"direct die" HDD watercooling

krazy · 02-17-2004, 12:22 PM

I've been trying to come up with a way to build my HDD into my watercooling loop when I rebuild my air box for water. By sifting through all the discussions about HDD watercooling around here, I've been able to determine several well-agreed upon points:

- HDDs don't put out a ton of heat, so watercooling is almost always overkill
- however, it can eliminate the need for a fan or two, makes the whole setup quieter, less dependant on airflow[, and is pretty cool]
- the two main sources of head in a HDD are the spindle motor and air friction/turbulance
- it doesn't matter too much where the block goes on the drive if you simply want to suck up excess heat
- a lot of people like putting blocks on the sides of drives to take advantage of the large, flat, & exposed surface area

I toyed with the copper-pipe-soldered-to-a-copper-fin idea, and decided that was kind of ghetto. I'm also not a great soldering artist, so my prototype turned out kind of ugly (junk HDD shown):
pic 1 pic 2

I also looked into the stereotypical design incorporating milled copper blocks screwed to the sides of the drive (like the Innovatek cooler or the one Ben designed in another thread). I would do this, but my access to the tools required to mill, seal, tap, and finish a block myself is pretty limited. I also couldn't think of a way to get the return loop around the front of the drive elegantly without adding tight corners. I also want to be able to cool any of the particularly hot ICs on the drive's circuit board that might experience problems in a zero-airflow environment.

In the past, I've daydreamed about drive manufacturers making drive chassis out of copper with integrated water paths running through them (effectively making a bigben design that is more compact and eliminates one more surface coupling to transfer heat through), but this will likely never happen and I can't do it on my own.

The revelation I had today [in the middle of a C++ lecture] was to build a block that epoxies onto the bottom of the drive, using the drive's aluminum chassis for the "baseplate". There is another thread floating around here somewhere that shows a guy testing temperatures of various drive surfaces to decide which was best to pick up heat. His conclusion was to use the bottom surface, and had bent a loop of copper tubing that would bend around and mate with this surface (somewhat ackwardly and with little surface area). How about doing away with the bottom of the tube, per say, and just using the drive to get more surface area and solve the ongoing dilemma of how to pick the heat up.

The drive I would be doing this to is a Maxtor 7200rpm 120gb SATA150. (can't remember model number). It has a good sized area of exposed chassis around the spindle motor, and this area has radially-oriented fins or ridges built into it. I'm thinking that a block could be made with 1/2" copper tubing for barbs and an exchanging housing made of formed copper sheet metal. It would be a rough rainbow shape to follow the half-round shape of the exposed metal and thin enough that the water would have to sort of surge up and down to go past the drive's formed ridges.

Any screw holes or other leak paths would have to be sealed with epoxy, and the whole housing would be glued onto the drive with a healthy bead of 30 minute epoxy to hold it together and seal it on.

The drive is aluminum and has a layer of what seems to be anodization. I doubt this could have any effect on thermal properties worth caring about, but what about galvonic troubles? People always seem to have issues with running copper and aluminum in the same system. Would a few square inches of mostly anodized aluminum have any effects of this nature?

Anyone have ideas for improvements or reasons it wouldn't work or would be a bad idea? (besides probably voiding the warranty on the drive)

Gooserider · 02-18-2004, 12:29 PM

Interesting concept, not sure how it would work in practice, but the theory sounds OK. Obviously it would be a brand specific solution, and not portable from drive to drive, but I don't think either of those are terrible problems

The 'block' and barbs I would probably make out of plexi or other plastic just to get the copper away from the AL, and you would absolutely need to use anti-corrosives if there were other copper in the system, but you probably could get away with it.

You would also need to make sure that you didn't cover up the required air bleed hole (which all non-mil-spec drives have) I'm not sure where it is on your drive.

