You really didn't need to do this! It sparked a nice in detail discussion and proved people are really eager for your in-depth videos. Thanks for your commitment and desire to go into details, but... you already did things perfectly (enough) round 1!
You really should make more vids discussing thermal conductivity, diffusivity, air temperatures, thermal camera stuff, COPPER SHIMS (huge personal favorite) etc, its incredibly interesting, really liked this video
@@pcoverthinkI suspect copper shims will become more popular in the future esp on rtx 4000 rdna3 and beyond, 8$ for 20 huge thick shims is an incredible value for anywhere between 10-50 (yes 50)c drop on gpu temp, amazing, only need like 10 shims on a gpu
Copper shim expansion under heat load and the potential for damage if the tolerances are too tight, would really be a fun video. For example, when used as replacements for the soft, pliable thermal pads on GPU memory is becoming fairly common, and examining the risks of damage from both the copper expansion as well as the memory module case material's own thermal expansion would be great to see, since if you've been doing this long enough, you know of the disasters this has resulted in for many people over the years as this breaks GPU memory module solder joints or the ICs themselves.
Incredible attention to detail, as always! I love these technical deep dives, breaking down what people assume they understand to the literal formula. Never stop, the industry needs people like you and Steve to keep the manufacturers (& community) honest!
Honesty is a manufacturers kryptonite, like Nvidia won't tell you if their connector catches fire and AMD won't tell you you actually need better cooling to offset the higher temperatures caused by their own design "to please everyone" Anyway, Keep it up Roman (& crew) I know it's not easy to do all this (and you're doing everything twice once in English and once in German others might just add subtitles instead of a reshoot)
I worked with one of these equipments at college and some alluminum alloys. Really interesting how a few percent chemical composition can change the conductivity. Material science is great!
I've got an MSc in Materials Science & Engineering, and I've learned something from this! (Or, at one point I did learn this at university, but I forgot...) One remark on the gold plating for pins and connectors: gold is relatively soft, so gold plated pins tend to offer less frictional resistance as well. That's not very important for pinned or LGA'ed CPUs, but it _is_ beneficial for connectors that make sliding contact, so USB, PCI-e, HDMI, DP, etc.
I'm about to do my first custom loop after about 15 years of building and being an enthusiast. Having to now take into account factors such the relationship between fluids (water, biocide, coolant) and metals (nickel, copper, aluminum), the flow of the pump in proportion to the dissipation of heat, and many other factors, is genuinely interesting and I can't wait to get it together
90% of the coolers will be nickel plated and not real silver. pure silver would work about 5 % better(as its about 5% more conductive electrically) , but cost about 10 -20 times what a regular 40 $ cooler would .
You are better off with just just copper or nickel plated copper anyway. I wont write a dissertation here about it but you want your loop to have like metals all the way though due to the galvanic effect. Silver doesn't play well with copper or pure nickel (nickel chrome alloys are better with silver). As most radiators are copper its best to just stick with the copper and use a biocide in your loop. Any benefit from the silver in the biocide department are simply not worth the cost.
I was wondering how we were going to convert electrical conductivity to thermal conductivity. The Wiedemann-Franz-Law was really neat to learn about . I'm still waiting for super materials to be used in cooling . Some graphene-copper alloy or synthetic diamond - graphene with some serious thermal and electrical conductivity. Thermal paste really makes a difference in all of these materials as well. The gold plating note was excellent . It's to prevent the corrosion as gold is supposed to be inert even to water. Most other materials will rust or breakdown but gold seems to hold up much longer.
Oh, sure. That's exactly what we need right now, exotic materials to make PC components just impossible to buy for the 99% of the people. Thanks for that.
@@tobiwonkanogy2975 I really appreciate that. But I was dead serious about it, exotic materials like the silver used on the Aqua computer example are just a perfect excuse to raise prices up for no gain or negligible boost on performance. I'm pretty sure those exotic materials are being used by NASA for purposes that makes more sense than playing with our gaming pc.
@@JP-xd6fm Marketing would definitely push its advantage even if it was within a range of error . my real want is for copper-graphene. wouldn't take much to figure out how make a batch . potentially just run simulations with known properties or might have make several batches until the right consistency is found . I've seen old space computers but nothing newer . the military and aerospace technology always trickles down to the common folk eventually . microwaves , flight, pc's etc etc all was mythological at one time.
It's really cool seeing how dramatically purity effects performance. I wonder if CPU playing is so thick to protect against wear. A really thin coating would be more efficient, but a lot easier to accidentally wear through.
Thank you for the detailed explanation and the 'short version' to, this went well beyond my field of knowledge, but I was actually very to know curious to know about the pro's and con's of copper vs silver thermal conductivity but didn't even know where to start. Nice video Der Bauer!
I like how systematic and logical your approach to the topic is. This is a top quality channel. On a side note - I wish this guy was my physics teacher. It would have made a lot more topics approachable to me.
Roman, in the thermal conductivity over temperature graph you showed early in the video it was noted that the thermal conductivity of alloys can INCREASE with temperature, unlike pure metals that DECREASE with temperature. Would it be worth testing both of the Aquacomputer IHS's with higher temperature? Is it possible, at say, 70C that the thermal conductivity of the sterling silver IHS increases enough to overtake that of the pure copper IHS?
