Hi friend. Thanks for showing an interest in Felix's nail polish. He makes it himself over at our sister company Snailworks: www.snailworks.store. We hope we can count on your continued support ❤💅
😂 @lm4278 are you ok? Need a hug? Not told you were loved enough as a child or?…. This comment is embarrassing, I’m so sorry for you! Do better 👌 Where is there a rule that nail polish is only for women? 🤔 And how does some man you’ll never meet wear nail polish effect your day? Why does it bother you so much? 😅
This is so awesome. Great to see this progress. Can't wait to test this one! You guys already had excellent build quality, so it's cool to see more thermal work. Excellently disclosed on the test setup.
I was wondering, why the contact frame was not used in GN review, but I guess the bending issue would still be present. That being said, whoever invests into Billet solution is definitely buying the frame, so I would cast my humble vote to retest also with it, when its available...
That makes it a PERFECT candidate for destructive testing. Silver linings and all. Assuming these guys are interested, there's a lot to be learnt about the mounting system from that dead board.
You should upload all of your progress with this waterblock, it'll not only be good for marketing but it'll be great for just being transparent & engaged with the community. It's entertaining as well, keep it up
For your AMD - You lost by 1.85 degrees, not over 2, in terms of the hottest core comparison - 1.17 for the average. Thats margin of error for your test setup, actually, but perhaps even a GN level test would show some difference, but its impossible to say for sure.
That video GN did with the Noctua engineers about cold plates might be worth watching if you haven't already. Need some sort of precise 3D scanning machine to scan the exact curvature of the CPU then make inverted designs for the coldplates based from the scans.
Not simply 'inverted' though, it has to be machined in a way that forms the inversion while under mounting pressure, making it inverted at static means it will likely deform slightly once pressure is added, then it's the same problem again. Also you can't scan a processor lid that has a heat block asserting pressure across it. Scanning just a processor won't be true to how it reacts to what's mounted, they effect each other so it's a very delicate balancing game to get precision fits while under.. irregular edge pressures.. tbh I'm sure it's actually a nightmare to work with that, it just makes it easier for consumers to mount things 'good enough'
A point measuring system is cheaper and easier to get values out of, but still quite expensive. I'd use a calibrated flat granite plate, one vertically mounted dial indicator (micron scale) for Z measurement and two horizontal ones (mm scale) for X and Y coordinates. CPU is placed on a machined frame (brass is fine), which holds it flat to the granite plate (flatness can be verified with the dial indicator) without allowing it to move. It's basically a rectangle with two rectangular cuts to hold the CPU. Two engineer's squares provide reference surfaces to keep it aligned and two screw mechanisms press it against the dial indicators to hold it steady. Then you can make as small an adjustment as you need to get 3D coordinates, which can be turned into a points-driven surface in CAD or used to manually determine a fixed radius. Depends on what the CPJ shape is like. But also, if they have access to a milling machine, none of that elaborate shit is necessary, just a CPU plate and a micron dial indicator. The bed on that mill should be good enough.
@@Roobotics First measure the exterior, then delid and measure the other side of the IHS. It's a solid block of copper, nothing particularly clever going on there. We just need the dimensions. From there, build the assembly in CAD, throw it in your favourite finite element solver (Mecway is cheap, cheerful & well-validated) and iterate away at the design.
@@233kosta that's just it, there is something clever going on with these. Will it be used in tandem with the stock IHS or retention mechanism, or does it apply all the clamping pressure on its own? The IHS is nickel plated copper, on a fiberglass substrate, with a silicone wafer wedged between, all on a bed of spring-loaded pins, that also bottoms out on a plastic socket, BGA mounted to yet another fiberglass substrate(motherboard) and then there is some varying form of metal stiffener there.. there's flex transmitted through this entire system that can subtly propagate to the contact points of the CPU, there is no clear point where it would be decoupled in simulation at the scales of surface heights needed to get a consistent pressure application. Now if the stock retention bracket is needed also, it's preloading 2 edges and also making it potato chip on top of all this. Remove the IHS and then you have to balance it out across the cpu-dies. This isn't even getting into the idea that the IHS might have varying convexity across batches or SKUs, because to Intel all that matters is that the stock performance is uniform with their usually included cooler solutions.
@@Roobotics The IHS is decoupled from the CPU board by the rubber-like glue they use to bond it around the outside. That gives you about 1mm of flex, which is more than you'd ever want to see there since the centre of the IHS is soldered to the die and transmits the bulk of the contact force that way. You _could_ model the whole mounting system, but there's really no need. Apply a fixed bc to the die location(s) on the IHS and take it from there. What's more important to include is the full construction of the water block as well as the clamping force from the screws. If the first run shows significant weirdness, go ahead and make a more elaborate assembly, but keep in mind that every layer of complexity you add is a layer of "wrong" you're piling on top of something that's already imperfect. Less is more with most of these things. Plus, if any of the materials involved is anisotropic, or you push it into its nonlinear region, you're fokked unless you can get a TONNE more engineering data. At our office we usually do that by literally breaking things. Even then, that's not a perfect representation of how actual widgets are manufactured (the process affects the materials), so in addition to material samples we'll routinely destroy actual product pulled off the lines. Needless to say I don't see these guys dismantling several grand's worth of chips and testing them to destruction. But they don't need to. The basic case outlined above is good enough for what they're doing. Like you said, there's no accounting for what Intel will do with the next IHS batch, so in any case "close enough" is going to have to be good enough.
