convection cooling with a standard water setup

[DigitalChaos]

has anyone ever tried using a standard water cooling setup to do convection cooling? as in just taking the rad and puting it face down in a tube with no fan... would there be enough air flow created with the heat? how low do you think temps would be?

[futRtrubL]

It might work if you leave it suspended horizontally outside of the case. No tube though, no need.
If only I had the equipment and the expertise I would love to turn the 'unused' side pannel of a case into a large passive heatsink. Might work if you soldered square copper tubes in the back, parallel like in a heater core. Nice fanless system, plus you might get good flow rates.
|-------|
|-------|
|-------|
|-------|
|-------|
|-------|
In Out

Edward

PS. Why aren't multiple spaces allowed? It makes asci art so difficult.

[DigitalChaos]

actually i thought about that!.. also thought of building my case out of copper square tubing... and then running water through the pipe, and the water would be cooled via radiator. that way all the drives would be supercool!
and that case cooling system would integrate into the chips too....
but then i realized that is way too much $$ so i though of convection.
right now i have a BIX which does a damn good job with a fan, but i am tryig to cool too much in series. so the end blocks heat up alot. so i was thinkin i should get a second rad and put it half way through. (parallel flows would be better, but i didnt want to)

[DigitalChaos]

i guess what i want to know.. is what is the best layout for an enclosure i can make that will produce the most airflow... only with convection produced from the radiator

[futRtrubL]

Do you have any exhaust fans or holes that are not covered with a rad? Or even just the inside vent of your psu? Just put a small rad there, you still get active cooling without extra fans.

Edward

[DigitalChaos]

actually im going for completely fanless
and the case will be custom from plexyglass once i have all the designing done

[Brad]

it should work, but you'd need a big rad, and a smallish heatload, I think you'd end up with fairly high water temps easily, unless you mounted the rad near some form of airflow

[UnaClocker]

I remember back when watercooling was a new thing in the overclocking world, I read an article where someone was doing convection cooling with his K6-2.. It worked ok, but I think it hung out around 60c (don't all K6-2's hang out at 60c anyways? heh). So it works, more or less, but not that well..
Pumps really aren't loud, and if you pad them well, they're silent. You'll definately run into a hard drive noise issue if you run without fans before you run into a pump issue.

[Twitchie]

If you intend to go passive/convection cooling, then what you need to get efficient cooling is a high temperature differential between your coolant and ambient.

So either you leave your computer somewhere really cold, or you let/make your coolant get really hot.

You could try peltiers to get a higher temp in your coolant, but you would almost have to go with stacking them to get the desired effect.

I guess it all depends on what cpu you intend to use...I thought about 'underclocking' a TBird 1.0G I had that would run @ 1.3V and see if I could get away w/o a fan, but I didn't see enough of an advantage over a low noise 80mm fan/adapter (which is what I ended up using), since hdd and chipset fan noise couldn't be avoided easily.

Think about how houses are insulated, and why styrofoam cups keep our drinks hot/cold...they trap a layer of air because it is such a good insulator...that's why we need fans

convection, conduction, and radiation you are going to be using one or more of these methods to transfer heat out of your system into another system

[Brad]

the last thing you want to do is run pelts passively cooled because of their huge amount of heat, if you want to run something passively, buy an intel tualatin

[DigitalChaos]

thanks for the info so far guys!

actually i AM going to be using a pump still.. its just the rad that i want to be convection...
any of you guys know of a good design that would create the most airflow through my rad with convection?

[icel0rd]

Well, its interesting that you should mention that. I decided to experiment and try something similar.

I took my BIX and mounted it underneath my yy cube and ran the barbs up through the bottom into the mobo side of the case. I figured that heat rises, and the air is cooler down there, and with wheels, the bix has clearance.

Well, I found out that if you care about your temps, dont bother. You NEED to blow air on that rad or your water gets hot under load rather quickly.

I'm leaving my rad where it is, but I am going to try and design some sort of high CFM blower to pass air over it underneath the chassis. I'm thinking of mounting something like a turbine or two together in a little box-like unit and the intake for them comes from underneath and they can blow together just under the rad.

My temps under load range from about 45 to 48c with a fan blowing on my rad underneath, which is acceptable to me. Ambient temp is around 31c. These figures are reported via my mobo util, so the accuracy is probably off. I know there is still a lot of air in the circuit though. I can see a 5 inch stretch of watterless space in the circuit when I shut down the system. I have just been too lazy to put some more water in. I'm not sure how much of a difference its going to make though.


-Ice-

[Leuf]

Originally posted by icel0rd
I'm leaving my rad where it is, but I am going to try and design some sort of high CFM blower to pass air over it underneath the chassis. I'm thinking of mounting something like a turbine or two together in a little box-like unit and the intake for them comes from underneath and they can blow together just under the rad.

-Ice- [/b]

http://www.grainger.com/Grainger/cat...i&CatPage=3613

Not exactly what you had in mind, but might give you some ideas.

[Brad]

better than that would probably be having a couple of 69cfm 120mm's all blowing in, and having the only exit as the rad, sealing all other exits off

[futRtrubL]

I think you would first off need a rad with quite an open fin design, larger the better. It needs to be oriented so that air flows up through the fins. Lastly it neeeds to be unobstructed access ,above and bellow, to room air with a clear column straight up to get some good convection. My thoughts would be to cantilever it off the side or back of the case, though if the comp is under a desk you loose the clear column of air above it.

Edward

[icel0rd]

Yeah, I thought about moving it to another location, which would make it easier to resolve my airflow issue on the rad.

Since I took I mounted it underneath the chassis, my options are somewhat limited, and performance may be just as limited as my options given the setup.

