The way I was taught in college to free hand lap was to do 10 strokes applying pressure in only one direction towards you with the most pressure being applied in the center with one finger. Then rotating the object 90 degrees and repeating the 10 strokes for all four directions. There were bonus marks for the ones that were able to wring the two lapped objects together and have them stick together proving their true flatness.
He needs to grind the shoulders off the bottom of the IHS and make sure BOTH the top and bottom are true flat as well. Also here is a question i've never seen answered about guage blocks: Can you wring ceramic to metal?
i would recommend you to go with a 280mm aio instead of using the high end air coolers for the best temperatures with direct die cooling. I initially tried cooling my 9900k using the noctua nh-d15 and it is possible but its not worth it to me. Right now i am running the 9900k direct die with liquid metal using the corsair h115i and the max load temperatures i am getting right now are 8-9c better than with the noctua nh-d15. Now i am thinking or reducing the die height to 0.4mm using this tool, but sadly derbaur did not included the before and after results.
@@bloomfield295 9900k has a TPD of 95w as stated by Intel, but reviews have popped up that 95w cooling is not enough. NH-D15 is rated for 165w. In theory, that's enough thermal dissipation for a good overclock, but in practice, it really is not. While I understand that CLC will bring more thermal headroom for overclocks, the usage of air coolers is still a good option for people that do not want to deal with CLC failures. Anyways, can you share your current temps with Δ's? Would appreciate it.
@@notyouraveragegamer130 95w is on stock frequencies and without turbo boost (or properly working turbo boost, one that doesn't allow you to boost indefinitely).
@@notyouraveragegamer130 Will share my results soon but as far as running a 9900k on stock clocks, a noctua nh-d15 is more than good. The thing is if some one is looking for running a cpu direct die then surely he is looking for the best thermal performance and to achieve that a 280mm 'aio' or a custom loop is needed, as i already mentioned, noctua nh-d15 will easily run 8-9c hotter than a 280mm aio and i am absolutely certain about that. Thats why i suggested against air cooling. But yeah air cooling has its own pros thats for sure.
No matter how good this product is you'll never be able to buy it. I've been waiting for the Ncore CPU block from these guys for well over a year, never any updates, never in stock, always pre-order, no date. Never replies to any emails. Hats off to this guy for making something unique and impressive but its pointless if you cant buy it, the guy is a hobbyist, not a business man.
Also a backer on the original Kickstarter for the Ncore, years later still noproduct in hand while he goes on to use kickstarter funds to buy CNC equipment and start making new products...
That tool is called micrometer. I used it to measure copperwires from 0,7mm to 0,01mm, when I worked in cable factory. Thanks for the video as allways.
If the cpu doesn't fit the tool, I guess they add a notch to catch the pcb, same as the packaging, no? Hopefuly, Roman will feedback and there'll be improvements
Sounds like buyers should measure the IHS, then apply some tape like painters tape to the top of the IHS before inserting in the tool. But if you are measuring the tool there is not much point in using the Acrylic Jig anyway. Just measure on all 4 corners and use the "walk around the barrel" sanding technique.
tape isn't perfect in thickness. nothing really is unless its machined for that. and when your talking about super precise measurements ,you want to have something guiding you. with the die itself, one wrong move and you will destroy it. and with the ihs. if it is a little off , you will have uneven mounting pressure against the die. this will cause it to crack when you to crack when you place the cooler. die lapping is risky business. way more risky then ihs top lapping. but i guess its the price you pay to have the best overclocks lol.using the tool stops you from rolling the edge when you sand. you have to take a ton of acrylic off before you would make an uneven sand. and you would see it happening buy the paint being removed more on one side .the tool isnt precise with its measurements. but it does make it way harder to mess up
Hey Derbaur, i would request you to share those die sanding results. If possible please sand down that 9900k to as low as 0.4mm and mount it direct die and please share the before and after results.
I could TOTALLY whip out a 3D printable design for these tools...one that would allow the "gauges" to be replaced after use because this acrylic unit looks like you must throw it away after lapping?
And a lab grade surface plate and an octoculimator to check the final result... No, seriously, calling it "CPU lapping" is flat out wrong. It is sanding it down. If we were talking lapping in the machinist sense, he would get better results with no liquid metal, as the surfaces would wring together like gauge blocks.
put a plate with a pin to fasten the drill to it, machine a flat plate with a pin sticking out to attach the drill and glue it to the tool then just go, either the drill press or hand drill, drill press would be the best. maybe not for the cpu though :p
Be interesting to see if it actually left the die flat, with the difference in material while sanding it should be easy to sand more off the tool with uneven pressure, maybe time to get a test indicator and testing the flatness of the die. I also think not lapping the top of the IHS flat seems like a waste would also negate the scratching,
I always use Break-Cleaner because it doesn't leave any residue. They also use that to see if (motor engine) valves are completely tight after polishing them.