Both alloys involved (copper and silver) have the same temperature/conductivity relationship. Bar some mad interaction between the two when alloyed, the trend will be the same. The point he was making was that by alloying brass in to an alloy, then you can benefit from the increasing conductivity trend brass has, as well as the raw conductivity of the (typically) copper..
Nope, the sterling and pure silver will outperform copper, Only thing better is diamond. The water block needs corrosion resistance so silver oxide doesn’t form on the waterside and insulate. Dissolving silver would increase thermal capacity but it would just clog things up. And any oxidized area traps heat
5:40 Looking at the graph, one important think about copper thermal connectivity is that it crashes by ~10% before reaching 100 deg C. You are doing the measurements at room temperature, but I think you should also test at ~70 to ~80 deg C (if the head of the measurement instruments permits that, I don't want to kill it).
I think you are misinterpreting the chart. From 200 Kelvin (-73°C) to 400 Kelvin (127°C) the conductivity drops by about 15 W/m*K which is not even 5%. So From the room temperature at which I measured to about 70 or 80°C it would change by about 1-2%. Not worth measuring
I wonder if given unlimited budget a diamond block would make much of a difference. Obviously it would cost a fortune to manufacture, but would be cool.
it was really interesting to see the rockitcool ihs tested here. its amazing how some LM and a slightly different profile IHS can drop temperatures more then 10c at max load. I agree that it should also be nickel plated.
So glad you pointed out that silver water coolers are worse, I've only bought copper water blocks for decades now because someone tested this years ago and said the same thing. Glad I have followed that advice and it helps when buying because they are cheaper. Thanks for this content ❤
thank you for letting people know the truth about gold plating! it's very important on the fingers of plugs that are frequently removed/inserted because right before the pins touch each other there will be a small electrical arc which would be damaging if not for the gold!
very informative.... so that slightly thicker Nickel plating on the AMD will work well with a ThermalGrizzly Kryonaut Liquid Metal Paste in protecting the IHS... and you gave me a good part to look into for a cpu waterblock....
Fantastic tool, I want one. Maybe I can persuade my professor to buy us one. XD These thermal and electrical conductivity problems are very common my field of research (applied superconductivity). The Wiedemann-Franz law especially is a sonofab*ch, when making current leads for experiments in liquid helium.
Hey Roman, what is the interaction between silver and gallium based liquid metal? Is silver immune to the gallium? If so I think that would be a big upsell. As you showed the nickel plating on the copper greatly affects the performance, so if the silver is safe for permanent use unlike the copper then the value might be there.
@@der8auer-en Gallium is corrosive to all metals except tungsten and tantalum, copper is no more resistance than silver, that said when we talking about a heat spreader on a CPU there is no real world difference. there is less then 1% difference in the thermal conductivity of materials on heat spreader that size at the working temperatures(not counting Effusive/Diffusivity) . So use copper its the best at transferring the heat at the price, unless we can make the heat spreader out of Graphite or Diamond.
Excellent video as always. This is something I have been struggling to wrap my head around, and wish I knew more about when creating my GPU copper modding videos. All I knew was adding copper greatly reduced the temperature of components, but the in-depth reason for WHY has been cloudy in my head. Love all of your videos and Thermal Grizzly products, you are truly the king of PC hardware engineering!
nice video and thorough theory behind the measurement. and still there is people who svallow a salesman salespoints than this facts. bear in mind that salesmen often don’t have other education than salesschool.
Roman, can you take another step and solve thermal diffusion equation? I'm asking because you are missing two variables in your comparisons: thermal capacity and density. These enter into the equation as well.
I would love to see this video with liquid metal vs stock cpu. Thats the only thing i missed in this video. I got about 20 C° difference with my old 4770k so it would be interesting to see your numbers
Thanks for the video and information Roman🙂 I really liked this video. Most of my subscriptions are either PC tech or physics channels, and this video was applied physics. Side note: I got into physics videos originally to learn more about Quantum tunnelling, like when a CPU gets hot, and what happens, also about how cosmic rays can either cause a pc to crash or the very least corrupt data.
If the silver goes all the way through and is used as the interface with the water in the loop - I understand why there is a silver version - silver is antibacterial, so the silver block will inhibit algae growth in the cooler liquid, and algae will clog up the microstructures inside the block; so although it might have slightly lower cooling capacity when new, it will maintain that over a much longer time than the pure copper/nickel plated copper.
ive been oC'ing anything from athlon xp (which i burn't) from q6600, didnty felt assleep, loved the cat and loved the video. keep making good footages please ;)
Roman, silver is used to destroy bacteria and something like that in a liquid system, that's why some people use it, but also didn't expect that has worse thermal conductivity than copper 🤯
The upshot I get is there is not a whole lot of point to lapping+liquid metal; other than a boat load of trouble down the road as everything corrodes. Plus points for an interesting physics lesson! 👍
As far as I know there is another reason for gold or silver plating: lower contact resistance. It's different than the electrical conductivity inside of the material. That's of course important for connectors, however digital transmission like with USB or HDMI cables does not see real gains from that.
Would it be possible to make an alloy of copper and silver that has better thermal conductivity than either element in their pure forms, especially at higher temperatures? Since you say that thermal conductivity generally decreases with temperature for pure elements, but rises for alloys.