Caution would also be indicated in the choice of adhesives, you need something that will bond to both materials, and that will withstand the coolant pressure and solvent action, it might be worth trying to figure out a way to put some mechanical support on to hold the block instead of just relying on adhesives.

Finally, I get nervous around data storage devices since failures are so potentially nasty with them - before I started experimenting with this, I would ensure that I had a good backup / restore system that gets used regularly.

Gooserider

bigben2k · 02-18-2004, 12:52 PM

Actually, you can build the one I presented with a hacksaw, and a torch. If you want to get fancy with it, then incorporate #Rotor's design into it (requires a drill press), but I don't think it's worth the effort.

krazy · 02-18-2004, 12:55 PM

Quote:

Originally Posted by Gooserider

Obviously it would be a brand specific solution, and not portable from drive to drive, but I don't think either of those are terrible problems

Some drives have a PCboard covering the entire base, and some have a huge area of smooth uninterrupted aluminum to work with. This would depend heavily on how the individual drive was laid out. I looked at a [dead] standard Western Digital 40gb drive I have laying around, and it has a very nice looking aluminum area. You would have to probably patch one screw hole shut however (or work right next to it and use it to hold the cooler on more securely)

Quote:

The 'block' and barbs I would probably make out of plexi or other plastic just to get the copper away from the AL, and you would absolutely need to use anti-corrosives if there were other copper in the system, but you probably could get away with it.

Hey; I like the acrylic/plexi idea! I didn't even think of that.

I'll be using a brass/copper heatercore, a [copper-base] d-tek WhiteWater, and two copper-base Innovatek chipset blocks. I'm a little worried about the galvonic implications. Is it possible to use the right additives and have zero problems with corrosion these days, or can you only hope to slow the effects?

What about coating the base of the HDD in a thin layer of epoxy first? It would hinder heat transfer a bit, but I imagine it's still plenty good for cooling a hard drive. This would [theoretically] isolate the aluminum from the block and make it entirely safe. Anyone care to speculate on if this will work to stop corrosion and how badly it will impede heat transfer?

Quote:

You would also need to make sure that you didn't cover up the required air bleed hole (which all non-mil-spec drives have) I'm not sure where it is on your drive.

I believe the vent hole is on the top of this particular drive somewhere. but this is a very valid point if this was attempted on other kinds of drives.

Quote:

Caution would also be indicated in the choice of adhesives, you need something that will bond to both materials, and that will withstand the coolant pressure and solvent action, it might be worth trying to figure out a way to put some mechanical support on to hold the block instead of just relying on adhesives.

I like this idea as well. Those screw holes on the bottom of the drive have to be good for something, I guess. I'm thinking a plexi housing to carry the water over the drive's aluminum surface. This housing would be tall enough for 1/2" copper tubing to be epoxied onto the ends to extend back to the rear of the drive, serving as barbs. Any hot chips on the board could easily have a small copper finger or nub protruding down from one of the tubes to sink the heat off.

The whole unit can be built as one rigid thing and be held onto the drive mainly by those four or six screws, but the plexi housing can be securely attached and sealed with epoxy. The screws will make the cooler a lot more secure and rigid.

Quote:

Finally, I get nervous around data storage devices since failures are so potentially nasty with them - before I started experimenting with this, I would ensure that I had a good backup / restore system that gets used regularly.

Good point, but now you're making me nervous too.

I have no backups to speak of, but I am about to RMA my old IBM deathstar 40gb that still works but makes horrible racket and bearing noise. I can probably get all my important stuff on the replacement for this and stick anything left over on friends' servers or CDs.

