Custom HDD Cooling

One_Hertz · 12-24-2005, 01:36 PM

Hey guys,

My dad, Andrei Shirobokov, owns Ace Data Recovery Engineering Inc. (www.acedre.com) that is currently the leading North American supplier of Data Recovery equipment and he is now developing a brand new product in that field. What he is looking for is some sort of custom cooling device to supply with his new product that will cool the HDDs being worked on, since they tend to work better (like most electronics) under cold conditions. He is not sure right now how cold he wants them since he only recently started doing R&D on that topic. The cooling device he is looking for should be able to connect to as many hard drives at once as possible and should also be able to do only one if needed. As for the temperature, the key point is that is must be under complete control starting from about 20C down to as cold as possible for now just to figure out at what point temperature has no effect on the readability of the drive. The temperature will eventually be controlled by the same software that the equipment uses. What I was thinking of is to use single stage phase change with a few heads on it that would be like hdd waterblocks. On the other hand the hard drives should be easily interchangeable and mounting a block on it takes some time. Condensation is another issue that i have no clue how to solve. Any ideas anyone?

P.S. already pmed chilly about it

yngndrw · 12-24-2005, 01:54 PM

Have you looked into chilled liquid cooling ?

sin0822 · 12-24-2005, 02:00 PM

how are you going to deal with condensation?

yngndrw · 12-24-2005, 02:04 PM

I think that as long as you have a good airflow, you wont have a problem with condensation. However, the colder it goes, the more airflow that you need. Dont hold me to this though.

afireinside · 12-24-2005, 02:11 PM

Hard drives don't like cold...

catkicker · 12-24-2005, 02:14 PM

Quote:

Originally Posted by afireinside

Hard drives don't like cold...

What do you consider "cold"

One_Hertz · 12-24-2005, 02:15 PM

Quote:

Originally Posted by afireinside

Hard drives don't like cold...

The ones in need of data recovery do, a lot. 5C can be a difference on whether an hdd is recoverable or not, I have seen this happen.

No I haven't looked into chilled watercooling since it will recquire maintenance once in a while and this will not be possible. I am looking for an industrial type machine so it can not be flaky at all.

sin0822 · 12-24-2005, 02:28 PM

how baout cold plates?

One_Hertz · 12-24-2005, 02:42 PM

Quote:

Originally Posted by sin0822

how baout cold plates?

Looked into that and did not really find a way to control that software wise. Would be nearly impossible to exactly controll temperatures too since the heat output of every drive is different. Looks like phase change is best for this, but other ideas are very welcome. Can evap temp be controlled and put at like 0C if needed, or is it only low negatives?

aussie_guy00000 · 12-24-2005, 04:15 PM

You need a freezer, an insulated box (like an esky) with an evap in it, that way you don't have to worry about condensation. Just open the box up whack as many hard drives as you want in, fire her up and it won't be hard to control by software. Obviously you'll need a drain for the evap as it'll suck the moisture out, some hard drive racks, and some sort of opening that allows you to pass the HD cables through and seal around them.
I'm currently designing such a system for Complete computer cooling, It'll look like a normal computer case (maybe a big larger) and have a hermatically sealing chamber with evap inside. You simply open the case up install your MB, graphics, HD, etc, although the optical drives and PSU fit in a chamber outside the sealed container. Seal him up and it drops the whole thing to probably about -20deg. I can't devulge too many of the details as I'm hoping to patent it and manufacture it, but I've got most of the details sorted out. This way you've got no worries with condensation (well the way i'm doing it you don't) and it cools everything from Hd to the MB. Obviously the effective ness of cooling your cpu and GPU will still rely on how good your heat sink is, but i'm working on that too. When I get the patents sorted out, i'll post some more details and pictures of the prototypes. Cheers

yngndrw · 12-24-2005, 04:30 PM

This any good?

wdrzal · 12-24-2005, 05:19 PM

I found if you just keep computer rooms at around 65 F and 40 to 60 % humidity & that stock air cooling keeps them working fine. since almost all equiptment comes with air cooling ,All the pc users we work for just elect to keep the rooms from overheating and the stock cooling does the rest. just a properly sized a/c unit with electronic air filtration to keep the air nice and clean and dust free is the best route.