Proper acetone doesn't leave any residue either. The stuff I buy at the local super market says "chemically pure acetone" on the bottle and it flashes off as cleanly as alcohol.
@der8auer I have one question which could be good video material. *Is there any actual efficiency difference between full copper water radiators compared to aluminium ones?* And are copper ones actually better? We all know, that copper has higher thermal conductivity, but from what I've researched, this alone is meaningless in radiator performance. Copper has *3.3* times the density of aluminium, while having only *1.75* times the thermal conductivity. That means that you need *1.7-1.85* times as much copper in weight to achieve similar overall thermal conductivity, which seems *not to be the case.* Not only are aluminium radiators heavier then necessary for theoretically identical results, but they're also *typically denser,* with more tubes and therefore higher overall area for cooling. The fins also seem to be thicker to offset lower conductivity. I've also found some topics on automotive forums, where copper radiators are not recommended for their inferior performance. Would you try checking that?
In theory, yes it also has been tested and proven although don't expect huge changes. The biggest impact comes from the cpu block which is now copper/nickel not from the radiators . And is not in a aluminium loop setup. Radiators like you mention come in al sizes prices config so if you have enough "radiator(s)" you will cool the loop. And i want to point that out because you see claims from 10 years ago saying a big 280, 360 is more than enough. i Will always have way more radiator than necessary
Wow lapping the die itself?! I thought I misunderstood when I first heard it on the video… o.O For a 2 degree difference that's just crazy but it's nice to have someone with enough money to do those crazy things we common folks can't afford to risk it lol
I wonder what the difference would be between lapping the CPU die and a silver heat spreader. Both idea are both equally crazy so it would be interesting to see if they have equal performance.
When are you going to make a AMD AM4 IHS lapping tool? I had an idea of using a broken AM4 board, just remove the entire socket and mounting mechanism, then remount it to a easy to handle block of acrylic
manufacturer used blue tac to hold in place, accurate measurement device = digital micrometer. could we just use a aftermarket copper ihs from an 8700K and skip the lapping?
What about keeping the IHS and die parallel? Sure the silicon is now thinner, but if the die and IHS are not parallel wouldn't that create a gap between the two leading to a potential decrease in thermal conductivity? The thicknesses matter, of course. For instance if you sand down 0.2 mm of silicon and now you have a 0.1 mm gap filled with Conductonaut on one side of the die overall there would probably be an improvement to thermal conductivity because silicon is roughly twice as thermally conductive as liquid metal and that gap would only exist on one side of the interface between the die and IHS. Thermal conductivities: Silicon 149 W/mK, Copper 400 W/mK, Thermal Girzzly Conductonaut 73 W/mK. Seems like if the two were not parallel there could be a decrease in thermal conductivity if using a TIM like Conductonaut. When sanding down the IHS tabs is enough material removed to cause interference between the die and the IHS before the IHS touches the interposer? This would allow the inside of the IHS to come in contact with the die and self align to it before the tabs touch the interposer. Sorry for all the questions, but I am new at this. A surface can be very flat and very smooth, but when you bring two flat, smooth surfaces together they need to be parallel to eliminate gaps. When you lap a CPU cooler this isn't an issue because it is allowed to self align to the top of the IHS because of how the cooler is mounted. For lapping the die, there could be an error stackup in parallelism between the die, the interposes, the tabs of the IHS and the inside of the IHS. Overall I like the video and appreciate your contribution to the technical content on TH-cam. Thanks.
I just watched the video from Oct 19, 2018 about sTIM on the 9900K. Moving from Indium solder in the stock thickness to Conductonaut made a big difference (8 °C). I can also see that the IHS was sanded down in this step to account for the reduction in overall height from removing the indium. Sanding the die and IHS from there provided about 1.25 to 1.5 °C per 0.05 mm. While I still have the question about maintaining a parallel interface between the die and IHS, I can see that reducing the thickness of the indium and silicon do make a big difference. I suspect the IHS tabs, or "feet" have been sanded down sufficiently to not make contact with the interposer before the IHS contacts the die.
lapping the top would not bring the IHS down to make contact with the die. i do see what you are saying tho, it does look like it would fit better that way.
Isn't laping the top of the IHS always the better solution? For the best cooling contact at least? Can you use this tool for that purpose? Or only inner side lapping?