You would think that what would be spoken about is resistance vs conductivity. Same thing I realize, but thermistors seem easier to talk about and resistance and temp track positively in pure metals.
Silver is too soft, that's why they use sterling. Its other components make it less soft than 999. It uses the silver because of biological growth in the water and cools decent enough. Maybe it would have been more idea to use a copper base and a silver lid, but that would be much more expensive.
Learned a lot from this. I had always thought gold was better, because NFS mentioned that the McLaren F1 had gold heatshields. Now I wonder why that was.
Good content! 👍 Much better than when you play the basic utube influencer! Which is a shame when being the CEO of a company. By the way I switched to gelid GC extreme, the price of the kryonaut and kryonaut extreme is too damn high here in France!!!!
Red copper or yellow is already a big difference, and then there is the purity to get the composition is the main role, and it should not bend, that is the 2nd main roll, Cast iron was popular before modern insulation because it took a long time to heat up and cool down. If the Victorian look is what you're after, stainless steel versions of old cast iron designs are available.
Hey Roman, I have the silver Aquacomputer block in the adjustable version with the 4 pins that can warp the profile of the block while mounted because the silver alloy they use is still softer than copper, so it makes it much less work to make it conform to unlapped blocks. Not that I need it anymore since I swapped to a Der8auer Edition 8086k - lapped copper version, not the fancy silver IHS one - but I also have no reason to get a new block, because as you said the temperature difference is only half a degree purely based on materials.
Wow really great vid I'm running one of the silver blocks I have both when buying I did question the purity why only 925 right? Now the silver has made my loop a mess too. Really love let's check the science videos. Off to clean my loop.
Silver is a better antimicrob then copper is. When your running straight distilled, with a silver block, your coolant over time would leach the silver ions and prevent nasty stuff from growing inside your system. The performance difference was very small, but it was more on not having to use additional additives in your coolant. Also Sterling Silver should not be used over pure Silver. Sterling Silver can cause corrosion which turned Silver products into Taboo for the hobby especially with nickle plating. We also ran silver kill coils, however most silver kill coils that came after Iandh usually were tainted or fake with iron and other metals mixed into it, that it caused corrosion. A wide ban on silver was brought about especially from eK, because of a few bad players who wanted to profit off fake kill coils. Iandh and I btw were the first ones to experiment with silver kill coils on XtremeSystem.
All those times you said, "You can see." I could not see. What is the conductivity of gold? Why was I taught in school that gold is the best conductor?
LM on copper just creates an alloy layer really and I have never needed to lap it to add more LM to it. LM on silver turns into rock hard solid bumps which are a nightmare to lap.
Basically, "Silver Plated" or even "Gold Plated" are just for marketing gimmick to make the product more "Premium" or "Luxury" but it perform worse than pure copper water block.
I didn't expect these results from the silver not being pure. As far as thermal conductivity goes, Silver is about 8-9% more thermally conductive than copper. Which is why silver is used for thermal compounds over copper. I mean, theoretically a silver loop is better than a copper loop. (if price is not a consideration) But I haven't seen pure silver components for a long time (because they didn't sell well due to price in the late 90's early 2000's) ESP haven't seen a silver radiator in a long time... which means you will get electrostatic migration when mixing metals, which is what I was assuming this video was going to talk about coming in. That migration is a pretty big problem, and is why I used to tell people to pick a metal and make the whole loop out of it. (Given, hardly anyone used to ask about a custom loop). Really strange they would ruin their own silver heatsink by using cheap impure silver.
There is an error at the start, around the 2:00 mark, where you state that multiplying the interface's Thermal Conductivity by the Temperature Difference across it and then dividing by the Thickness (through which the conduction takes place), results in Watts of Heat being transferred through the interface. This is incorrect: λ * ΔT / L has dimentions of (W/(m*K)) * (K) / (m) = W/(m^2); i.e. one obtains the Heat *Flux* (Watts per surface area). Using the given example's dimentions of 100m * 100m = 10,000 (m^2) for the area of each of the large faces, the correct Rate of Heat Flow can be calculated by multiplying the obtained Heat Flux for the 2 material cases (first 400 W/(m^2) and then 200 W/(m^2)) by the total face area (10,000 (m^2)), giving 4MW and 2MW respectively: 400 W/(m^2) * 10,000 (m^2) = 4e+6 Watts, and 200 W/(m^2) * 10,000 (m^2) = 2e+6 Watts. As for the emphasis on steady-state thermal conductivity, I think for modern CPUs, especially when overclocking and using water cooling, such a simple steady state analysis obscures some very significant transient differences between materials' thermal diffusivities. Even at stock, modern CPUs and GPUs already make use of transient thermal headroom by alternating between boosting over the steady state TDP limit and dropping far below it, many times faster than the cooling system's characteristic thermal time constant. (Basically, the waterblock *doesn't* behave like a steady-state thermal conductor; but rather 'absorbs' short bursts of transient heating, followed by periods of below-average heat input, during which thermal diffusion and ultimately conduction into the cooling water, cools the block back down.) Therefore, I would very much like to see what difference a heat spreader (or waterblock) made of *pure* silver would make, seeing as Silver's Thermal Diffusivity is 49% higher than that of Copper!