Sidewinder · 12-24-2005, 05:22 PM

From what i read it sounds like that the cooling solution should be flexible, have a variable temperaturerange and a variable power (Qo, Kälteleistung in German, English word?)

A waterchiller with PMEV for temperaturecontrol and "something" to control the power, could be a frequency...?? (Frequenzumrichter), or a reservoir.
the discs can be connected any time to the system, no need to work on the coolingcircle. connect the discs serial would use less space but may be inefficent...
A "thingie" like yngndrw explained could be used as heattransformer between watercircle and discs.

Sorry, but my English gone horrible...

aussie_guy00000 · 12-24-2005, 05:28 PM

Well I guess it really depends on what you temps you need. I assumed that since you posted this in the phase change thread, that you're after below zero temps. If this isn't the case, then what wdrzal said is the go, just do it in a room with the aircon cranked. If you use below zero water chilling or other phase change methods, condensation is an issue. I still think my idea would be the go in the below zero region, but obvioulsy if could be designed to do say 25c - -25c and you just set the temp but computer programming, obviously would require a controller but.

Aphex_Tom_9 · 12-24-2005, 06:22 PM

Quote:

Originally Posted by afireinside

Hard drives don't like cold...

that seems to be the consensus. exactly how cold? i think anything under 10c is considered too cold. not sure on this though.

One_Hertz · 12-24-2005, 06:29 PM

I am sorry i didnt really make myself clear. It is something we will need to ship to our customers along with the equipment, like a compartment to keep the drives cool while the hardware does what it needs to. That company does not actually do data recovery, just sells equipment and does consulting. The new product we have gets the drives ridiculously hot so the normal aircooling can not keep up well and if left without any cooling the drive will be dead withing hours. It has to cool real well and down to about 0C from what we can tell right now to keep up with the extra stress that the pcb will be recieving. Our engineers do not THINK that below 0 will be needed but it is also something we need to experiment on, that is why I thought phase change will be the best way. It has to be real simple and controlled by software so the people who buy the product can use it no problem and without any knowledge of cooling and such.

aussie_guy00000 · 12-24-2005, 06:42 PM

well to remove that much heat that quick, you'll probably need a direct conduction cooler (chilled water, or Direct evap), but it makes it difficult to do it on my then one at a time.

Bloody_Sorcerer · 12-24-2005, 07:08 PM

sounds to me like a solid use of thermostat-controlled liquid chilling, some custom-designed HDD waterblocks, and a quick mounting solution.

epion2985 · 12-24-2005, 07:25 PM

Magnetic materials dont like cold. As ugly once said freezing magnetic materials can scramble stored data and cause alteration (unwanted alteration) of the write strength of the heads. The exact temp at which you lose cohesion depends on the specific density of the drives data and the ability of magnetic layer to hold the data. Stay away from temeratures below zero. And I would recomend stay over 10 degrees C period to be safe. Also going between room temperature hot and cold means stress due to thermal expantion and contraction, stress the hard drives are not designed for, meaning stress that can have very bad consequences.

Water cooling is a good way to keep the temperatures in check. However I advise against cooling that water below room temperature.

STEvil · 12-24-2005, 08:25 PM

first would be to test what temperatures are what you should aim for.

To do this I would suggest a water loop using a copper plate (as the waterblock) you can strap/bolt to the HDD to keep the HDD cool. Adding a chiller to bring down the temperature slowly and test approx every 5c to 10c drop in temps (starting at about room temp or slightly below most likely) and measuring the affect this has on data recovery would be the way to go....

A decent chiller should allow you to get down to -20c on a hdd and can be built from something as simple as a water cooler (although that type might not get you down to -20c).

If you keep getting better results as you keep going down then using a phasechange cooler and adjusting the temperature manually will be the next step.. I'm no expert there, but there are valve you can add to control refrigerant flow and you can influence temperature by amount of refrigerant in the system.. as well as by increasing/decreasing the compressor speed etc... so I assume if you're getting into it this far that you will have someone who can do it for you or good enough directions from other members here or other people in the refrigeration buisness on how to do this stuff exactly.