I'll start playing with designs and layouts when I get some free time again. Keep the ideas flowing!

|kbn| · 02-18-2004, 01:41 PM

ive been thinking about watercooling hdd's for a while. i wouldnt do it just to remove the heat, thats not really worth the effort. if i did it would be to remove the 7200rpm whine that they make, even though its not much, at night it is clearly noticable. thats from a 120gb 7200rpm 2mb pata maxtor and a 160gb 7200rpm 8mb sata maxtor.

recently i set up a second computer in my room, this has 2 hdd's ones 3.2gb and the other is 4gb. both were annoyingly loud. i thought about it, and decided they must be very low rpm compared to my 7200's, so heat wouldnt be much of a problem. so in the end they both found there way into cardboard boxes, surrounded by 1 1/2" thick foam/sponge packaging stuff, which makes them compleatly silent. if i watercooled my 120gb and 160gb drives, they would find there way into similar cardboard boxes

i dont htink there would be any point in direct cooling on a hdd, its not worth the problems with corosion it would cause imo. the copper pipe idea is what i would do, having 3 lenghs going around each drive with 90 degree elbows, or possibly just one lenght with two elbows to join the two drives pipes - just cooling one side of each. what it looks like wouldnt really matter, as they would be inside a box, the type of box im no sure about. cardboard is easy, but you can get alu boxes that are about the right size too...

krazy · 02-18-2004, 02:15 PM

kbn: do you have any way of cooling the ICs on the board? or are they sitting there getting scaldingly hot inside that insulating box? Once you remove the normal airflow from something, you have to think about what else was being cooled by that air spilloff and find a way to make sure it can remain cool enough without the air.

My plan will allow me to cool both the drive's motor and any hot ICs on the drive's control board.

Arcturius · 02-18-2004, 02:47 PM

I think this is an interesting idea, but you're probably creating more problems than you'll solve...

I don't have a new Maxtor in front of me ATM, but the one's I've worked with in the past seem to fit your description--the front (away from connectors) 1/3 of the aluminum casting has a recessed area in the bottom with structural ridges that look roughly like radial and circumferential lines on a platter. The ones I've seen also have the sides and front of the casting extending in thin walls downward to make the disk more 'square'.
If this is the case, I'm assuming you intend to make a single vertical wall towards the rear of this recession, and then bolt another plate across the bottom of the drive using the standard mounting holes, to make a box, right? Do you intend to run your outlets through the front of the drive, or down out of the box (down, I would presume).

Basically, this sounds like a big sealing nightmare, if you ask me...
Since some added bulkiness doesn't appear to be a big issue, what about bolting an AMD block to the side of your drive? The 102mm distance between holes on a HDD matches exactly the larger distance of the AMD mounting holes. Then, the cooler could move between drives later, and wouldn't void your warranty. Plus, you wouldn't have to worry about seals as much.

From my experience with Maxtors, a valid warranty and good back-up plan will be your two best friends.

|kbn| · 02-20-2004, 01:37 PM

afaik the ic's on the bottom dontmake nearly enough heat. on a old hdd i wouldnt bother but a newish one it might be good idea to stick a passive heatsink or heatspreader one. that allows it to store/disipate the heat without airflow...
maybe just get a flat sheet of copper and stick that onto the ic's and have that going round to the side and soldered to the copper pipe? that would be plenty...

krazy · 02-20-2004, 09:05 PM

Quote:

Originally Posted by |kbn|

maybe just get a flat sheet of copper and stick that onto the ic's and have that going round to the side and soldered to the copper pipe? that would be plenty...

That's more or less what I have in mind. Having the two copper pipes running along the bottom of the drive instead of the top will allow me to mount the drive in the stock drive cage instead of having to use a CDrom bay and the pipes will already be very close to the IC's, making the length of the copper lead cooling them much shorter. It could probably be done with a chunk of really fat copper wire with the end hammered flat.