@der8auer : 8:35 The micrometer used has a 0.001mm resolution. I will the NLap CPU tool to lap my delidded i9-9900K. Is sufficient to use instead a caliper with 0.01mm resolution to track the amount of silicon effectively removed? Should I plan to sand the top of my der8auer direct die Intel 9th Gen. OC-Frame (anodised aluminium) by the same amount? If so nail varnish will be enough to protect it against corrosion from liqui metal contact? Thank you all for inputs.
You need to get a small but extremely precise cnc milling machine. You could have milled the die and ihs and finish it off with lapping for a smooth surface finish. I would be interested to see you take this concept to the extreme. Find some silicone lottery, shave it down as low as you can, and then do the same thing on a 2080 ti or titan. You can even use the machine to make your own custom parts for builds. Custom blocks, distroplates, case parts, etc.
Gotta give Intel some credit for that stunt they pulled going from Sandy Bridge to Ivy Bridge. Nobody in their right mind would delid, use liquid metal, lap an IHS or do direct-die 5-10 years ago. Content like this probably wouldn't exist.
On my 9900KF I lapped the top of the IHS because it was quite convex, and the temps would jump around and spike very quickly and the CPU would thermal throttle under heavy load, my temps have dropped by over 10c, at idle and around 20c under heavy load, I'm not using liquid metal. I only use normal thermal paste (it is a Cooler Master brand paste) and I have not de-lidded the CPU. I was considering de-lidding it and getting a direct die cooling kit but I'm not too confident in doing it to such an expensive CPU. But considering it now runs stable at 5.1GHz all cores with a max temp of 80c. I have been satisfied with the outcome in regards to temperature. So my question is; is it ok to lap the top of the IHS as I have done and not using liquid metal or should I be aware of a potential future issue caused by doing this?
@der8auer, I have delided and used a direct die cooling with your OC frame in my Eurocome Sky X7c Clevo chassis laptop. I used a 3mm vaccum chamber thick and it increased my z-height by up to 0.4mm. I am planning on lapping the CPU to improve my contacts and cooling system tolerance. How much of the i9-9900k die could I lap to stay above OC frame? Should I plan to lap the bottom of the OC frame to offset it? Anyone has an idea of the thichkness of the STIM solder or the height pcb to the top of the IHS before delidding the i9-9900K (lost measurement after delidding)?
Can u make vid where u try overclock stuff with a PSU that only has a +Plus -ceritivication or lower :P and compare it mayeb with some higher end ones just for funzies, might be some unexpected results :P
Not really, don't expect more than 2 degree less. If you sand both the cpu and the coldplate maybe something more than 2 degrees but for all the work isn't really worth it
Ehh this have so many flaws... What about flatness of the acrylic "thing", what about of CPU PCB flatness? what about glass panel flatness, what about finger pressure - all those are imperfect and will affect final result ... You don't even know how hard is to be perfectionist. Nice job as always
Those of us with machinist training think of lapping as something very different than what overclockers think. They say lapping but they mean sanding. Period. If you are interested in how to make a proper lap (that could be used for a CPU if you like) go check oxtoolco here on youtube. He goes through the classic three lap method using only copper roundstock, a hack saw and 5my alumina abrasive.
you would have to make a new nickel coating then to prevent the bare copper on the IHS to react with the indium alloy from the liquid metal on top of it.
9900ks = 9900k with some binning and a "special edition" on the box, nothing more. But you shouldn't buy this crap, if you really have money to burn, wait for 10th gen, they'll be here soon.
@der8auer All these years and still no one has made a motor device for lapping. disappointing. It's not that impossible to design and make a device! Simple concept, have a rotating circle surface with perfect mirror finish. Make it rotatable with a small cheap motor, make a arm that pushes the CPU to the circle surface where you have the sandpaper.
If you are going for overclocking to the extreme levels, then 2 to 3 degrees is a lot. Look at this way, most people who do overclocking are colouring book and crayons. This is level of overclocking where they are painting something like the Mona Lisa quality.