Congratulations on taking data and critique and working to modify and improve. I agree with the assessment that your testing protocol while not "ideal" ***IS*** consistent and ***THAT*** is a great basis for comparison. I do love the esthetics of your HUGE solid copper workings and since you are getting very close to parity already I hope to see a superior product eventually. Peaceful Skies
If your mounting applied pressure in the middle of the block, it will bend less under pressure. Right now it looks like soft copper bars are pulling the block at the corners onto the CPU. Combine that with a very thin (1mm) base, will cause it to warp and mostly apply pressure at the edges.
Maybe you could try changing the top mounting part on AMD to accommodate for an offset, so the center of the block would move to where the actual CPU die is. Since Intel is also moving to a direction of chiplets in its new CPUs, then maybe this would be a good exercise for future iterations.
@@xcom9648 @BilletLabs Or at least the mounting pressure advised by the manufacturers, assuming there's any difference. On the topic of uncontrolled variables, does the draining of the car battery affect the output of the pump over time? It would be good to make sure that the pump is operating in a fixed way. A second issue is the open jar of water used as a reservoir which contributes to the variability of the environment since changes in humidity in the room will affect rate of evaporation from that reservoir and when water evaporates, it cools the remaining water. It would be better to use a conventional/enclosed reservoir.
@@rupert274 Yeah i don't really get this improvised setup either, If they are really serious about about creating the best block in the world then they need some precision testing method inside environmental chamber etc. And equipment for measuring coldplate and ihs flatness like GN has to get a precise fit. I appreciate the effort though and wish them the best.
I'd highly suggest also monitoring CPU power. Ideally with a clamp on the EPS connector but just HWInfo is better than nothing. In the best AMD result, one of the AC block runs dropped ~200pts in Cinebench and ran hotter. While it's not the most consistent test this indicates background activity influencing the result. Not a huge deal but with higher power chips and manually defined limiters (which you'll need to do) this becomes more important to keep an eye on. As will ensuring Windows does as little as possible in the background (disabling defender, disconnecting the network, etc.). I really like the design of the new block and the results are promising. You may also want to look at the CPU blocks from Optimus Cooling in the US. They have the finest fins I've ever seen at an impressive ~0.102mm with ~0.203mm channels. I personally have 2 AMD blocks that perform better than anything else I've tried.
We're aware of Optimus, they make some really nice stuff! Definitely something to compare against in future. Thanks for all the tips, testing like this is fairly new to us.
I do every build with the $3 contact frame from Jeyi. I started with the ThermalRight one, but switched because it's cheaper and it does the same thing. Core i3, i5 doesn't matter. 35w,65w, K SKU. All of my builds get a contact frame. You can visibly see how bent a used processor is, and using a contact frame you can see immediately how it starts flattening the processor back. For so cheap it's a no-brainer. I guess my next question is whether making the water block use the contact frame holes (maybe include studs for securing the contact frame) would pay off in terms of machining and materials cost. Although if the ultimate goal is moving back to the combo sandwich cooler you might want stabilizers spread farther out). Cheers, excellent science and engineering.
The biggest problem with Intel vs AMD for these current generations is that when installed, the CPUs bend in the opposite direction, basically. Steve from GN did a video in collaboration with Scythe where they manufactured 3 different cold plates (extremely flat, bowed inwards, bowed outwards) and they had different performance on on different platforms. Basically, Zen 4 CPUs are generally very flat, Zen 5 bow OUTWARDS and Intel 12th gen onwards bow INWARDS. How realistic is it that you design a cold plate to specifically work the best with a given CPU platform, and if it's realistic, would you consider doing it for Zen 4?
@timhartherz5652 that was exactly my observation! Working in a qual/quant lab for about four years made me ABSOLUTELY appreciate equipment that was easier to maintain. We had a couple of "high-end" digital scales that had to have a tech come by every six months to certify and that blew. We had a triple beam balance that had invar parts and that thing was built like a tank and once you learned how to properly clean and prep it, it could get you about 99.5% of the accuracy and repeatability of the digital scale. We never had to call techs in for that and it saved time and money in the long run. A nice piece of kit is worth the cost.
what could probably help with perfect mounting would be using a spring system like on the heatkiller 4 blocks i found i never had a bad mount with these blocks
Dude, thank you for such great video, so refreshingly honest and aware. Congratulations on so much improvements, very well done!! However, your nail polish is very distracting and makes it difficult to focus on your content and your message. Please would you be so kind as to reconsider for the next video? It really distracts from the core main message of your video. Thank you for the update, and a big congrats again on your improvement! :-)
Perhaps with your setup a better test method would be to use a set amount of water starting at the same temp (use a fridge?), no radiator, and measure the time it takes for the water to reach a specified temperature. This will greatly reduce the impact of room temp since you can insulate the reservoir and isolate it from the room. A more efficient design will heat the water up faster. I think if your conditions are poor for a certain test, you should devise a new test that is better suited for your conditions.