|kbn| · 02-20-2004, 09:30 PM

heat wont go from the mechanical part of the drive through the IC's to your w/c. the top will get too hot. thats where 80% or more of the heat is made... the top doesnt seem as hot though, thats becuase the heat is spread out to a large area..
i touched bottom of my 160gb maxtor (i couldnt see where i was touching btw) and there were a few very hot IC's, doubt they make much heat, but they odnt disipate very well under there....
so how about this idea? first teh drive is mounted in 5 1/4" bays, tbh i prefer them, they are easyer to get drives in and out of, and dont get in the way of compents on the mobo, like the 3.5" ones do.... this is good unless you have a LOT of drives. also means you can take out your 3.5" bays so that you can fit more w/c stuff in the bottom

pipe goes round the drive, using copper elbows, it goes all the way round exept for the back end where the conenctors are. then you get a piece of flat copepr sheet which is as long as the drive, and the width is the same as the drive + 2x the hieght. this allows you to bend it up the sides. you solder the pipe onto the sheet. the sheet also has screw holes for holding the drive. also the sheet has some kind of clip, so that this style of mounting is possible easly, http://www.spodesabode.com/content/a...hddnoise/print
also with the drive mounted, you can fit thick peices of packaging foam around it to reduce all of the niose - youve already got rid of the vibrations with the bands.
the only problem i can see is the tubing. i cant find anywhere selling 12mm OD copepr pipe, so im using 15mm which is standard size over hear... its the only pipe i can get.. tubing is 12mm ID, so that causes a problem. the tubing takes ages to fit on, but is possible with most tubing..

02-17-2004, 12:22 PM	#1
krazy Cooling Savant Join Date: Mar 2003 Location: St. Paul, Minnesota, USA Posts: 123	"direct die" HDD watercooling idea I've been trying to come up with a way to build my HDD into my watercooling loop when I rebuild my air box for water. By sifting through all the discussions about HDD watercooling around here, I've been able to determine several well-agreed upon points: - HDDs don't put out a ton of heat, so watercooling is almost always overkill - however, it can eliminate the need for a fan or two, makes the whole setup quieter, less dependant on airflow[, and is pretty cool] - the two main sources of head in a HDD are the spindle motor and air friction/turbulance - it doesn't matter too much where the block goes on the drive if you simply want to suck up excess heat - a lot of people like putting blocks on the sides of drives to take advantage of the large, flat, & exposed surface area I toyed with the copper-pipe-soldered-to-a-copper-fin idea, and decided that was kind of ghetto. I'm also not a great soldering artist, so my prototype turned out kind of ugly (junk HDD shown): pic 1 pic 2 I also looked into the stereotypical design incorporating milled copper blocks screwed to the sides of the drive (like the Innovatek cooler or the one Ben designed in another thread). I would do this, but my access to the tools required to mill, seal, tap, and finish a block myself is pretty limited. I also couldn't think of a way to get the return loop around the front of the drive elegantly without adding tight corners. I also want to be able to cool any of the particularly hot ICs on the drive's circuit board that might experience problems in a zero-airflow environment. In the past, I've daydreamed about drive manufacturers making drive chassis out of copper with integrated water paths running through them (effectively making a bigben design that is more compact and eliminates one more surface coupling to transfer heat through), but this will likely never happen and I can't do it on my own. The revelation I had today [in the middle of a C++ lecture] was to build a block that epoxies onto the bottom of the drive, using the drive's aluminum chassis for the "baseplate". There is another thread floating around here somewhere that shows a guy testing temperatures of various drive surfaces to decide which was best to pick up heat. His conclusion was to use the bottom surface, and had bent a loop of copper tubing that would bend around and mate with this surface (somewhat ackwardly and with little surface area). How about doing away with the bottom of the tube, per say, and just using the drive to get more surface area and solve the ongoing dilemma of how to pick the heat up. The drive I would be doing this to is a Maxtor 7200rpm 120gb SATA150. (can't remember model number). It has a good sized area of exposed chassis around the spindle motor, and this area has radially-oriented fins or ridges built into it. I'm thinking that a block could be made with 1/2" copper tubing for barbs and an exchanging housing made of formed copper sheet metal. It would be a rough rainbow shape to follow the half-round shape of the exposed metal and thin enough that the water would have to sort of surge up and down to go past the drive's formed ridges. Any screw holes or other leak paths would have to be sealed with epoxy, and the whole housing would be glued onto the drive with a healthy bead of 30 minute epoxy to hold it together and seal it on. The drive is aluminum and has a layer of what seems to be anodization. I doubt this could have any effect on thermal properties worth caring about, but what about galvonic troubles? People always seem to have issues with running copper and aluminum in the same system. Would a few square inches of mostly anodized aluminum have any effects of this nature? Anyone have ideas for improvements or reasons it wouldn't work or would be a bad idea? (besides probably voiding the warranty on the drive) __________________ Abit NF7-S ... xp1800+ ... Abit GF4ti4200 ... Maxtor 120gb SATA150 Currently gathering parts to water cool everything! (read about the project) Last edited by krazy; 02-18-2004 at 09:25 AM.