For the IHS should be a magnet placed inside the acrylic to hold IHS from the top, for the CPU should be like a delidding tool, with screw(s) from the (both) side(s) to hold it
@@honzamoravka9430 true, but I used neodyme magnets from hdd and put second one below the ihs, I found some are just thin enough to fit and not interfere when lapping
And result of all you're work is a 9900k at +100mghz OC all cores constantly at 82-90c air and 75-79 under a 240 aio at 300 wats from the wall ... pls tell me it ain't true
This is no different than people tuning cars to the edge. You can get a fair performance boost for a little money and a little effort. After that it becomes more and more expensive and more and more labor intensive.... Chasing the last 1% can quickly cost $100k and take months of work.
it believe the "accurate measurement device" is called "MICROMETER"? Yes? Why not call it by its name? Afraid Steves gonna come memeing? XD (they just have some meme video about nano and micrometers or something? XD)
I have had a 9900K lapped and direct-die cooled with liquid metal onto an EK block since shortly after launch. 5.4 with silly voltages, 5.2 daily. No ill effects yet, but I haven't touched the block since installing so I don't know what it looks like
I guess having a 9900K that only requires 1.290v to run at 5.2GHz, only reaching 90°c after a 20 loop run in R20 with an AIR cooler defeats the purpose of this tool. I also lucked out on the 1.280v 5.3GHz 8700K that i got open box, also topping out at 90°c on AIR. What's with all the water cooling...lol Nothing against AIO's. I've been through many and the only ones that actually cooled better than some of the best AIR coolers was Corsair's V115 and V80 Extreme V2's with the double thick radiators and some vardars. Though, with that amount of money involved in AIO's and some worthy fans, you gotta be an enthusiast. I just like testing old and massive AIR coolers. Anyone remember the first 8 heatpipe Silver Arrow monster, or how about the ZALMAN 9900MAX, 9900MAX DF or Triple fan 12X. Might sound funny to some but, beats paying $100+ for something that essentially puts up the same numbers as a 5 to 10 year old AIR cooler. That can be found for $25 to $40. Gotta have a grade a chip though.
And if the only source of plastic residues in the world was from people like Der8auer lapping a cpu, let's assume that the problem would be really, really not a big deal
He's perfectly understandable, and his accent is pretty good for a German, better than a lot of native English speakers even. Maybe you should listen with better knowledge of the English language?
That t-shirt, i need that in my life
Me too.. I was wondering where to get it
The way I was taught in college to free hand lap was to do 10 strokes applying pressure in only one direction towards you with the most pressure being applied in the center with one finger. Then rotating the object 90 degrees and repeating the 10 strokes for all four directions. There were bonus marks for the ones that were able to wring the two lapped objects together and have them stick together proving their true flatness.
This...
He needs to grind the shoulders off the bottom of the IHS and make sure BOTH the top and bottom are true flat as well. Also here is a question i've never seen answered about guage blocks: Can you wring ceramic to metal?
wondering what major did you study ?
@@Alhussainba CS
@@NVMDSTEvil ceramic breaks too easily for that
3dlabs, 3dlabs.. i know its not remotely consumer, but seeing one of those old enterprise beasts running ut2004 would make my year.
pls talk about direct die cooling with large air coolers next
i would recommend you to go with a 280mm aio instead of using the high end air coolers for the best temperatures with direct die cooling. I initially tried cooling my 9900k using the noctua nh-d15 and it is possible but its not worth it to me. Right now i am running the 9900k direct die with liquid metal using the corsair h115i and the max load temperatures i am getting right now are 8-9c better than with the noctua nh-d15. Now i am thinking or reducing the die height to 0.4mm using this tool, but sadly derbaur did not included the before and after results.
@@bloomfield295 9900k has a TPD of 95w as stated by Intel, but reviews have popped up that 95w cooling is not enough. NH-D15 is rated for 165w. In theory, that's enough thermal dissipation for a good overclock, but in practice, it really is not.
While I understand that CLC will bring more thermal headroom for overclocks, the usage of air coolers is still a good option for people that do not want to deal with CLC failures.
Anyways, can you share your current temps with Δ's? Would appreciate it.
@@notyouraveragegamer130 I doubt many people who are unwilling to deal with the dangers of water cooling and delidding their CPUs.
@@notyouraveragegamer130 95w is on stock frequencies and without turbo boost (or properly working turbo boost, one that doesn't allow you to boost indefinitely).
@@notyouraveragegamer130 Will share my results soon but as far as running a 9900k on stock clocks, a noctua nh-d15 is more than good. The thing is if some one is looking for running a cpu direct die then surely he is looking for the best thermal performance and to achieve that a 280mm 'aio' or a custom loop is needed, as i already mentioned, noctua nh-d15 will easily run 8-9c hotter than a 280mm aio and i am absolutely certain about that. Thats why i suggested against air cooling. But yeah air cooling has its own pros thats for sure.
Man did seasonic rip off winamp with that intro or what?! lol
All they did was whip the Llama's ass.
it really kicks the lamas ass
Ah, good old times.
Haha it really sounds like the winamp intro xD
@@MrMacroVision whips*
No matter how good this product is you'll never be able to buy it. I've been waiting for the Ncore CPU block from these guys for well over a year, never any updates, never in stock, always pre-order, no date. Never replies to any emails.