I pretty much have 2 options. The first would be to design something to cool the rad with limited space underneath the chassis, especially when the rad already killed a lot of my under-chassis clearance.

The other option would be to just accept the fact that it was probably a dumb idea and move it to one of the two remaining viable spots. I could top mount it and go with Brad's airflow suggestion, or mount it in the same side of the chassis as the mobo and put it where the slots are to accommodate the huge PCI cards, where I would have to do a lot more customization (well, somewhat more) to get it behind those vertical rails. 2 120mm fans right there would be sweet. I may even consider isolating the rad (air-wise like I tried to do by bottom mounting) and make a bottom air intake and have dual 120 mm fans blow the exhaust out of the font of the chassis.

I really wanted to cut down on the darn fan noise though. The aircraft sounding machine used to sound cool, but over time it got really annoying, so I dropped scsi, dropped my heatpipe cooler with the loud fans, and wanted to get things as quiet and efficient as possible.

I appreciate the suggestions as always. You guys are a great help and give me lots of inspiration and motivation to try things.

Maybe I will try the blower box over my rad underneath and see how well it works. If the temps still suck, I will go with one of the above approaches.

[UnaClocker]

The difference between running a radiator with NO fan, and running one with say a 69cfm 120mm at 7v (about 40cfm) is AMAZING. If you were to take say, the thin 120mm 69cfm fan they sell at CaseEtc, which is already under 30db (like 24 if I remember right) and run it at 7v, it would not make a sound. That's what I think you should do.

[Brad]

I agree with that una, they'd be silent.


Also, I'm thinking about another idea, get a long bar of Al off DDen, cross drill it, then mount series of heatsinks on it.

Costing would be $4.5 for a 2" x 12" x 3/4" slab of Al, you could buy two of them, running them in parrallel, side by side, with 120mm's blowing over the heatsinks, which could be something like http://www.kdcomputers.com/eui/xNuPH...4/19/component as they are very cheap

for the richer people, you could cut the slabs of Al into 2" x 4" sections, and mount http://www.kdcomputers.com/eui/xNuPH...4/20/component on them

sounds like fun?

[sloppycoder]

have you considered leaving the fans on 7v but taking away the pump?

that sounds mad but..

there was a guy (trubador) on the OCS forums who wanted to see how long his cpu would last if the pump failed. the cpu was an athlon xp1500+ overclocked to 491mhz over stock. so he turned off the pump and waited, and it went from 40 odd degrees to 57 degrees and stayed there. this was seemingly because the radiator was mounted in the top of his case above the waterblock. so the hot water rose to the rad, was cooled and went back down again. also this rad was a noname, copper, double pass continuous winding tube. it looked like that pants one from 3DCOOL, and with one panaflow 120mm on it.


slop

[Leuf]

Originally posted by DigitalChaos
i guess what i want to know.. is what is the best layout for an enclosure i can make that will produce the most airflow... only with convection produced from the radiator

Okay you want to make it out of plexi, but what if you make the top out of aluminum (or copper if you're so inclined). Specifically, what if the entire top of the case was a big thin AL box, which you could make by sandwiching two plates on bar stock I guess. Wouldn't that give you a big passive radiator/reservoir? I don't know if you'd have enough surface area.

You could even mount the hdds directly to the bottom of it to cool them, using it as a big waterblock.

[Brad]

sloppy, the point is, how loud is a pump vs fans. If you run an Eheim in a resivior, mounted on rubber standoff's, I doubt it'd make any noise at all, and the fans on the rad could be pabst 12dba models

[redleader]

Get a big enough heater core and mount it perpendicular to the case and you can cool anything passively. Rising hot air should cool them if you have enough surface area. Coolant wouldn't need to be too much warmer. I'm too lazy to do the math, but with a a Cu or Al rad, a delta T of 8-10C should be plenty if you have a good deal of surface area. Get two and your comp will look like a plane

You could try peltiers to get a higher temp in your coolant, but you would almost have to go with stacking them to get the desired effect.

Heh maybe you should think that though. More heat != Easier to cool.

[icel0rd]

Ok, I just finished the handiwork on my system. The temps are WAY better now.

I drained the whole circuit, put in distilled + a little antifreeze to prevent algae and the murky water syndrome. I recleaned the processor core and block surface with alcohol and cleared the cheap silicone thermal compound. I used Antec reference 99% silver micronized compound (performs ~ as well as Artic Silver). The Antec compound runs circles around the other crap I used.

Now, I didn't want to add any more fans than I needed to.. sooooooooo.... I modified my cube by cutting out a fan hole through the sheet metal and the plastic case face (bottom left). The fan is pulling the air through the rad out of the front of the case. When I mounted the fan, I put a rubber gasket on the fan hole I cut, made custom rubber washers where the screws mount the fan to the sheet metal, and used rubber coating dip on the back side of the fan that touches the rad so there is 0 vibration noise (back side of the fan is all rubber now). I went out of my way to do things right this time and not be lazy.

The fan is not that loud at 12v, but now I think I will go ahead and put it on 7v as you all suggested and see how things go for noise reduction and performance.

The new temps are far superior to the BS I was running last time. 28C ambient 38C under full load for over 1 hour running seti@home.

Thanks for the suggestions gang.

[WebMasta33]

While your at the case door cooling thing... you might wanna make a misting tower...

The act of misting water cools off the water. Get about 20 misters, and you could mist off half your flow, and rad off the other half. In theory, it does work, but I don't know practical it would be to setup.

[icel0rd]

I'd like to see something like that in action. I can't imagine where I would put the parts to do it.

What you are talking about sounds something like the cray cooling blocks. I saw a video of one somewhere in the forum where that sucker was misting pretty hard.

I'm not sure how you could neatly contain the misting equipment or even what you could use that would fit and be practical. It would be neat to see though.


-Ice-