Introducing my new watercooling design concept!

chylld · Post by **chylld** » Tue Aug 03, 2004 6:57 pm

Straker wrote:passive watercooling will fit into the picture in a year or two when passive heatsinks (or even normal HSFs, if we're unlucky) for any modern CPU/GPU are so big you can't get air through the case any more.

i think passive watercooling in a confined space is better done with heatpipes. it's cheaper, much much less prone to failure and probably performs close to it as well.

i'm not doing a u-turn on watercooling here btw, i'm just posing some questions that are good to think of since they help everyone get a more overall view of watercooling.

Straker · Post by **Straker** » Tue Aug 03, 2004 7:57 pm

chylld wrote:i think passive watercooling in a confined space is better done with heatpipes. it's cheaper, much much less prone to failure and probably performs close to it as well.

funny that you said that, i said something almost identical in a diff thread a few days ago, but that was regarding passive watercooling with no pump even, which is just like a crappy heatpipe.

i didn't mean no pump and i didn't mean confined space, but you don't really have any other options with a normal heatsink (or heatpipe), i don't think it'd be possible to get the heat very far away fast enough... whereas water is easy to get out of the case and back.

chylld wrote:i'm not doing a u-turn on watercooling here btw, i'm just posing some questions that are good to think of since they help everyone get a more overall view of watercooling.

yeah, i actually didn't even read all 7 pages before my first reply, i thought this was just another WC thread at first.

i know it's not really relevant now that the CSP750 exists, but this guy managed to trim an Eheim 1250 way down to fit into a 1048 housing, scroll about 3/4 down:
http://bit-tech.net/article/119/

just demonstrates that since you're making your own housing for the system as a whole, you might be able to save a lot of room inside the case.

mpteach · Post by **mpteach** » Tue Aug 03, 2004 8:32 pm

Chyld its so sad to hear that. Hmm .. was really hoping for something better than the reserator.

I find two distint trends in technology, the mobile trend, characterized by low power laptops, wireless, passive, cellphones etc, and The power trend characterized by exponentially powerfull and heathog computers and fiber optic links, suvs, nvidia 6800s etc. Computers havent event caught up to the human body for heat production.

it seems that increasing technology can be used for effieciency or power.d Both trends compliment each other, like fiber optic links connected to Wireless points. I dont see myself living in a trailer, driving a hybrid and watching the death of watercooling.

EDIT, chyld if you think heatpipes and passive are the way, look at the cost of the zalman heatpipe computer case, and the cooling of the semi-passive reserator.

mpteach · Post by **mpteach** » Tue Aug 03, 2004 8:37 pm

If yove completly given up, can you share the plans?

chylld · Post by **chylld** » Wed Aug 04, 2004 4:46 am

hi mpteach,

if i had enough time to draw up the plans on the computer and label everything in plain english then i might as well just make a prototype and write a report on it. unfortunately my time is too constrained for that kind of activity so i'll be keeping my plans to myself for the time being. not to be selfish, but it's something that i think has serious potential and when i have the time i do intend to pursue it to completion. things are just being put on hold at the moment.

sthayashi · Post by **sthayashi** » Wed Aug 04, 2004 5:07 am

Dammit... Now I'm going to have to go reserator for my watercooling kit needs.

chylld · Post by **chylld** » Wed Aug 04, 2004 5:15 am

sorry guys

i thought it was my responsibility to let you all know that things had been pushed back, rather than to pointlessly string you along for who-knows-how-long...

mpteach · Post by **mpteach** » Wed Aug 04, 2004 2:53 pm

Hey honesty's the best policy, well at least for the middle class.

mpteach · Post by **mpteach** » Thu Aug 26, 2004 12:45 pm

Its obvious that there are going to be 3 settings, on off and auto. The auto will be hooked to a relay of some sorts.

Rather than hook the relay to a complicated finicky power sensing cuircuit, hook it up to a PSU extention cable, that will hook up to a PSU molex connector inside the case, run out the back of the case, and connect to your watercooling box. When the PSU is off it outputs 0 volts, with it on it outputs 5v and 12v.

If reliability was a concern you could make the end of the extention that hooks to the watercooling box, some sort of locking connector or screw on connector. I doubt a few meters of wire and 2 connectors would cost much.

chylld · Post by **chylld** » Thu Aug 26, 2004 2:59 pm

i thought about that but i decided it's not worth the risk. there is the possibility that the user will forget to plug that cable in and so the computer will turn on without the unit. you can imagine what happens next.

also another cable coming out of the box would look ugly - ideally there'd only be 2 tubes and a power cable, that's it.

Gooserider · Post by **Gooserider** » Thu Aug 26, 2004 3:43 pm

Did you ever get a look at what Bladerunner did with his PSU? he put a circuit in it with a magnetic reed switch by the PSU input plug, and put a magnet in the power cord plug so that the PSU would only turn on if the specific power plug that also fed his pump was plugged in. A neat solution, but probably not workable since you wouldn't be able to mod the customers PSU.

I think the most reliable / simplest solution would be to have two cords in the cooler box, one being AC in, and the second having a female plug that would connect to the PC PSU. Make the computer power cord connect to the same switch that turns on the power to the pump. Assuming the customer follows the directions on how to hook up the PC properly, then you in effect have a switched outlet such that the PC CAN'T be turned on w/o turning on the pump. It also seems like a fairly elegant solution in that it doesn't require any additional AC outlets for the cooling box.