Hats off to this guy for making something unique and impressive but its pointless if you cant buy it, the guy is a hobbyist, not a business man.
Also a backer on the original Kickstarter for the Ncore, years later still noproduct in hand while he goes on to use kickstarter funds to buy CNC equipment and start making new products...
just use a lathe
@@monad_tcp A lathe? To make the tool or to lap the CPU? Can't imagine how you'd use a lathe to do either
That tool is called micrometer. I used it to measure copperwires from 0,7mm to 0,01mm, when I worked in cable factory. Thanks for the video as allways.
If the cpu doesn't fit the tool, I guess they add a notch to catch the pcb, same as the packaging, no?
Hopefuly, Roman will feedback and there'll be improvements
Sounds like buyers should measure the IHS, then apply some tape like painters tape to the top of the IHS before inserting in the tool. But if you are measuring the tool there is not much point in using the Acrylic Jig anyway. Just measure on all 4 corners and use the "walk around the barrel" sanding technique.
tape isn't perfect in thickness. nothing really is unless its machined for that. and when your talking about super precise measurements ,you want to have something guiding you. with the die itself, one wrong move and you will destroy it. and with the ihs. if it is a little off , you will have uneven mounting pressure against the die. this will cause it to crack when you to crack when you place the cooler. die lapping is risky business. way more risky then ihs top lapping. but i guess its the price you pay to have the best overclocks lol.using the tool stops you from rolling the edge when you sand. you have to take a ton of acrylic off before you would make an uneven sand. and you would see it happening buy the paint being removed more on one side .the tool isnt precise with its measurements. but it does make it way harder to mess up
I've lapped mine imperfectly, and it's still holding up since I got the 9900k around launch.
Hey Derbaur, i would request you to share those die sanding results. If possible please sand down that 9900k to as low as 0.4mm and mount it direct die and please share the before and after results.
In 2020 ...45 minutes = Forever
I could TOTALLY whip out a 3D printable design for these tools...one that would allow the "gauges" to be replaced after use because this acrylic unit looks like you must throw it away after lapping?
Sound like you need hand lapping plates and an optical flat to take this to next level.
And a lab grade surface plate and an octoculimator to check the final result... No, seriously, calling it "CPU lapping" is flat out wrong. It is sanding it down. If we were talking lapping in the machinist sense, he would get better results with no liquid metal, as the surfaces would wring together like gauge blocks.
put a plate with a pin to fasten the drill to it, machine a flat plate with a pin sticking out to attach the drill and glue it to the tool then just go, either the drill press or hand drill, drill press would be the best. maybe not for the cpu though :p
I always imagined die lapping was done with an angle grinder
The LED panels have an interesting.. pattern 😅
o no
ure just seeing things in everything :D
Haaaa
Be interesting to see if it actually left the die flat, with the difference in material while sanding it should be easy to sand more off the tool with uneven pressure, maybe time to get a test indicator and testing the flatness of the die. I also think not lapping the top of the IHS flat seems like a waste would also negate the scratching,
I would advise you to look up the 'walk around the barrel technique' for making a plane surface
I always use Break-Cleaner because it doesn't leave any residue.
They also use that to see if (motor engine) valves are completely tight after polishing them.
Proper acetone doesn't leave any residue either. The stuff I buy at the local super market says "chemically pure acetone" on the bottle and it flashes off as cleanly as alcohol.
Nice Video Roman, Thank you for the advice.
@der8auer I have one question which could be good video material.
*Is there any actual efficiency difference between full copper water radiators compared to aluminium ones?* And are copper ones actually better?
We all know, that copper has higher thermal conductivity, but from what I've researched, this alone is meaningless in radiator performance.
Copper has *3.3* times the density of aluminium, while having only *1.75* times the thermal conductivity. That means that you need *1.7-1.85* times as much copper in weight to achieve similar overall thermal conductivity, which seems *not to be the case.*
Not only are aluminium radiators heavier then necessary for theoretically identical results, but they're also *typically denser,* with more tubes and therefore higher overall area for cooling. The fins also seem to be thicker to offset lower conductivity.
I've also found some topics on automotive forums, where copper radiators are not recommended for their inferior performance.
Would you try checking that?
In theory, yes it also has been tested and proven although don't expect huge changes. The biggest impact comes from the cpu block which is now copper/nickel not from the radiators . And is not in a aluminium loop setup. Radiators like you mention come in al sizes prices config so if you have enough "radiator(s)" you will cool the loop. And i want to point that out because you see claims from 10 years ago saying a big 280, 360 is more than enough. i Will always have way more radiator than necessary
6:00 yes, always move it just left right/ use rotating lapping machine
@der8auer do you have NCore V1 waterblock for test or can u test it with 8700k and 9900k ?