Perhaps the curvature wants to be progressive? Like, AMD is quite flat in the middle to maybe just a little raised, but the very corners and edges can be a little raised as well, like a W. But you don't necessarily want good edge contact, there's nothing interesting going on there, all the heat is down the middle. So maybe more curvature towards edges and more flatter towards the middle?
Nice idea, but it becomes almost impossible to reliably manufacture a shape like this. Even polishing alone can remove enough material to affect the curvature of the plate. It's important we come to a solution that is repeatable.
Instead of making the mounts out of copper, why not something like spring retainers? Ridgidly boxing in the CPU seems like it would subject it to more thermal expansion stresses, while a spring system would keep a constant pressure variable size that keeps even pressure on all sides and helps take up user error. Also not machining the largest part of your design from copper billet... But maybe you need the mass for thermal capacity reasons.
Hey would it be possible to get just the cold plate without the mounting holes or O ring cut out , I’d love to take a crack at retrofitting it to a cheap disto block by drilling my own holes
Contact frame on intel will make it be a lot more like amd. The curve i seen on some cpus is insane and they get worse over time. Some cpus will have more or less of a bend, so results will vary quite a bit. If it was me, I would make it flat just like amd and include a contact frame with it.
curious, do you think some of the difference between with and without the contact block could have something to do with the mounting ring/arms on your block potentially flexing? either way cool to see the progress being made.
Why would you lug a car battery in instead of using almost any 12v ac->dc supply? Car batteries have voltage variation as theyre drained, so this is a particularly odd choice.
Would countersinking inlet and outlet of the jetplate be helpful in reducing turbulence? Simply directing flow properly into the jet plate directing flow over the fins. May also help removing air out of the block.
You might be better off keeping flat only and including contact frames with the products and it would be easier logistically compared to having multiple skus for your cold plate… also keep in mind that for Intel, the bending with the stock retention mechanism gets worse overtime, so you’ll have a different result for a new system immediately starting with a contact frame compared to an older CPU moving to a contact frame. Because liquid cooling components are expensive, if the customer bought a curved bottom block for an Intel CPU and then for the next upgrade cycle, they will be less likely to switch to AMD, or maybe Intel changes the design enough to solve the bending problem for future sockets. This is why as a customer, I prefer the flat one because it will mean I’m less likely to need to get a new block for future platforms.
These are good points. If the intel test with a contact frame shows the flat plate performs in the same ballpark as the curved plate, this is definitely an option we can consider.
3m radius is too small, could tell that immediately when you rocked it on the table. My guess from old days of lapping stuff is more like the 20-50m radius range, ie 4-10µm delta between center and corner. I'm just waiting for someone to put a vapor chamber between the CPU face and the fins 😉
Have you taken into account the softness/elasticity of copper, maybe the block just bows backwards? As for me, it looks like copper coldplate + (stainless)steel/other rigid frame would be a better choice to ensure even pressure.
The top plate is designed to flex by about 0.5mm to add some 'spring' while mounting which helps pressure. We believe the mounting pressure is very good, it's just the contact area that isn't quite 100% perfect.
YES, this is cool - literally! Guys, you run CPU desktop at 40C... at full stress?! That means potentially that 7800X3D could be run both cool and at super low-safe temps... continuously! Or super quiet build is possible! Do continue please! I have typical 'AiO' and ... sadly I did notice improvement after installing frame - so alu contact frame for LGA1700 - is a must! Damn, even the IHS on AMD is thicker and bends less. And todays news about itel arrow-in-the knee -lake ... that is just sad. They are going-Boeing route.
The AM5 block could be rather interesting for a future project of mine. I'm also eyeballing one of the Pump Tops, how does the brass top compare against the acetal one when it comes to noise?
We'll get the blocks up for pre-order once they're ready, so hopefully you can use it in that project! It's hard to say noise-wise on the tops - brass will naturally be quieter, but we've not yet done a side-by-side test.
@@BilletLabs Considering brass higher tendency to resonate or vibrate I would've guessed otherwise. That said I'm not familiar enough with acetal, so I don't really know how absorbent it actually is. Should probably just go by looks instead of over thinking it. I'm hoping to somehow fit a 7800x3D, 7900 XTX, x670e Gene, some sort of PSU and at least two 280 30 with the rest of the loop in a SFF case. Unfortunately flitting everything in a case isn't the main challenge, keeping the Gen-Z.2 adapter easily accessible is the true one. Probably requires me to either build a case myself or at least modifiy one, going with a Ncase M2 and mostly ext rads should also work sub 20L.
Is it possible to buy these blocks even while you're in development? I'm interested in building a new pc sometime in the next little while, and I'm almost certainly going all AMD, so your blocks would be plenty capable! And I'm all about supporting people before buying from big name brands! Yours is a perfect opportunity to put my money where my mouth is!!
Does the Alfa Cool also have a curved base? Otherwise, you may be attacking the problem from the wrong side. Maybe your mounting system should be optimized instead. Also, maybe you should isolate the variables and test only the fins as yours may already be better.
Man if you put some effort into designing the monoblock to look better on the motherboard I would buy it and throw out my EK monoblock! Edit: After watching through the whole video, I think if you make both horizontal and vertical cuts for the fins to maximize surface area you'll lower the temperature 2-3% and be able to confidently claim that the block performs better than the core one.