Is there a differnce in conductivity between: Material to Material or Air to material ? Air|paste|Copper[------]Copper|TIM|chip I'm guessing the thermal conductivity of copper is talking about the copper itsself?
The reason why that product is explained in 5:58. Your tests might be the issue. You should have backed it up with real tests, and with different frequency settings and at say +100°C as a second shot. If all those result are the same as your initial result. That only means they have poor quality silver.
So the best we'd want is the backside that touches the chip(s) to be copper, & water side plated nickel? Guess people could sand off the nickel of the back sides of their water blocks, this would be great to shown what difference this would make.
With this tool is it possible to do a COMPREHENSIVE test of the most popular thermal compounds and then compare the results to their "advertised" numbers? Would be interesting to see firstly if what most manufacturers advertise is true and then secondly based on these now accurate results, which thermal compound is REALLY the best.
tldw; You can afford a much more powerful and higher quality cooler for less than screwing around with silver and copper corrodes less. Ancillary info: One of the problems that AMD is facing with their multi-chip components is that the different chips will have different thermal ranges to a certain engineering factor... this means that on occasion there may be intentional things like that thicker nickel plating on their packages. This is actually fairly common in older compact systems (power supplies, electrical management, etc.. including old heatsinked transmitter components) with heatsinks where some heatsinks have an increased thermal mass to them to better regulate to the operating temperature lifespan of given components and others may contain segments of copper or be made largely of copper instead of aluminum. In computer components the metal type doesn't matter so much as it does in high frequency equipment where the heatsink may literally be Tin (Sn). Thinner spreader+cooler+faster is NOT best. Also, there are several university projects where they are attempting to _grow_ diamond heat spreaders onto chips. Its only nearly 5x more conductive than silver... It sounds like the kryos silver has a mild disruption of the plating interface to me... heat it up to 65°C and re-test it.
Isn't silver used in loops as a biocide? Seems that using the silver for the block would serve a similar function albeit for far more money than required/reasonable.
You mentioned the silver water block is 92.5% Silver and 7.5% "Probably Copper" . . . but what if it's an alloy, like brass, and it's thermal conductivity either increases with temperature, or, stays stable and does not decrease with temperature?
Re-Upload: Fixed the W unit mistake in the first part of the explanation so this should be solved. Thanks for the feedback
You really didn't need to do this! It sparked a nice in detail discussion and proved people are really eager for your in-depth videos. Thanks for your commitment and desire to go into details, but... you already did things perfectly (enough) round 1!
@@FreakyAngelus it was almost no effort and I agree that it should be following SI units
uh that's why I found it private and then reuploaded.. well looks like I saw the correct one then, thanks :)
I appreciate the dedication to accurate explanations :)
Fyi pure silver doesnt tarnish but its waay to soft
You really should make more vids discussing thermal conductivity, diffusivity, air temperatures, thermal camera stuff, COPPER SHIMS (huge personal favorite) etc, its incredibly interesting, really liked this video
@@pcoverthinkI suspect copper shims will become more popular in the future esp on rtx 4000 rdna3 and beyond, 8$ for 20 huge thick shims is an incredible value for anywhere between 10-50 (yes 50)c drop on gpu temp, amazing, only need like 10 shims on a gpu
@@pcoverthink just build a 1lb heatsink out of copper shims :kekbomb:
Copper shim expansion under heat load and the potential for damage if the tolerances are too tight, would really be a fun video. For example, when used as replacements for the soft, pliable thermal pads on GPU memory is becoming fairly common, and examining the risks of damage from both the copper expansion as well as the memory module case material's own thermal expansion would be great to see, since if you've been doing this long enough, you know of the disasters this has resulted in for many people over the years as this breaks GPU memory module solder joints or the ICs themselves.
@@bjn714 copper shim good
@@bjn714 PCoverthink's sandwich would seem a good compromise.
Incredible attention to detail, as always!
I love these technical deep dives, breaking down what people assume they understand to the literal formula.
Never stop, the industry needs people like you and Steve to keep the manufacturers (& community) honest!
Quick explanation: He is German. 😅
Honesty is a manufacturers kryptonite, like Nvidia won't tell you if their connector catches fire and AMD won't tell you you actually need better cooling to offset the higher temperatures caused by their own design "to please everyone"
Anyway, Keep it up Roman (& crew) I know it's not easy to do all this (and you're doing everything twice once in English and once in German others might just add subtitles instead of a reshoot)
I worked with one of these equipments at college and some alluminum alloys. Really interesting how a few percent chemical composition can change the conductivity. Material science is great!
I've got an MSc in Materials Science & Engineering, and I've learned something from this! (Or, at one point I did learn this at university, but I forgot...)
One remark on the gold plating for pins and connectors: gold is relatively soft, so gold plated pins tend to offer less frictional resistance as well. That's not very important for pinned or LGA'ed CPUs, but it _is_ beneficial for connectors that make sliding contact, so USB, PCI-e, HDMI, DP, etc.
hm
I'm about to do my first custom loop after about 15 years of building and being an enthusiast. Having to now take into account factors such the relationship between fluids (water, biocide, coolant) and metals (nickel, copper, aluminum), the flow of the pump in proportion to the dissipation of heat, and many other factors, is genuinely interesting and I can't wait to get it together
In theory, as silver is a natural biocide, it's possible the silver block was made to prevent growth. I've always wondered about a pure silver cooler.