I wonder if you could get same results on a gpu die?
Wow lapping the die itself?! I thought I misunderstood when I first heard it on the video… o.O
For a 2 degree difference that's just crazy but it's nice to have someone with enough money to do those crazy things we common folks can't afford to risk it lol
I wonder what the difference would be between lapping the CPU die and a silver heat spreader. Both idea are both equally crazy so it would be interesting to see if they have equal performance.
Thank You for that Shirt. Now if you will excuse me I have to get back to catching them all.
When are you going to make a AMD AM4 IHS lapping tool? I had an idea of using a broken AM4 board, just remove the entire socket and mounting mechanism, then remount it to a easy to handle block of acrylic
manufacturer used blue tac to hold in place, accurate measurement device = digital micrometer. could we just use a aftermarket copper ihs from an 8700K and skip the lapping?
What about keeping the IHS and die parallel? Sure the silicon is now thinner, but if the die and IHS are not parallel wouldn't that create a gap between the two leading to a potential decrease in thermal conductivity? The thicknesses matter, of course. For instance if you sand down 0.2 mm of silicon and now you have a 0.1 mm gap filled with Conductonaut on one side of the die overall there would probably be an improvement to thermal conductivity because silicon is roughly twice as thermally conductive as liquid metal and that gap would only exist on one side of the interface between the die and IHS.
Thermal conductivities: Silicon 149 W/mK, Copper 400 W/mK, Thermal Girzzly Conductonaut 73 W/mK.
Seems like if the two were not parallel there could be a decrease in thermal conductivity if using a TIM like Conductonaut. When sanding down the IHS tabs is enough material removed to cause interference between the die and the IHS before the IHS touches the interposer? This would allow the inside of the IHS to come in contact with the die and self align to it before the tabs touch the interposer.
Sorry for all the questions, but I am new at this. A surface can be very flat and very smooth, but when you bring two flat, smooth surfaces together they need to be parallel to eliminate gaps. When you lap a CPU cooler this isn't an issue because it is allowed to self align to the top of the IHS because of how the cooler is mounted. For lapping the die, there could be an error stackup in parallelism between the die, the interposes, the tabs of the IHS and the inside of the IHS.
Overall I like the video and appreciate your contribution to the technical content on TH-cam. Thanks.
I just watched the video from Oct 19, 2018 about sTIM on the 9900K. Moving from Indium solder in the stock thickness to Conductonaut made a big difference (8 °C). I can also see that the IHS was sanded down in this step to account for the reduction in overall height from removing the indium. Sanding the die and IHS from there provided about 1.25 to 1.5 °C per 0.05 mm.
While I still have the question about maintaining a parallel interface between the die and IHS, I can see that reducing the thickness of the indium and silicon do make a big difference. I suspect the IHS tabs, or "feet" have been sanded down sufficiently to not make contact with the interposer before the IHS contacts the die.
Just an idea - naked die water cooling? Submerging the die in liquid?
Direct die cooling is too dangerous. It can break so easy. Die is glass and glass breaks
I wonder if they intended for the tool to be used to lap the top of the IHS for some reason, it'd probably fit in there better the other way around.
lapping the top would not bring the IHS down to make contact with the die. i do see what you are saying tho, it does look like it would fit better that way.
Isn't laping the top of the IHS always the better solution? For the best cooling contact at least? Can you use this tool for that purpose? Or only inner side lapping?
@der8auer :
8:35 The micrometer used has a 0.001mm resolution. I will the NLap CPU tool to lap my delidded i9-9900K. Is sufficient to use instead a caliper with 0.01mm resolution to track the amount of silicon effectively removed?
Should I plan to sand the top of my der8auer direct die Intel 9th Gen. OC-Frame (anodised aluminium) by the same amount?
If so nail varnish will be enough to protect it against corrosion from liqui metal contact?
Thank you all for inputs.
I wonder if it's possible to diy a fresh diffusion barrier after lapping. It's silicon nitride, right?
You need to get a small but extremely precise cnc milling machine. You could have milled the die and ihs and finish it off with lapping for a smooth surface finish. I would be interested to see you take this concept to the extreme. Find some silicone lottery, shave it down as low as you can, and then do the same thing on a 2080 ti or titan. You can even use the machine to make your own custom parts for builds. Custom blocks, distroplates, case parts, etc.
This product is genius. Like how many times do you lap the same CPU... lol
Gotta give Intel some credit for that stunt they pulled going from Sandy Bridge to Ivy Bridge. Nobody in their right mind would delid, use liquid metal, lap an IHS or do direct-die 5-10 years ago. Content like this probably wouldn't exist.