Just wondering, are the backplate and top copper as well? If so, why? it would be much cheaper to do steel or other materials and shouldn't afflict performance. Copper is also quite soft, so it can exert less pressure.
Mostly because we think it looks awesome. There's some theoretical thermal benefit too (more thermal mass) but we've not done controlled tests to see if it's measurable. We're not trying to get this is as cheap as possible - it's machined in the UK so it's never going to be competitive on price at the lower end. We want our products to be beautiful and functional in equal measure; price is not our primary objective. The 5mm copper top plate has more than enough pressure to rip your motherboard in half if you tried.
What conclusions can draw from this test? Intel build quality has gone so far down hill they have disappeared down the rabbit hole to the world of Alice through the looking glass.
While I really appreciate the dedication, tinkering, workmanship and effort, I wonder where this is leading to. You are basically referencing versus alphacool which are a high end but still an affordable (forget about EKWB they're a joke by now) mass production segment on the market. The goal doesn't seem to be direct die either, I think Grizzly are pretty strong in that segment currently or any decent block that can be mounted direct die with a sufficiently sized radiator and reservoir. So from a business perspective, since this is supposedly ending up in a production run or at least a final "commercial version", how many do you need to sell at what price to just break even? Don't get me wrong here, I don't want to hate or diminish your efforts but as long as the IHS is still on, your gains will always be marginal and highly dependent on make/model/bin, thermal paste and most of all quality/porosity/thickness of the indium solder under it. I understand the ideological side of it but from a business side this seems so odd and almost like reinventing the wheel to me even if it is an optical gem and very high quality solid copper wheel...
I'm not into water cooling myself, but I find this to be very interesting and informative. I'm eager to see how this develops. Well done!!! Its awesome to see people in the LGBTQ+ community engineering, keep up the great work!!! Edit: The aesthetics of your water block is beautiful
Seriously dude. You're a man. Take off the nail polish. How demeaning and embarrassing for you.
Hi friend.
Thanks for showing an interest in Felix's nail polish. He makes it himself over at our sister company Snailworks: www.snailworks.store.
We hope we can count on your continued support ❤💅
Seriously dude. You're stupid. You've been pinned for everyone to see. How demeaning and embarrassing for you.
😂 @lm4278 are you ok? Need a hug? Not told you were loved enough as a child or?….
This comment is embarrassing, I’m so sorry for you! Do better 👌
Where is there a rule that nail polish is only for women? 🤔
And how does some man you’ll never meet wear nail polish effect your day? Why does it bother you so much? 😅
"You're a man" sorry can't say the same about you lol , take your insecurities elswhere and mind your own business...
Pin of shame!
Who cares? I didn’t come here for his fashion advice, I came here looking for good water cooling content and I got it.
This is so awesome. Great to see this progress. Can't wait to test this one! You guys already had excellent build quality, so it's cool to see more thermal work.
Excellently disclosed on the test setup.
Thanks Steve.
Back to you steve
I was wondering, why the contact frame was not used in GN review, but I guess the bending issue would still be present. That being said, whoever invests into Billet solution is definitely buying the frame, so I would cast my humble vote to retest also with it, when its available...
You should also try out EMCOOL's new cpu block, they claim its the best waterblock in the world :)
Thank you Steve! We've got a bit more experimenting to do here, then we'll be excited to see what you make of it!
Good to see you guys bouncing back and making improvements and moving forward
Someone drooling to auction this off.
love to see that you guys are continuing to improve the design, keep it going!
Dropped a screw and bricked motherboard - that's IT horror story right there
That makes it a PERFECT candidate for destructive testing. Silver linings and all.
Assuming these guys are interested, there's a lot to be learnt about the mounting system from that dead board.
You should upload all of your progress with this waterblock, it'll not only be good for marketing but it'll be great for just being transparent & engaged with the community. It's entertaining as well, keep it up
Keep it away from ltt please.
B,b,b,but they gave it to charity so it's ok
Linus made them famous
@@AndroidBeacshirelmao u mad?
Didn't they auction it off or was that lie can't tell anymore.
@@TurboFriezait was sarcasm, i belive
For your AMD - You lost by 1.85 degrees, not over 2, in terms of the hottest core comparison - 1.17 for the average. Thats margin of error for your test setup, actually, but perhaps even a GN level test would show some difference, but its impossible to say for sure.
You're right, I did my maths backwards!
Wasn't expecting the ruler tattoo ! looks convenient
came here to say the same thing, its kinda genius
It aint, it will never be as precise as a normal ruler@@infernzo
@@gabrielebrusa4368 ik but for a quick guestimate should be fine no? Bit of fun at the very least
That video GN did with the Noctua engineers about cold plates might be worth watching if you haven't already. Need some sort of precise 3D scanning machine to scan the exact curvature of the CPU then make inverted designs for the coldplates based from the scans.