90% of the coolers will be nickel plated and not real silver. pure silver would work about 5 % better(as its about 5% more conductive electrically) , but cost about 10 -20 times what a regular 40 $ cooler would .
You are better off with just just copper or nickel plated copper anyway. I wont write a dissertation here about it but you want your loop to have like metals all the way though due to the galvanic effect. Silver doesn't play well with copper or pure nickel (nickel chrome alloys are better with silver). As most radiators are copper its best to just stick with the copper and use a biocide in your loop. Any benefit from the silver in the biocide department are simply not worth the cost.
@@amindamok lol some rich mofo now contemplating building a loop with all solid silver parts. Jks but I'm sure someone will do it at some point.
Greetings from Brazil! descent of German here !! Great work! tks! for this masterpiece video .
I was wondering how we were going to convert electrical conductivity to thermal conductivity. The Wiedemann-Franz-Law was really neat to learn about . I'm still waiting for super materials to be used in cooling . Some graphene-copper alloy or synthetic diamond - graphene with some serious thermal and electrical conductivity. Thermal paste really makes a difference in all of these materials as well. The gold plating note was excellent . It's to prevent the corrosion as gold is supposed to be inert even to water. Most other materials will rust or breakdown but gold seems to hold up much longer.
Oh, sure. That's exactly what we need right now, exotic materials to make PC components just impossible to buy for the 99% of the people. Thanks for that.
@@JP-xd6fm you and J B are forever jocking my stuff. thank you for never stopping your typing crusade.
@@tobiwonkanogy2975 I really appreciate that. But I was dead serious about it, exotic materials like the silver used on the Aqua computer example are just a perfect excuse to raise prices up for no gain or negligible boost on performance.
I'm pretty sure those exotic materials are being used by NASA for purposes that makes more sense than playing with our gaming pc.
@@JP-xd6fm Marketing would definitely push its advantage even if it was within a range of error . my real want is for copper-graphene. wouldn't take much to figure out how make a batch . potentially just run simulations with known properties or might have make several batches until the right consistency is found . I've seen old space computers but nothing newer . the military and aerospace technology always trickles down to the common folk eventually . microwaves , flight, pc's etc etc all was mythological at one time.
It's really cool seeing how dramatically purity effects performance. I wonder if CPU playing is so thick to protect against wear. A really thin coating would be more efficient, but a lot easier to accidentally wear through.
Science over ignorance, myths and misinformation. Thats why I like this channel, keep it up man.
Thank you for the detailed explanation and the 'short version' to, this went well beyond my field of knowledge, but I was actually very to know curious to know about the pro's and con's of copper vs silver thermal conductivity but didn't even know where to start. Nice video Der Bauer!
I knew about lots of these things but not to the extent that was explained. As always great video. Very informative.
Great detail! Just had PTSD triggered when i heard "heat conductivity of copper" and the memories came flooding... MAN HEAT TRANSFER WAS A PAIN!
I like how systematic and logical your approach to the topic is. This is a top quality channel.
On a side note - I wish this guy was my physics teacher. It would have made a lot more topics approachable to me.
Roman, in the thermal conductivity over temperature graph you showed early in the video it was noted that the thermal conductivity of alloys can INCREASE with temperature, unlike pure metals that DECREASE with temperature. Would it be worth testing both of the Aquacomputer IHS's with higher temperature? Is it possible, at say, 70C that the thermal conductivity of the sterling silver IHS increases enough to overtake that of the pure copper IHS?
Both alloys involved (copper and silver) have the same temperature/conductivity relationship. Bar some mad interaction between the two when alloyed, the trend will be the same. The point he was making was that by alloying brass in to an alloy, then you can benefit from the increasing conductivity trend brass has, as well as the raw conductivity of the (typically) copper..
Nope, the sterling and pure silver will outperform copper,
Only thing better is diamond.
The water block needs corrosion resistance so silver oxide doesn’t form on the waterside and insulate. Dissolving silver would increase thermal capacity but it would just clog things up. And any oxidized area traps heat
5:40 Looking at the graph, one important think about copper thermal connectivity is that it crashes by ~10% before reaching 100 deg C. You are doing the measurements at room temperature, but I think you should also test at ~70 to ~80 deg C (if the head of the measurement instruments permits that, I don't want to kill it).
I think you are misinterpreting the chart. From 200 Kelvin (-73°C) to 400 Kelvin (127°C) the conductivity drops by about 15 W/m*K which is not even 5%. So From the room temperature at which I measured to about 70 or 80°C it would change by about 1-2%. Not worth measuring
@@der8auer-en I agree. Yet the engineer in me screams: I want to test it and see it ;)
I see why DerBauer name is so big in the community, this young man is engineer material and I see why people listen to him.
Good job and keep it up.
thank you!
I wonder if given unlimited budget a diamond block would make much of a difference. Obviously it would cost a fortune to manufacture, but would be cool.
I love when you go in depth behind the physics like this
it was really interesting to see the rockitcool ihs tested here. its amazing how some LM and a slightly different profile IHS can drop temperatures more then 10c at max load. I agree that it should also be nickel plated.