On my 9900KF I lapped the top of the IHS because it was quite convex, and the temps would jump around and spike very quickly and the CPU would thermal throttle under heavy load, my temps have dropped by over 10c, at idle and around 20c under heavy load, I'm not using liquid metal.
I only use normal thermal paste (it is a Cooler Master brand paste) and I have not de-lidded the CPU.
I was considering de-lidding it and getting a direct die cooling kit but I'm not too confident in doing it to such an expensive CPU.
But considering it now runs stable at 5.1GHz all cores with a max temp of 80c.
I have been satisfied with the outcome in regards to temperature.
So my question is; is it ok to lap the top of the IHS as I have done and not using liquid metal or should I be aware of a potential future issue caused by doing this?
Wonder if you've tried the lapped 9900K with your direct die frame? Curious to know just how low you can go with 9900K temps.
LOL Awesome shirt bro
That quite precise tool is called a digital screw gauge..........😊😊😊
@der8auer, I have delided and used a direct die cooling with your OC frame in my Eurocome Sky X7c Clevo chassis laptop. I used a 3mm vaccum chamber thick and it increased my z-height by up to 0.4mm. I am planning on lapping the CPU to improve my contacts and cooling system tolerance. How much of the i9-9900k die could I lap to stay above OC frame? Should I plan to lap the bottom of the OC frame to offset it? Anyone has an idea of the thichkness of the STIM solder or the height pcb to the top of the IHS before delidding the i9-9900K (lost measurement after delidding)?
Man I miss the old days of AMD Socket A with direct die cooling.
oh hell no, cracked so many die's.
accurate measurement devices are good to have
I don't believe lapping a cpu is worth it, lapping the heat sink on both cpu and cooler is worth it though.
Is there any point in lapping a cpu?
Bye bye warranty. Not that it covers overlocking thou
I need zen 4 to run on x470.
You mean Zen 3.
Where did u get that T-shirt? :D
www.qwertee.com/product/surprised-pikachu (He Postet this in the german vid)
www.qwertee.com/product/surprised-pikachu
Can u make vid where u try overclock stuff with a PSU that only has a +Plus -ceritivication or lower :P and compare it mayeb with some higher end ones just for funzies, might be some unexpected results :P
Nice video
use the die tool to make the die smaller then put on the Ncore directly?
only way I would ever lap a cpu would be by hiring derBauer to do it for me...
Quick question: Is there any benefit in lapping the IHS a R5 3600 on a 280mm CLC? The temps currently are quite good, but just curious.
Not really, don't expect more than 2 degree less. If you sand both the cpu and the coldplate maybe something more than 2 degrees but for all the work isn't really worth it
Not worth it.
There might be a slight benefit, sure, but there's really not any reason to do so.
No, there isn't much overclocking performance on 3600 and isn't worth it.
nice
What about laser lapping
Ehh this have so many flaws... What about flatness of the acrylic "thing", what about of CPU PCB flatness? what about glass panel flatness, what about finger pressure - all those are imperfect and will affect final result ...
You don't even know how hard is to be perfectionist. Nice job as always
Those of us with machinist training think of lapping as something very different than what overclockers think. They say lapping but they mean sanding. Period.
If you are interested in how to make a proper lap (that could be used for a CPU if you like) go check oxtoolco here on youtube. He goes through the classic three lap method using only copper roundstock, a hack saw and 5my alumina abrasive.
Sad face for not having a link to the Pikachu shirt =( jk
that looks very handy tool, shame it doesnt work as intended
oh man, you're already lapping the direct die, might as well lap the top of the IHS.
you would have to make a new nickel coating then to prevent the bare copper on the IHS to react with the indium alloy from the liquid metal on top of it.
Hey roman can you make a video about intel i9 9900ks cpu can we delid that cpu or not plz make a video about it i want make new pc by using that cpu
Should be exactly the same process as with a 9900k and I'm pretty sure he made a video on that
9900ks = 9900k with some binning and a "special edition" on the box, nothing more.
But you shouldn't buy this crap, if you really have money to burn, wait for 10th gen, they'll be here soon.
I usually lap my CPU in 58 second
I have not lapped a CPU in years.
pity is non of it works with a Haswell
@der8auer All these years and still no one has made a motor device for lapping. disappointing.
It's not that impossible to design and make a device!
Simple concept, have a rotating circle surface with perfect mirror finish.
Make it rotatable with a small cheap motor, make a arm that pushes the CPU to the circle surface where you have the sandpaper.