Gooserider

chylld · Post by **chylld** » Thu Aug 26, 2004 3:48 pm

Gooserider wrote:I think the most reliable / simplest solution would be to have two cords in the cooler box, one being AC in, and the second having a female plug that would connect to the PC PSU. Make the computer power cord connect to the same switch that turns on the power to the pump. Assuming the customer follows the directions on how to hook up the PC properly, then you in effect have a switched outlet such that the PC CAN'T be turned on w/o turning on the pump. It also seems like a fairly elegant solution in that it doesn't require any additional AC outlets for the cooling box.

that's exactly how i intended to do it.

i really want to get back into it once i find the time. the show i'm working for opens on the 15th of september so i should be finished before then.

mpteach · Post by **mpteach** » Thu Aug 26, 2004 8:35 pm

chylld wrote:i thought about that but i decided it's not worth the risk. there is the possibility that the user will forget to plug that cable in and so the computer will turn on without the unit. you can imagine what happens next.

also another cable coming out of the box would look ugly - ideally there'd only be 2 tubes and a power cable, that's it.

Theres the possibility that the user could forget to plug the unit into the wall, and turn on the computer. I expect people who can watercool can double check and plug in the ac cord and the molex. Locking connectors could help too. Theres always the leak test, why not make a wiring test?

As for cables being ugly, look at the back of any modern computer, i have over 15 cables.

mpteach · Post by **mpteach** » Thu Aug 26, 2004 8:53 pm

chylld wrote:
Gooserider wrote:I think the most reliable / simplest solution would be to have two cords in the cooler box, one being AC in, and the second having a female plug that would connect to the PC PSU. Make the computer power cord connect to the same switch that turns on the power to the pump. Assuming the customer follows the directions on how to hook up the PC properly, then you in effect have a switched outlet such that the PC CAN'T be turned on w/o turning on the pump. It also seems like a fairly elegant solution in that it doesn't require any additional AC outlets for the cooling box.
that's exactly how i intended to do it.

i really want to get back into it once i find the time. the show i'm working for opens on the 15th of september so i should be finished before then.

I still think the molex method is better. I happen to be very fond of putting my computer into standby, everything shuts down except for RAM, its completly silent and uses almos no power. In the morning i touch my mouse and the computer wakes up within a few seconds. With the dual cord method this isnt possible.

I prefer having the computer controlling the cooling, and not the cooling controlling the smarter computer, it just gives you more options, and will cause less problems.

When the molex is hooked up to the computer certain wires are grounded so a simple cuircuit (like 2 transistors) could make the unit run contunously if the molex if physically disconnected from the computer.

I dont understand the ugly wire argument.
1AC + 1 Molex = 1 AC + 1 AC, both methods require 2 cords

chylld · Post by **chylld** » Thu Aug 26, 2004 8:56 pm

If you use the wall --iec-> wc box --iec-> computer method, then there won't be any issues with the computer going into standby. you'd just put it into standby like normal and the wc box would keep on pumping.

the pump i was intending to use was an eheim pump for reliability reasons, which is AC powered, so if i were to use a molex cable then it'd require 3 cables: 1 for the computer, 1 for the wc box and 1 from the computer to the wcbox.

mpteach · Post by **mpteach** » Fri Aug 27, 2004 6:28 am

chylld wrote:If you use the wall --iec-> wc box --iec-> computer method, then there won't be any issues with the computer going into standby. you'd just put it into standby like normal and the wc box would keep on pumping.

the pump i was intending to use was an eheim pump for reliability reasons, which is AC powered, so if i were to use a molex cable then it'd require 3 cables: 1 for the computer, 1 for the wc box and 1 from the computer to the wcbox.

Why would i want to leave my box pumping when my computer is in standby? Thats a waste.

Some people use leak detection systems. If theres a leak and the computer shuts down, why would i want the wc box to keep pumping water into the rig thus ensuring major damage? The molex methods allows people to easily add smart controls to the box.

1 molex cable + 1 AC cable is still only 2 cables connected to the wc box. Only 2 cables you have to design, either way. Sure youd need an extra AC cable that connects from the computer to the UPS. Big deal, i already have 13 AC cables behind my desk already, and i think it would be worth 1 more.

A molex controlled Solid state relay should be easy so id probably mod my future WC box if it didnt have one.

BTW, where did the ugly wire concept come from? I though you said that asthtetics werent much of a concern and your just going with a wood box.

Keep up the good work Chylld, love the wc idea.

chylld · Post by **chylld** » Fri Aug 27, 2004 6:51 am

if i was just building it for myself then aesthetics aren't a concern no. but when you're building a product for the commercial market, every little detail counts.

a molex would be trivial to implement, and would look quite neat if i have the in/out tubes coming through one of the pci slots. but i don't want a standard molex cable running out the back, in my opinion that's ugly. it'd only need 12v anyway so i'd probably use some cheap connector like a 3.5mm stereo/mono jack.

mpteach · Post by **mpteach** » Fri Aug 27, 2004 7:45 am

chylld wrote:if i was just building it for myself then aesthetics aren't a concern no. but when you're building a product for the commercial market, every little detail counts.

a molex would be trivial to implement, and would look quite neat if i have the in/out tubes coming through one of the pci slots. but i don't want a standard molex cable running out the back, in my opinion that's ugly. it'd only need 12v anyway so i'd probably use some cheap connector like a 3.5mm stereo/mono jack.

Good idea, molex cables and connectors are ugly and designed for much higher currents than would be needed.

However I would suggest against using a 3.5mm stereo/mono jack because they are used primarily for audio.
A standard power/communication jack such as a DIN connector would be better. Plus you'd have a few extra pins for future use, for example a simple continuity cuircuit that would detect if the the conector was attached or not. A (mini-DIN)/(PS/2) connector would also work.

I just have visions of sombody finding a cable coming out the back of thier computer and then pluging it into their sound card.