I just want to say you have a great proportionality Mr Der8auer.
So glad you pointed out that silver water coolers are worse, I've only bought copper water blocks for decades now because someone tested this years ago and said the same thing. Glad I have followed that advice and it helps when buying because they are cheaper.
Thanks for this content ❤
Sir you are the best Ifo vloger to educate people of needed Computer sciences thank you.
thank you for letting people know the truth about gold plating! it's very important on the fingers of plugs that are frequently removed/inserted because right before the pins touch each other there will be a small electrical arc which would be damaging if not for the gold!
You're all casual here, I'm using diamond IHS for good thermal conductivity. Copper and silver is so amateur 🥸😏
Love the detailed testing and physics background. Incredible work and science communicating.
Would love to see some actual benchmarks in a custom loop on those 2 "Aqua computer" heatsinks, copper vs silver.
Nice to see sheik gracing us with her presence, even if it was very breifly. thank you for giving the people what they really want..
I’m going to use this video in my resume for my next job. I’m now qualified to use this machine. 😁
very informative.... so that slightly thicker Nickel plating on the AMD will work well with a ThermalGrizzly Kryonaut Liquid Metal Paste in protecting the IHS... and you gave me a good part to look into for a cpu waterblock....
Fantastic tool, I want one. Maybe I can persuade my professor to buy us one. XD These thermal and electrical conductivity problems are very common my field of research (applied superconductivity). The Wiedemann-Franz law especially is a sonofab*ch, when making current leads for experiments in liquid helium.
Liquid helium is interesting indeed cause of the temps it can achieve and what it can do to metal at those temperatures.
Hey Roman, what is the interaction between silver and gallium based liquid metal? Is silver immune to the gallium? If so I think that would be a big upsell. As you showed the nickel plating on the copper greatly affects the performance, so if the silver is safe for permanent use unlike the copper then the value might be there.
From the limited testing we did with silver it seemed like it was actually worse than copper. But we only tried it on few IHS and limited time
@@der8auer-en Ah ok. So a nickel plated copper contact surface is still the best. Nice to know you guys already tested it. 👍
@@der8auer-en Gallium is corrosive to all metals except tungsten and tantalum, copper is no more resistance than silver, that said when we talking about a heat spreader on a CPU there is no real world difference. there is less then 1% difference in the thermal conductivity of materials on heat spreader that size at the working temperatures(not counting Effusive/Diffusivity) .
So use copper its the best at transferring the heat at the price, unless we can make the heat spreader out of Graphite or Diamond.
silver 92.5% is very prone to oxidation and gallium eat away the passivated coating
@@Thundra74 why would you use 925? that like using 92% copper both are very poor metals.
Learning PC Hardware also learning about metallurgy. Nice info as always 👍👍
That out take was rather amusing :D
Excellent video as always. This is something I have been struggling to wrap my head around, and wish I knew more about when creating my GPU copper modding videos. All I knew was adding copper greatly reduced the temperature of components, but the in-depth reason for WHY has been cloudy in my head. Love all of your videos and Thermal Grizzly products, you are truly the king of PC hardware engineering!
nice video and thorough theory behind the measurement.
and still there is people who svallow a salesman salespoints than this facts. bear in mind that salesmen often don’t have other education than salesschool.
Roman, can you take another step and solve thermal diffusion equation? I'm asking because you are missing two variables in your comparisons: thermal capacity and density. These enter into the equation as well.
I would love to see this video with liquid metal vs stock cpu. Thats the only thing i missed in this video.
I got about 20 C° difference with my old 4770k so it would be interesting to see your numbers
Thanks for the video and information Roman🙂
I really liked this video.
Most of my subscriptions are either PC tech or physics channels, and this video was applied physics.
Side note: I got into physics videos originally to learn more about Quantum tunnelling, like when a CPU gets hot, and what happens, also about how cosmic rays can either cause a pc to crash or the very least corrupt data.
Hey, im answering for rockitcool here since I know the answer, he changed the design of the IHS to make it work with a thermal grizzly contact frame!
Great and very underrated video.
Do you have any of those graphene thermal pad sheets? I would be really interested to see what reading those give you.
If the silver goes all the way through and is used as the interface with the water in the loop - I understand why there is a silver version - silver is antibacterial, so the silver block will inhibit algae growth in the cooler liquid, and algae will clog up the microstructures inside the block; so although it might have slightly lower cooling capacity when new, it will maintain that over a much longer time than the pure copper/nickel plated copper.
Another reason for the gold on contact points isn't just corrosion - it reduces the contact resistance as well.
I like that your cat likes to help you.
ive been oC'ing anything from athlon xp (which i burn't) from q6600, didnty felt assleep, loved the cat and loved the video. keep making good footages please ;)
Roman, silver is used to destroy bacteria and something like that in a liquid system, that's why some people use it, but also didn't expect that has worse thermal conductivity than copper 🤯
Koolance used to sell water blocks that they claimed were gold plated. I have a complete Exxos kit from forever ago and it certainly looks like gold.
Appreciate the content . It keeps my brain hunger in check
this was super informative, thanks a lot for taking the time to test this stuff.