Good luck.

epion2985 · 12-25-2005, 12:29 AM

Quote:

Originally Posted by STEvil

first would be to test what temperatures are what you should aim for.

To do this I would suggest a water loop using a copper plate (as the waterblock) you can strap/bolt to the HDD to keep the HDD cool. Adding a chiller to bring down the temperature slowly and test approx every 5c to 10c drop in temps (starting at about room temp or slightly below most likely) and measuring the affect this has on data recovery would be the way to go....

A decent chiller should allow you to get down to -20c on a hdd and can be built from something as simple as a water cooler (although that type might not get you down to -20c).

If you keep getting better results as you keep going down then using a phasechange cooler and adjusting the temperature manually will be the next step.. I'm no expert there, but there are valve you can add to control refrigerant flow and you can influence temperature by amount of refrigerant in the system.. as well as by increasing/decreasing the compressor speed etc... so I assume if you're getting into it this far that you will have someone who can do it for you or good enough directions from other members here or other people in the refrigeration buisness on how to do this stuff exactly.

Good luck.

Wrong and a bad idea at best.

1. -20C will most likely destroy the drive, its data, its capability to store, read or write date, or all of the above.

2. Testing by going lower and lower if the previous step was weathered by the drive is a recipe for a disaster. First of all not all drives are made alike, even different drives of same brand model and make are not identical and may or may not react identically. Further more the damage may be not visible immediately. The drive may fail as a result; a day or a year later, and the data may be alterned or corrupted without your notice.

What remains true are two things. One, the hard drives of today are not made to weather stresses from thermal expansion and contraction subzero temperatures would induce on them.

And two, freezing magnetic materials changes magnetic properties of the there of materials and hence will change the way the drive will operate, most likely resulting in complete failure since any variation from how they are made to operate equals different operation for which they are not made to work under.

Carlz0r · 12-25-2005, 12:55 AM

For your dad's harddrive cooling, what about using TEC's with varied voltages? You could probably use some form of microcontroller to vary the resistance in the current path, thereby lowering the voltage which in turn lowers the temperature of the TEC. Couple that with a few high cfm fans making sure condensation doesn't occur and you'd be able to control in between 30 degrees to subzero temperatures (dependant on tec wattage). Hopefully that idea helps.

epion2985 · 12-25-2005, 01:17 AM

TEC's are a waste of power. His dad runs a company and I am sure he has a budget. Water cooling with a coolant to air heat exchanger would be a much more efficient and apropriate solution.

Thrilla · 12-25-2005, 01:35 AM

Low temperature might alter the physical properties inside the HDD, at extremely low temperature the magnet will become a super conductor, so, phase change might be a bad idea. And things usually shrink at low temp. Unless you have a HDD made for low temps, I wouldn't risk it.

Chill water (10-20'C) with good condensation proof would be nice.

Water cooling is the most affordable and efficient method, just make some special blocks so water goes in from one side, and comes out from the opposite to reduce flow resistance. Then you can controll the temp with water flow, pump+powerful fan controller.

Or just use air, use a low profile HS

TEC is a waste of power, they could be useful, but I don't think it's worth it, just use chilled water insdead, and tune(underclock) it to almost ambient

epion2985 · 12-25-2005, 03:28 AM

Actually this water cooling solution for hard drives from Asetek does incredibly well. High flow rate as well. It cools the bottom of the drive which is most effective because that is where the spindle is as well as the circitry are. However this solution is only as effective as the contact, so they sell a thermal conduction pad to fill the spaces between the block and the bottom of the drive. Like thermal compound but its a pad, alot cleaner

Block is ~$40-$50, pad ~$20

Tests show a 20C degree improvement all over the board, idle as well as full load. Thats very nice in my book. I am personaly planning to use these with my 4 15k rpm SAS drives. At firs when I saw it I thought it would preform like crap because the bottom of the drive is not a smooth surface, but that was before I knew about the pad.

Review:
http://www.bjorn3d.com/read.php?cID=714

My 2 cents, use this block with the pad and chill water but do not take it lower then 15C