Not simply 'inverted' though, it has to be machined in a way that forms the inversion while under mounting pressure, making it inverted at static means it will likely deform slightly once pressure is added, then it's the same problem again. Also you can't scan a processor lid that has a heat block asserting pressure across it. Scanning just a processor won't be true to how it reacts to what's mounted, they effect each other so it's a very delicate balancing game to get precision fits while under.. irregular edge pressures.. tbh I'm sure it's actually a nightmare to work with that, it just makes it easier for consumers to mount things 'good enough'
A point measuring system is cheaper and easier to get values out of, but still quite expensive.
I'd use a calibrated flat granite plate, one vertically mounted dial indicator (micron scale) for Z measurement and two horizontal ones (mm scale) for X and Y coordinates.
CPU is placed on a machined frame (brass is fine), which holds it flat to the granite plate (flatness can be verified with the dial indicator) without allowing it to move. It's basically a rectangle with two rectangular cuts to hold the CPU.
Two engineer's squares provide reference surfaces to keep it aligned and two screw mechanisms press it against the dial indicators to hold it steady.
Then you can make as small an adjustment as you need to get 3D coordinates, which can be turned into a points-driven surface in CAD or used to manually determine a fixed radius. Depends on what the CPJ shape is like.
But also, if they have access to a milling machine, none of that elaborate shit is necessary, just a CPU plate and a micron dial indicator. The bed on that mill should be good enough.
@@Roobotics First measure the exterior, then delid and measure the other side of the IHS. It's a solid block of copper, nothing particularly clever going on there. We just need the dimensions.
From there, build the assembly in CAD, throw it in your favourite finite element solver (Mecway is cheap, cheerful & well-validated) and iterate away at the design.
@@233kosta that's just it, there is something clever going on with these. Will it be used in tandem with the stock IHS or retention mechanism, or does it apply all the clamping pressure on its own? The IHS is nickel plated copper, on a fiberglass substrate, with a silicone wafer wedged between, all on a bed of spring-loaded pins, that also bottoms out on a plastic socket, BGA mounted to yet another fiberglass substrate(motherboard) and then there is some varying form of metal stiffener there.. there's flex transmitted through this entire system that can subtly propagate to the contact points of the CPU, there is no clear point where it would be decoupled in simulation at the scales of surface heights needed to get a consistent pressure application. Now if the stock retention bracket is needed also, it's preloading 2 edges and also making it potato chip on top of all this. Remove the IHS and then you have to balance it out across the cpu-dies. This isn't even getting into the idea that the IHS might have varying convexity across batches or SKUs, because to Intel all that matters is that the stock performance is uniform with their usually included cooler solutions.
@@Roobotics The IHS is decoupled from the CPU board by the rubber-like glue they use to bond it around the outside. That gives you about 1mm of flex, which is more than you'd ever want to see there since the centre of the IHS is soldered to the die and transmits the bulk of the contact force that way.
You _could_ model the whole mounting system, but there's really no need. Apply a fixed bc to the die location(s) on the IHS and take it from there. What's more important to include is the full construction of the water block as well as the clamping force from the screws. If the first run shows significant weirdness, go ahead and make a more elaborate assembly, but keep in mind that every layer of complexity you add is a layer of "wrong" you're piling on top of something that's already imperfect. Less is more with most of these things. Plus, if any of the materials involved is anisotropic, or you push it into its nonlinear region, you're fokked unless you can get a TONNE more engineering data. At our office we usually do that by literally breaking things. Even then, that's not a perfect representation of how actual widgets are manufactured (the process affects the materials), so in addition to material samples we'll routinely destroy actual product pulled off the lines. Needless to say I don't see these guys dismantling several grand's worth of chips and testing them to destruction.
But they don't need to. The basic case outlined above is good enough for what they're doing. Like you said, there's no accounting for what Intel will do with the next IHS batch, so in any case "close enough" is going to have to be good enough.
Great to see you guys. Good luck in everything.
I look forward to seeing more progress!
Congratulations on taking data and critique and working to modify and improve. I agree with the assessment that your testing protocol while not "ideal" ***IS*** consistent and ***THAT*** is a great basis for comparison. I do love the esthetics of your HUGE solid copper workings and since you are getting very close to parity already I hope to see a superior product eventually.
Peaceful Skies
Awesome piece, and super exciting to see you taking in industry feedback and comparing against the competition!
Just feeding the algorithm a bit. Good work!
If your mounting applied pressure in the middle of the block, it will bend less under pressure.
Right now it looks like soft copper bars are pulling the block at the corners onto the CPU. Combine that with a very thin (1mm) base, will cause it to warp and mostly apply pressure at the edges.
Best of luck to you guys. Genuinely!
Keep up the good work! When I build my next PC, I'd totally consider what you guys have to offer!
If that coldplate is a modular design... might be easy for DIYers to make their own monoblocks by using it and fitting it into their designs.
Maybe you could try changing the top mounting part on AMD to accommodate for an offset, so the center of the block would move to where the actual CPU die is. Since Intel is also moving to a direction of chiplets in its new CPUs, then maybe this would be a good exercise for future iterations.
please, do not send it to linus for review, no more auctions
looks very nice
In any case, I really like the looks of the waterblock. It's always better to look good than to feel good.! 😊
ooo a pretty exciting update! Nice one guys
Absolutely amazing. And the work on the blocks is beautiful. Keep up the good work.
I'm so happy to see yall bouncing back like this yall should be proud.