02-18-2004, 12:29 PM	#2
Gooserider Cooling Savant Join Date: Jun 2003 Location: North Billerica, MA, USA Posts: 451	Interesting concept, not sure how it would work in practice, but the theory sounds OK. Obviously it would be a brand specific solution, and not portable from drive to drive, but I don't think either of those are terrible problems The 'block' and barbs I would probably make out of plexi or other plastic just to get the copper away from the AL, and you would absolutely need to use anti-corrosives if there were other copper in the system, but you probably could get away with it. You would also need to make sure that you didn't cover up the required air bleed hole (which all non-mil-spec drives have) I'm not sure where it is on your drive. Caution would also be indicated in the choice of adhesives, you need something that will bond to both materials, and that will withstand the coolant pressure and solvent action, it might be worth trying to figure out a way to put some mechanical support on to hold the block instead of just relying on adhesives. Finally, I get nervous around data storage devices since failures are so potentially nasty with them - before I started experimenting with this, I would ensure that I had a good backup / restore system that gets used regularly. Gooserider __________________ Designing system, will have Tyan S2468UGN Dual Athlon MOBO, SCSI HDDS, other goodies. Will run LINUX only. Want to have silent running, minimal fans, and water cooled. Probably not OC'c

02-18-2004, 12:52 PM	#3
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Actually, you can build the one I presented with a hacksaw, and a torch. If you want to get fancy with it, then incorporate #Rotor's design into it (requires a drill press), but I don't think it's worth the effort. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

02-18-2004, 02:15 PM	#6
krazy Cooling Savant Join Date: Mar 2003 Location: St. Paul, Minnesota, USA Posts: 123	kbn: do you have any way of cooling the ICs on the board? or are they sitting there getting scaldingly hot inside that insulating box? Once you remove the normal airflow from something, you have to think about what else was being cooled by that air spilloff and find a way to make sure it can remain cool enough without the air. My plan will allow me to cool both the drive's motor and any hot ICs on the drive's control board. __________________ Abit NF7-S ... xp1800+ ... Abit GF4ti4200 ... Maxtor 120gb SATA150 Currently gathering parts to water cool everything! (read about the project)

02-18-2004, 02:47 PM	#7
Arcturius Cooling Savant Join Date: Aug 2001 Location: Chippewa Falls, Wisconsin Posts: 294	I think this is an interesting idea, but you're probably creating more problems than you'll solve... I don't have a new Maxtor in front of me ATM, but the one's I've worked with in the past seem to fit your description--the front (away from connectors) 1/3 of the aluminum casting has a recessed area in the bottom with structural ridges that look roughly like radial and circumferential lines on a platter. The ones I've seen also have the sides and front of the casting extending in thin walls downward to make the disk more 'square'. If this is the case, I'm assuming you intend to make a single vertical wall towards the rear of this recession, and then bolt another plate across the bottom of the drive using the standard mounting holes, to make a box, right? Do you intend to run your outlets through the front of the drive, or down out of the box (down, I would presume). Basically, this sounds like a big sealing nightmare, if you ask me... Since some added bulkiness doesn't appear to be a big issue, what about bolting an AMD block to the side of your drive? The 102mm distance between holes on a HDD matches exactly the larger distance of the AMD mounting holes. Then, the cooler could move between drives later, and wouldn't void your warranty. Plus, you wouldn't have to worry about seals as much. From my experience with Maxtors, a valid warranty and good back-up plan will be your two best friends. __________________ Everything aircooled (I honestly don't know why you guys let me stay a member...) Always folding: WXP Pro: KD7R + Palomino 1800+ W2K3: 4x P3 Xeon 500 512KB (4 clients) Linux boxes: none Had to put them to use doing real work.

