update this post with some of my recent findings. When doing the design studies to conceive of the prototype for this redesign, I realized that the original copper IHS with the expanded surface area achieves this extra surface area by raising the lip that goes all the way around the stock IHS up to the level of the hot plate that interacts with the cooler, leaving only the tabs that the stock Intel ILM clasps onto. This presents a problem in that it is no longer possible to even out the mounting pressure without designing a brace that partially obstructs contact between the cold plate and IHS in order to apply even mounting pressure. Any brace that evens out the mounting pressure with the earlier model rockit IHS would have to press down on the top of the IHS since there is no longer any peripheral lip to push down on. Putting part of the mounting brace between the cold plate and IHS would frankly be stupid, so there isn't really a solution to evening out ILM mounting pressure with the larger rockitcool copper IHS in place.

 

Then we get to the merits (or lack of) for having the extra IHS peripheral surface area that this version of the rockitcool copper IHS provides. The 15% or so of extra IHS area is debatably consequential in heat transfer since it is entirely peripheral and temps don't change noticeably if thermal paste is there or not. The fact is that thermal returns diminish more and more quickly the further you get away from the IHS-die interface so I highly doubt that there is any real benefit from that extra surface area. Considering that this extra area comes with the inability to have even mounting pressure, it's best to relid with the stock intel IHS so that you can make use of existing aftermarket braces like DrBauer's. One minor benefit of the rockitcool IHS design is the flatness of the IHS, but even this does not really justify its use since it is easier and cheaper to sand or lap your stock IHS and your cold plate if necessary. In my case, both my cold plate and IHS are flat enough that lapping or sanding is probably not worth the trouble.

 

The currently available version of the rockitcool IHS reintroduced the surrounding lip and is now compatible with all of the aftermarket contact frames for Intel 12th gen which also means that the IHS hot plate surface area is now identical to the Intel stock IHS by necessity. Now the new copper IHS is basically just a very marginally flatter version of Intel's stock IHS with all of the nickel plating stripped off, so if you're thinking of using it with a contact frame for intel 12th or 13th gen, you'd be better served by simply sanding or lapping and relidding with your stock IHS. You are much more likely to benefit thermally from sticking with Intel's stock IHS and using one of the aftermarket ILM replacements instead of using the stock ILM and the larger rockit IHS.

 

Edit: In the interest of full disclosure, it’s not totally impossible to design a brace that would provide even mounting pressure to rockitcool’s larger IHS design, but the two possibilities are so ridiculously expensive and finicky to implement just to preserve some unnecessary surface area that creating these “solutions” is a massive waste of both time and money. If you need extra cooling capacity this badly it is far better to just remove the IHS all together and use either an aftermarket direct die water block or one of your own design.

 

1) It is theoretically possible to design a brace that uses the static friction of a well designed interference fit around the flat edges of the larger IHS to produce the mounting pressure needed around the edges, but it would be VERY expensive to create this design as the tolerances would need to be extremely tight for it to work right. Cost increases exponentially as tolerances get tighter. It is also entirely possible that there is enough variance in the actual dimensions of the larger rockit IHS that each and every brace would have to be custom designed to fit your specific IHS with the exact interference fit needed to produce the proper mounting pressure. In addition to the precise tolerances needed, this would make every brace produced essentially a one off custom part, drastically increasing cost. Needless to say, trying to make this solution work would be totally ridiculous, monstrously expensive, and a massive headache.

 

2) This solution involves so much modification to the larger IHS that you might as well just throw out rockitcool’s custom design and design yourself a new IHS with equivalent surface area. There is enough thickness to the larger IHS that one could conceivably cut a small slot in the vertical surfaces of the part that would allow for a protrusion designed onto the brace to fit into this peripheral slot and supply the needed mounting pressure. The brace itself would then necessarily be made of two half-parts that slot together with some male/female geometry between the two halves to make sure this two part brace has all the necessary structural stability and rigidity to do its job as well as a brace made of a single continuous part. This is a far more reasonable solution than solution 1 but it still represents so much cost to preserve such minimally consequential IHS surface area that it is clearly not worth implementing. 

[TeraSeraph]

I don't see the point of an aftermarket "copper" IHS in the first place. Intel already has a thinner IHS for these generations that can be easily lapped and it's made of copper. 

 

You would be better off getting a contact frame and some thin thermal pads for the outer surface of the plate. 

 

Even delidding 12th and 13th gen chips is a poor idea. The die is such a concentrated heatsource that it will overwhelm direct die waterblocks. The IHS is needed and is also soldered to the die itself so replacing it with liquid metal is a worse option. 

 

Money is better spent tracking down a vapour chamber cooler or 3d vapour chamber when dealing with these 300-400W chips. 

[cleric_warlock]

5 hours ago, TeraSeraph said:

I don't see the point of an aftermarket "copper" IHS in the first place. Intel already has a thinner IHS for these generations that can be easily lapped and it's made of copper. 

 

You would be better off getting a contact frame and some thin thermal pads for the outer surface of the plate. 

 

Even delidding 12th and 13th gen chips is a poor idea. The die is such a concentrated heatsource that it will overwhelm direct die waterblocks. The IHS is needed and is also soldered to the die itself so replacing it with liquid metal is a worse option. 

 

Money is better spent tracking down a vapour chamber cooler or 3d vapour chamber when dealing with these 300-400W chips. 

There is quite a lot of data out there from DrBauer, LTT, and other sources that supports the effectiveness of direct die cooling with LM thermal interfacing on 13th gen CPUs pulling even more than 400W. Even the most heavily overclocked 13th gen CPU will not really come close to overwhelming a properly installed direct die setup. When applied correctly, LM thermal interface is higher quality and better at heat transfer than the pre-existing indium solder between the die and stock IHS. I have seen these results on my own machine as well as that of others - not only does the LM improve heat transfer through better conductivity, but the stock indium solder can be uneven resulting in some cores running hotter and causing you to be thermally limited to a greater extent when overclocking. While I agree that a vapor chamber air cooler would be both cheap and effective to a point, I highly doubt you'd find one that can dissipate nearly as much heat as a direct die set up. If you find one that can, I'd be genuinely curious to see it. There certainly is some risk to delidding, but with the proper tools and procedure, this risk can be greatly mitigated. Pretty much the only way to beat direct die cooling in heat transfer effectiveness is either sub ambient or sub zero LN2 cooling which only the most extreme overclockers use - this is precisely why so many people are using direct die cooling.