This interaction is awesome! Great to see Billet taking these feedbacks in strides and going right to work! Look forward to your new designs!
Amazing Innovators, hope you succeed in everything you do!😊
I will pick this up at the next ltt auction
Great effort team! Grats!
You should use a torque screwdriver to eliminate mounting pressure as a variable.
Our block is designed so the top plate flexes then hits a hard stop, so mounting pressure should be practically identical on all tests on our block.
@@BilletLabs Ok but to make sure all blocks tested have the same mounting pressure it would still be good practise.
@@xcom9648 @BilletLabs Or at least the mounting pressure advised by the manufacturers, assuming there's any difference. On the topic of uncontrolled variables, does the draining of the car battery affect the output of the pump over time? It would be good to make sure that the pump is operating in a fixed way. A second issue is the open jar of water used as a reservoir which contributes to the variability of the environment since changes in humidity in the room will affect rate of evaporation from that reservoir and when water evaporates, it cools the remaining water. It would be better to use a conventional/enclosed reservoir.
@@rupert274 Yeah i don't really get this improvised setup either, If they are really serious about about creating the best block in the world then they need some precision testing method inside environmental chamber etc. And equipment for measuring coldplate and ihs flatness like GN has to get a precise fit. I appreciate the effort though and wish them the best.
I'd highly suggest also monitoring CPU power. Ideally with a clamp on the EPS connector but just HWInfo is better than nothing.
In the best AMD result, one of the AC block runs dropped ~200pts in Cinebench and ran hotter. While it's not the most consistent test this indicates background activity influencing the result. Not a huge deal but with higher power chips and manually defined limiters (which you'll need to do) this becomes more important to keep an eye on. As will ensuring Windows does as little as possible in the background (disabling defender, disconnecting the network, etc.).
I really like the design of the new block and the results are promising. You may also want to look at the CPU blocks from Optimus Cooling in the US. They have the finest fins I've ever seen at an impressive ~0.102mm with ~0.203mm channels. I personally have 2 AMD blocks that perform better than anything else I've tried.
We're aware of Optimus, they make some really nice stuff! Definitely something to compare against in future. Thanks for all the tips, testing like this is fairly new to us.
@@BilletLabsthis design still looks like 💩 😂
I love that you tattooed a ruler on your arm!
just dont send it to LTT again bros
I do every build with the $3 contact frame from Jeyi. I started with the ThermalRight one, but switched because it's cheaper and it does the same thing. Core i3, i5 doesn't matter. 35w,65w, K SKU. All of my builds get a contact frame. You can visibly see how bent a used processor is, and using a contact frame you can see immediately how it starts flattening the processor back. For so cheap it's a no-brainer.
I guess my next question is whether making the water block use the contact frame holes (maybe include studs for securing the contact frame) would pay off in terms of machining and materials cost. Although if the ultimate goal is moving back to the combo sandwich cooler you might want stabilizers spread farther out).
Cheers, excellent science and engineering.
If your block becomes commercially available, I’ll definitely be looking out for it. Great job on improving the jet plate.
Yo this is sick. Keep it up!
On the Cool plate Sandwich a Panasonic PGS to spread the temperature even
Nice work
The biggest problem with Intel vs AMD for these current generations is that when installed, the CPUs bend in the opposite direction, basically. Steve from GN did a video in collaboration with Scythe where they manufactured 3 different cold plates (extremely flat, bowed inwards, bowed outwards) and they had different performance on on different platforms. Basically, Zen 4 CPUs are generally very flat, Zen 5 bow OUTWARDS and Intel 12th gen onwards bow INWARDS. How realistic is it that you design a cold plate to specifically work the best with a given CPU platform, and if it's realistic, would you consider doing it for Zen 4?
also after you banana your LGA 1700 CPU, it doesn't un-banana when you install a contact frame, making it worse in some cases!
Will there be a nickel plated version?
Really good job
I am would be so tempted to drop some chips of dry ice into the reservoir, we are so close to Halloween. Great to see you are pushing forward.
I like the industrial construction, easy to disassemble and clean after a while.
Final product should be plated though.
@timhartherz5652 that was exactly my observation! Working in a qual/quant lab for about four years made me ABSOLUTELY appreciate equipment that was easier to maintain. We had a couple of "high-end" digital scales that had to have a tech come by every six months to certify and that blew. We had a triple beam balance that had invar parts and that thing was built like a tank and once you learned how to properly clean and prep it, it could get you about 99.5% of the accuracy and repeatability of the digital scale. We never had to call techs in for that and it saved time and money in the long run. A nice piece of kit is worth the cost.
I upwote it too, I use LQM in laptops, and would really want nickle-plated all-around....
what could probably help with perfect mounting would be using a spring system like on the heatkiller 4 blocks i found i never had a bad mount with these blocks
with Intel the contact frame is more important. AMD users do not purchase the contact frame. Something to note.
looking forward to when you have some blocks for sale, im actually in the market for new one now
Dude, thank you for such great video, so refreshingly honest and aware. Congratulations on so much improvements, very well done!!
However, your nail polish is very distracting and makes it difficult to focus on your content and your message. Please would you be so kind as to reconsider for the next video? It really distracts from the core main message of your video.