Kingpin has a whole setup.But i don't see me building buying something like that for that 1 CPU a year ;)
nice shirt
It's not accurate, but it is precise
Da funktionieren zwei Lampen nicht oh nein
All that for 2 or 3 degree's??? Not worth all that time to me.
it is for extreme overclockers
If you are going for overclocking to the extreme levels, then 2 to 3 degrees is a lot. Look at this way, most people who do overclocking are colouring book and crayons. This is level of overclocking where they are painting something like the Mona Lisa quality.
Two to three degrees can help you increase your imc voltage by quite a bit.
Dat price...samething when i saw dat ekwb price with shipping for dat 011xl d5 distro plate..
Tool you are using is a digital Micrometer.
All that for just 2-3°? Oof
Pika pika
I thought this video was about lap dancing.
Never mind.
нахер его вообще тереть то ? че с ним не так ?
понимаю еще чтобы без термопасты собрать
но это не про это
For the IHS should be a magnet placed inside the acrylic to hold IHS from the top, for the CPU should be like a delidding tool, with screw(s) from the (both) side(s) to hold it
IHS is copper and copper is not magnetic...
@@honzamoravka9430 true, but I used neodyme magnets from hdd and put second one below the ihs, I found some are just thin enough to fit and not interfere when lapping
And result of all you're work is a 9900k at +100mghz OC all cores constantly at 82-90c air and 75-79 under a 240 aio at 300 wats from the wall ... pls tell me it ain't true
Milligram hertz
I got i boot like .... im so happy omfg yeye
This is no different than people tuning cars to the edge. You can get a fair performance boost for a little money and a little effort. After that it becomes more and more expensive and more and more labor intensive.... Chasing the last 1% can quickly cost $100k and take months of work.
When you have a tool that measures 0.001 millimeters, and you just call it "quite accurate", you know you're in Germany :D
You mean a micrometer
it believe the "accurate measurement device" is called "MICROMETER"? Yes? Why not call it by its name? Afraid Steves gonna come memeing? XD (they just have some meme video about nano and micrometers or something? XD)
Ich mag katzen
no sheik !!!!!!!! un subbed lol
I have had a 9900K lapped and direct-die cooled with liquid metal onto an EK block since shortly after launch. 5.4 with silly voltages, 5.2 daily. No ill effects yet, but I haven't touched the block since installing so I don't know what it looks like
Nice, do u game with it?
I guess having a 9900K that only requires 1.290v to run at 5.2GHz, only reaching 90°c after a 20 loop run in R20 with an AIR cooler defeats the purpose of this tool. I also lucked out on the 1.280v 5.3GHz 8700K that i got open box, also topping out at 90°c on AIR. What's with all the water cooling...lol
Nothing against AIO's. I've been through many and the only ones that actually cooled better than some of the best AIR coolers was Corsair's V115 and V80 Extreme V2's with the double thick radiators and some vardars.
Though, with that amount of money involved in AIO's and some worthy fans, you gotta be an enthusiast. I just like testing old and massive AIR coolers.
Anyone remember the first 8 heatpipe Silver Arrow monster, or how about the ZALMAN 9900MAX, 9900MAX DF or Triple fan 12X. Might sound funny to some but, beats paying $100+ for something that essentially puts up the same numbers as a 5 to 10 year old AIR cooler. That can be found for $25 to $40. Gotta have a grade a chip though.
How to create microplastics on a larger scale. Consider where the acrylic goes if you use water .
nowhere because this amount of water dries after 20 minutes
@@der8auer Hello man, thanks for your answer. Thank you for your cool videos as well!
And if the only source of plastic residues in the world was from people like Der8auer lapping a cpu, let's assume that the problem would be really, really not a big deal
Stop driving then, thats the real difference :P Anyway fun video, but lapping on cooler prolly enough.
That’s a shit lot of work . Hehe
lol get a normal micrometer . digital is too easy .
U know u can make a better tool 😉
NICE
I'm first xD
Wieso nur in Englisch?
th-cam.com/video/iscSUL1hWx0/w-d-xo.html
DE
German version Pls ;-(
th-cam.com/video/iscSUL1hWx0/w-d-xo.html
Your accent makes it hard to understand you, can tou talk with a better accent so we can understand you?
He's perfectly understandable, and his accent is pretty good for a German, better than a lot of native English speakers even. Maybe you should listen with better knowledge of the English language?
Go watch the german version if you find his english hard to understand. You should be grateful he takes the time to even do an english vid.
Yes ,Roman just be born somewhere else to switch to a "better accent"....(fucking hell people are stupid)
Can you improve your English spelling and grammar?