02-18-2004, 01:41 PM	#5
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	ive been thinking about watercooling hdd's for a while. i wouldnt do it just to remove the heat, thats not really worth the effort. if i did it would be to remove the 7200rpm whine that they make, even though its not much, at night it is clearly noticable. thats from a 120gb 7200rpm 2mb pata maxtor and a 160gb 7200rpm 8mb sata maxtor. recently i set up a second computer in my room, this has 2 hdd's ones 3.2gb and the other is 4gb. both were annoyingly loud. i thought about it, and decided they must be very low rpm compared to my 7200's, so heat wouldnt be much of a problem. so in the end they both found there way into cardboard boxes, surrounded by 1 1/2" thick foam/sponge packaging stuff, which makes them compleatly silent. if i watercooled my 120gb and 160gb drives, they would find there way into similar cardboard boxes i dont htink there would be any point in direct cooling on a hdd, its not worth the problems with corosion it would cause imo. the copper pipe idea is what i would do, having 3 lenghs going around each drive with 90 degree elbows, or possibly just one lenght with two elbows to join the two drives pipes - just cooling one side of each. what it looks like wouldnt really matter, as they would be inside a box, the type of box im no sure about. cardboard is easy, but you can get alu boxes that are about the right size too...

02-20-2004, 01:37 PM	#8
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	afaik the ic's on the bottom dontmake nearly enough heat. on a old hdd i wouldnt bother but a newish one it might be good idea to stick a passive heatsink or heatspreader one. that allows it to store/disipate the heat without airflow... maybe just get a flat sheet of copper and stick that onto the ic's and have that going round to the side and soldered to the copper pipe? that would be plenty...

02-20-2004, 09:30 PM	#10
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	heat wont go from the mechanical part of the drive through the IC's to your w/c. the top will get too hot. thats where 80% or more of the heat is made... the top doesnt seem as hot though, thats becuase the heat is spread out to a large area.. i touched bottom of my 160gb maxtor (i couldnt see where i was touching btw) and there were a few very hot IC's, doubt they make much heat, but they odnt disipate very well under there.... so how about this idea? first teh drive is mounted in 5 1/4" bays, tbh i prefer them, they are easyer to get drives in and out of, and dont get in the way of compents on the mobo, like the 3.5" ones do.... this is good unless you have a LOT of drives. also means you can take out your 3.5" bays so that you can fit more w/c stuff in the bottom pipe goes round the drive, using copper elbows, it goes all the way round exept for the back end where the conenctors are. then you get a piece of flat copepr sheet which is as long as the drive, and the width is the same as the drive + 2x the hieght. this allows you to bend it up the sides. you solder the pipe onto the sheet. the sheet also has screw holes for holding the drive. also the sheet has some kind of clip, so that this style of mounting is possible easly, http://www.spodesabode.com/content/a...hddnoise/print also with the drive mounted, you can fit thick peices of packaging foam around it to reduce all of the niose - youve already got rid of the vibrations with the bands. the only problem i can see is the tubing. i cant find anywhere selling 12mm OD copepr pipe, so im using 15mm which is standard size over hear... its the only pipe i can get.. tubing is 12mm ID, so that causes a problem. the tubing takes ages to fit on, but is possible with most tubing..

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)