The upshot I get is there is not a whole lot of point to lapping+liquid metal; other than a boat load of trouble down the road as everything corrodes. Plus points for an interesting physics lesson! 👍
Could you make video on how much more efficient cooling gets when the temperature ran as high as possible? (E.g. with lower fan speed)
At 7:01, it says Gold (Ag) and Silver (Au). The chemical symbols ought to be flipped.
As far as I know there is another reason for gold or silver plating: lower contact resistance. It's different than the electrical conductivity inside of the material. That's of course important for connectors, however digital transmission like with USB or HDMI cables does not see real gains from that.
Results with silver was very interesting, thanks!
I LOVE VIDEO LIKE THIS!! So good for arguing with ignorant people!!
Would it be possible to make an alloy of copper and silver that has better thermal conductivity than either element in their pure forms, especially at higher temperatures? Since you say that thermal conductivity generally decreases with temperature for pure elements, but rises for alloys.
You would think that what would be spoken about is resistance vs conductivity. Same thing I realize, but thermistors seem easier to talk about and resistance and temp track positively in pure metals.
Electrical and thermal conductivity directly proportionate? Just blown my mind!
Great video with high quality of technical details, as usual!
So I'm lapping my 5950X now! That's the only thing that really matters! 🤣
Silver is too soft and oxidizes thickly, if it's a harder silver alloy...it loses conductivity. There's a reason copper is so hard to beat.
Silver is too soft, that's why they use sterling. Its other components make it less soft than 999. It uses the silver because of biological growth in the water and cools decent enough. Maybe it would have been more idea to use a copper base and a silver lid, but that would be much more expensive.
Learned a lot from this. I had always thought gold was better, because NFS mentioned that the McLaren F1 had gold heatshields. Now I wonder why that was.
Really great man. Love the explanation on how to use the equation. Thank you!
Great video, very good explanation in complex and simple. And the Silver water block will feel like it works better because of the wallet factor...
Pero que maravilla de vídeo. Saludos desde Argentina. Gracias.
Great video! How do you calculate thermal conductivity on non conductive thermal paste?
Good content! 👍
Much better than when you play the basic utube influencer!
Which is a shame when being the CEO of a company.
By the way I switched to gelid GC extreme, the price of the kryonaut and kryonaut extreme is too damn high here in France!!!!
Kyronaut is also garbage for anything with shit heatsink-die contact i.e laptops.
t. happy GC-Extreme user
Red copper or yellow is already a big difference, and then there is the purity to get the composition is the main role, and it should not bend, that is the 2nd main roll, Cast iron was popular before modern insulation because it took a long time to heat up and cool down. If the Victorian look is what you're after, stainless steel versions of old cast iron designs are available.
yellow copper is a misleading term that just means brass or another copper alloy.
Hey Roman, I have the silver Aquacomputer block in the adjustable version with the 4 pins that can warp the profile of the block while mounted because the silver alloy they use is still softer than copper, so it makes it much less work to make it conform to unlapped blocks.
Not that I need it anymore since I swapped to a Der8auer Edition 8086k - lapped copper version, not the fancy silver IHS one - but I also have no reason to get a new block, because as you said the temperature difference is only half a degree purely based on materials.
Wow really great vid I'm running one of the silver blocks I have both when buying I did question the purity why only 925 right? Now the silver has made my loop a mess too. Really love let's check the science videos. Off to clean my loop.
Could you test the conductivity of copper that has been stained with liquid metal, next?
Silver is a better antimicrob then copper is. When your running straight distilled, with a silver block, your coolant over time would leach the silver ions and prevent nasty stuff from growing inside your system. The performance difference was very small, but it was more on not having to use additional additives in your coolant. Also Sterling Silver should not be used over pure Silver. Sterling Silver can cause corrosion which turned Silver products into Taboo for the hobby especially with nickle plating.
We also ran silver kill coils, however most silver kill coils that came after Iandh usually were tainted or fake with iron and other metals mixed into it, that it caused corrosion.
A wide ban on silver was brought about especially from eK, because of a few bad players who wanted to profit off fake kill coils.
Iandh and I btw were the first ones to experiment with silver kill coils on XtremeSystem.
All those times you said, "You can see." I could not see. What is the conductivity of gold? Why was I taught in school that gold is the best conductor?
I'm a vehicle mechanic and use copper paste and always wondered what it would be like to use compared to cpu paste?
waves with snaps quick flashes and a fan cooling conductive to insure it can take the extra away
LM on copper just creates an alloy layer really and I have never needed to lap it to add more LM to it. LM on silver turns into rock hard solid bumps which are a nightmare to lap.
Good Job, I give you a 99.93%
Basically, "Silver Plated" or even "Gold Plated" are just for marketing gimmick to make the product more "Premium" or "Luxury" but it perform worse than pure copper water block.
Wow, this was a great video on this topic. Thank you for the class.
from one engineer to another, this was beautifully explained 👌👌👌👌
Cool bloopers 😄 and nice master class
I felt like ı am at the Üniversity class again. Thank you professor
Amazingly informative video. Also, about the silver, would that be enough to act as a biocide?
No, it wouldn't leak ions the same way a kill coil does. Stirling silver is much more stable than the silver used in a coil.
@@FirmB1ade thanks! That answers another question I ask below about copper ions being enough of a biocide.