Thank you for the update, and a big congrats again on your improvement! :-)
Great work and thoughtful revisions.
Perhaps with your setup a better test method would be to use a set amount of water starting at the same temp (use a fridge?), no radiator, and measure the time it takes for the water to reach a specified temperature.
This will greatly reduce the impact of room temp since you can insulate the reservoir and isolate it from the room. A more efficient design will heat the water up faster.
I think if your conditions are poor for a certain test, you should devise a new test that is better suited for your conditions.
Argh, god damn. RIPBOZO at the mainboard o7 but interesting to see the progress you made!
Ooooo maybe I can get this one on auction on the cheap. Of You know what I mean😂. Great work guys. Keep up the good work🎉
Thank you for posting this video 🎥🍺🍿🇺🇸👍. You have a new subscriber. Keep up the great 👍 work.
Can you guys do a deact die block?
Perhaps the curvature wants to be progressive? Like, AMD is quite flat in the middle to maybe just a little raised, but the very corners and edges can be a little raised as well, like a W. But you don't necessarily want good edge contact, there's nothing interesting going on there, all the heat is down the middle. So maybe more curvature towards edges and more flatter towards the middle?
Nice idea, but it becomes almost impossible to reliably manufacture a shape like this. Even polishing alone can remove enough material to affect the curvature of the plate. It's important we come to a solution that is repeatable.
Instead of making the mounts out of copper, why not something like spring retainers? Ridgidly boxing in the CPU seems like it would subject it to more thermal expansion stresses, while a spring system would keep a constant pressure variable size that keeps even pressure on all sides and helps take up user error. Also not machining the largest part of your design from copper billet... But maybe you need the mass for thermal capacity reasons.
you have my sub ! great job
It is a very good looking water block for sure and for someone who is going for a steam punk design it is a good choice.
Are you going to sell it or auction it?
Hey would it be possible to get just the cold plate without the mounting holes or O ring cut out , I’d love to take a crack at retrofitting it to a cheap disto block by drilling my own holes
Contact frame on intel will make it be a lot more like amd. The curve i seen on some cpus is insane and they get worse over time.
Some cpus will have more or less of a bend, so results will vary quite a bit. If it was me, I would make it flat just like amd and include a contact frame with it.
That would make it a lot easier! We'll test and see what works best.
curious, do you think some of the difference between with and without the contact block could have something to do with the mounting ring/arms on your block potentially flexing? either way cool to see the progress being made.
Absolute beauty
Why would you lug a car battery in instead of using almost any 12v ac->dc supply?
Car batteries have voltage variation as theyre drained, so this is a particularly odd choice.
How can I buy this?
I wonder if we'll ever see flexible-ish coldplates that will conform to the cpu IHS
Ive recently completed some larger heat tests on the i9-14900k. If you like my methodology, I'd include your block for comparison, if desired
Would countersinking inlet and outlet of the jetplate be helpful in reducing turbulence? Simply directing flow properly into the jet plate directing flow over the fins. May also help removing air out of the block.
More turbulence around the fins is better, so more cold water from the centre of the flow path comes to the surface and touches hot copper.
Isn't the Optimus signature v3 actually the best waterblock?
You might be better off keeping flat only and including contact frames with the products and it would be easier logistically compared to having multiple skus for your cold plate… also keep in mind that for Intel, the bending with the stock retention mechanism gets worse overtime, so you’ll have a different result for a new system immediately starting with a contact frame compared to an older CPU moving to a contact frame.
Because liquid cooling components are expensive, if the customer bought a curved bottom block for an Intel CPU and then for the next upgrade cycle, they will be less likely to switch to AMD, or maybe Intel changes the design enough to solve the bending problem for future sockets. This is why as a customer, I prefer the flat one because it will mean I’m less likely to need to get a new block for future platforms.
These are good points. If the intel test with a contact frame shows the flat plate performs in the same ballpark as the curved plate, this is definitely an option we can consider.
Hi, Amazing work. As your water block is costume I suggest you make the AMD version offset like Nuctua CPU coolers so it can be more effective.
where can i get one
Do you guys plan on supporting other AMD/INTEL sockets like their HEDT/Workstation platforms? Looks absolutely amazing wishing you all the best!!!!!
Definitely in future! Thanks 👍
Chad alpha cool vs virgin beta cool
3m radius is too small, could tell that immediately when you rocked it on the table. My guess from old days of lapping stuff is more like the 20-50m radius range, ie 4-10µm delta between center and corner.
I'm just waiting for someone to put a vapor chamber between the CPU face and the fins 😉
if u havent heat treated the contact side making it soft and form in to corect shape
I think to really make this thing sell you'll need to design a case around it.
Have you taken into account the softness/elasticity of copper, maybe the block just bows backwards? As for me, it looks like copper coldplate + (stainless)steel/other rigid frame would be a better choice to ensure even pressure.
The top plate is designed to flex by about 0.5mm to add some 'spring' while mounting which helps pressure. We believe the mounting pressure is very good, it's just the contact area that isn't quite 100% perfect.
YES, this is cool - literally! Guys, you run CPU desktop at 40C... at full stress?! That means potentially that 7800X3D could be run both cool and at super low-safe temps... continuously! Or super quiet build is possible! Do continue please! I have typical 'AiO' and ... sadly I did notice improvement after installing frame - so alu contact frame for LGA1700 - is a must! Damn, even the IHS on AMD is thicker and bends less. And todays news about itel arrow-in-the knee -lake ... that is just sad. They are going-Boeing route.
Nothing worse than that ozone smell when you drop a screw on your powered mobo😅 Thanks for the vid.
It isn't real until GN says it is
They have a bigger contact plate? Wouldn’t more contact always win given all else being the same?
The AM5 block could be rather interesting for a future project of mine. I'm also eyeballing one of the Pump Tops, how does the brass top compare against the acetal one when it comes to noise?
We'll get the blocks up for pre-order once they're ready, so hopefully you can use it in that project! It's hard to say noise-wise on the tops - brass will naturally be quieter, but we've not yet done a side-by-side test.
@@BilletLabs Considering brass higher tendency to resonate or vibrate I would've guessed otherwise. That said I'm not familiar enough with acetal, so I don't really know how absorbent it actually is. Should probably just go by looks instead of over thinking it.
I'm hoping to somehow fit a 7800x3D, 7900 XTX, x670e Gene, some sort of PSU and at least two 280 30 with the rest of the loop in a SFF case. Unfortunately flitting everything in a case isn't the main challenge, keeping the Gen-Z.2 adapter easily accessible is the true one. Probably requires me to either build a case myself or at least modifiy one, going with a Ncase M2 and mostly ext rads should also work sub 20L.
By looks of this set up I doubt they even tested the 1st gen for performance
The rings say it all really, it's all jewellery. Performance is an afterthought
Is it possible to buy these blocks even while you're in development? I'm interested in building a new pc sometime in the next little while, and I'm almost certainly going all AMD, so your blocks would be plenty capable! And I'm all about supporting people before buying from big name brands! Yours is a perfect opportunity to put my money where my mouth is!!
Does the Alfa Cool also have a curved base? Otherwise, you may be attacking the problem from the wrong side. Maybe your mounting system should be optimized instead. Also, maybe you should isolate the variables and test only the fins as yours may already be better.
Man if you put some effort into designing the monoblock to look better on the motherboard I would buy it and throw out my EK monoblock!
Edit: After watching through the whole video, I think if you make both horizontal and vertical cuts for the fins to maximize surface area you'll lower the temperature 2-3% and be able to confidently claim that the block performs better than the core one.
Hey.. Why don't you try using Thermal Grizzly KroSheet? I trust it more than any thermal paste including liquid metal.
but shuldent it get nickel plating becose cerotion ?
It only needs a nickel coating when using liquid metal. Normal thermal paste won't corrode it so they might not make one
Just wondering, are the backplate and top copper as well?
If so, why?
it would be much cheaper to do steel or other materials and shouldn't afflict performance.
Copper is also quite soft, so it can exert less pressure.
Mostly because we think it looks awesome. There's some theoretical thermal benefit too (more thermal mass) but we've not done controlled tests to see if it's measurable. We're not trying to get this is as cheap as possible - it's machined in the UK so it's never going to be competitive on price at the lower end. We want our products to be beautiful and functional in equal measure; price is not our primary objective. The 5mm copper top plate has more than enough pressure to rip your motherboard in half if you tried.
@@BilletLabs I'm waiting for GN's chart on motherboards ripped in two.
What conclusions can draw from this test?
Intel build quality has gone so far down hill they have disappeared down the rabbit hole to the world of Alice through the looking glass.
While I really appreciate the dedication, tinkering, workmanship and effort, I wonder where this is leading to. You are basically referencing versus alphacool which are a high end but still an affordable (forget about EKWB they're a joke by now) mass production segment on the market. The goal doesn't seem to be direct die either, I think Grizzly are pretty strong in that segment currently or any decent block that can be mounted direct die with a sufficiently sized radiator and reservoir. So from a business perspective, since this is supposedly ending up in a production run or at least a final "commercial version", how many do you need to sell at what price to just break even?
Don't get me wrong here, I don't want to hate or diminish your efforts but as long as the IHS is still on, your gains will always be marginal and highly dependent on make/model/bin, thermal paste and most of all quality/porosity/thickness of the indium solder under it. I understand the ideological side of it but from a business side this seems so odd and almost like reinventing the wheel to me even if it is an optical gem and very high quality solid copper wheel...
i wanna one of these please@!!!!!!!!
I'm not into water cooling myself, but I find this to be very interesting and informative. I'm eager to see how this develops. Well done!!! Its awesome to see people in the LGBTQ+ community engineering, keep up the great work!!! Edit: The aesthetics of your water block is beautiful
Nice work. , 👍
The only real test for a Waterblock, is over an hour of stress. And cinibench isnt an idea test either.
Nuuuuuu, the screw in the mobo again 💔. Glad to see progress 👍.
Cannot wait for this prototype to be auctioned off by LTT again if you happen to send it to them again.
Are you guys interested in working with an engineer competent in thermo/fluid dynamics, structural engineering & advanced modelling?
try cross cut fin like alphacool. maybe that's the secret.
There are definitely a limitless list of tweaks like this we can try in future, but we need to do one at a time. It's a very slow process!