This takes me back to my thermodynamics class! My prof was also German. LOL
I didn't expect these results from the silver not being pure. As far as thermal conductivity goes, Silver is about 8-9% more thermally conductive than copper. Which is why silver is used for thermal compounds over copper.
I mean, theoretically a silver loop is better than a copper loop. (if price is not a consideration) But I haven't seen pure silver components for a long time (because they didn't sell well due to price in the late 90's early 2000's)
ESP haven't seen a silver radiator in a long time... which means you will get electrostatic migration when mixing metals, which is what I was assuming this video was going to talk about coming in. That migration is a pretty big problem, and is why I used to tell people to pick a metal and make the whole loop out of it. (Given, hardly anyone used to ask about a custom loop).
Really strange they would ruin their own silver heatsink by using cheap impure silver.
I'd say there's some chance that Aquacomputer offer silver as a cosmetic option, since the perforrmance should be essentially identical.
There is an error at the start, around the 2:00 mark, where you state that multiplying the interface's Thermal Conductivity by the Temperature Difference across it and then dividing by the Thickness (through which the conduction takes place), results in Watts of Heat being transferred through the interface.
This is incorrect: λ * ΔT / L has dimentions of (W/(m*K)) * (K) / (m) = W/(m^2); i.e. one obtains the Heat *Flux* (Watts per surface area).
Using the given example's dimentions of 100m * 100m = 10,000 (m^2) for the area of each of the large faces, the correct Rate of Heat Flow can be calculated by multiplying the obtained Heat Flux for the 2 material cases (first 400 W/(m^2) and then 200 W/(m^2)) by the total face area (10,000 (m^2)), giving 4MW and 2MW respectively:
400 W/(m^2) * 10,000 (m^2) = 4e+6 Watts, and
200 W/(m^2) * 10,000 (m^2) = 2e+6 Watts.
As for the emphasis on steady-state thermal conductivity, I think for modern CPUs, especially when overclocking and using water cooling, such a simple steady state analysis obscures some very significant transient differences between materials' thermal diffusivities.
Even at stock, modern CPUs and GPUs already make use of transient thermal headroom by alternating between boosting over the steady state TDP limit and dropping far below it, many times faster than the cooling system's characteristic thermal time constant.
(Basically, the waterblock *doesn't* behave like a steady-state thermal conductor; but rather 'absorbs' short bursts of transient heating, followed by periods of below-average heat input, during which thermal diffusion and ultimately conduction into the cooling water, cools the block back down.)
Therefore, I would very much like to see what difference a heat spreader (or waterblock) made of *pure* silver would make, seeing as Silver's Thermal Diffusivity is 49% higher than that of Copper!
Is there a differnce in conductivity between:
Material to Material
or
Air to material
?
Air|paste|Copper[------]Copper|TIM|chip
I'm guessing the thermal conductivity of copper is talking about the copper itsself?
The reason why that product is explained in 5:58. Your tests might be the issue. You should have backed it up with real tests, and with different frequency settings and at say +100°C as a second shot. If all those result are the same as your initial result. That only means they have poor quality silver.
So the best we'd want is the backside that touches the chip(s) to be copper, & water side plated nickel? Guess people could sand off the nickel of the back sides of their water blocks, this would be great to shown what difference this would make.
Fantastic - learned so much - thank you - will be watching it again
Thank you very much. This stuff is boring, but you are doing a wonderful service by your careful methods.
With this tool is it possible to do a COMPREHENSIVE test of the most popular thermal compounds and then compare the results to their "advertised" numbers? Would be interesting to see firstly if what most manufacturers advertise is true and then secondly based on these now accurate results, which thermal compound is REALLY the best.
tldw; You can afford a much more powerful and higher quality cooler for less than screwing around with silver and copper corrodes less.
Ancillary info: One of the problems that AMD is facing with their multi-chip components is that the different chips will have different thermal ranges to a certain engineering factor... this means that on occasion there may be intentional things like that thicker nickel plating on their packages. This is actually fairly common in older compact systems (power supplies, electrical management, etc.. including old heatsinked transmitter components) with heatsinks where some heatsinks have an increased thermal mass to them to better regulate to the operating temperature lifespan of given components and others may contain segments of copper or be made largely of copper instead of aluminum. In computer components the metal type doesn't matter so much as it does in high frequency equipment where the heatsink may literally be Tin (Sn).
Thinner spreader+cooler+faster is NOT best.
Also, there are several university projects where they are attempting to _grow_ diamond heat spreaders onto chips. Its only nearly 5x more conductive than silver...
It sounds like the kryos silver has a mild disruption of the plating interface to me... heat it up to 65°C and re-test it.
18:18 Silver edition meant to be a biocide, not for conductivity purpose
I learned something.
Isn't silver used in loops as a biocide? Seems that using the silver for the block would serve a similar function albeit for far more money than required/reasonable.
copper/brass is a biocide as well. Once door handles were done with it just for this reason.
@@marsovac they actually started doing it in hospitals
@@marsovac they still are made of copper, brass is just not cheap but it's a beautiful material, nonetheless.
You mentioned the silver water block is 92.5% Silver and 7.5% "Probably Copper" . . . but what if it's an alloy, like brass, and it's thermal conductivity either increases with temperature, or, stays stable and does not decrease with temperature?
925 silver is always with copper
Awesome video and keep up the great work.