BEEEP wtf is that beeping... ARGH why me...

[Snowman]

Well came in the living room sunday morning to an odd beeping... it was my pump telling me it was trying to run dry i said gee thats odd... and gee thats a puddle of coolant on the bottom of my case...:facepalm:
http://knoxware.org/snowman/2010-11-28%2017.26.31.jpeg
Have to rebuild my loop before I know the total damage but I am pretty sure the wireless n card that filled with coolant is probably dead.

[mDust]

*slowly plays Taps*

I left my computer on when a friend picked me up to go get Crysis the day it released. We were excited to finally get to play this awesome new game we've been waiting on for almost a year. When we got back, there was some jacked up test pattern on my TV (monitor) after I turned it on. Adrenaline shot through my veins. I checked to see if the computer would respond. Then I looked in the side window only to be further disappointed. Somehow the CPU barb leaked a ton of coolant directly onto my GPU and then magically stopped leaking. It had sat like this for about 40 minutes with the power on while we were gone. The GPU was dead and burnt, as was the PCIe slot on the motherboard. I know it was fine before I left as I had just inspected it before I put the side of the case back on for the first time in months.

Anyway, yours was a good machine. Hopefully it will pull through and play another hundred games in its lifetime.

[Oneslowz28]

Oh no! Where did it leak from?

[Snowman]

best guess would be an improperly seated o ring on the mcw360.. if there is such a thing, best possible place it could have leaked at because it did so away from the board and may have only killed a d link wireless card.

[Twigsoffury]

THIS

is why i stick with air cooling.

[mDust]

THIS

is why i stick with air cooling.
If you never take risks, you're never going to have fun!

[Snowman]

I brought my video card down 30 degrees alone with no so great flow, I am thinking of replacing the pump/res for an actual res and pump if I can find one in 1/4"

[Twigsoffury]

If you never take risks, you're never going to have fun!

knocking 3.8Ghz on my X4 955 :P

955/1330mhz on the HD 5770

[mDust]

I brought my video card down 30 degrees alone with no so great flow
This.

knocking 3.8Ghz on my X4 955

955/1330mhz on the HD 5770
You're brushing up against fun, but you aren't quite having it...:)
Imagine what you could do with water!

[Snowman]

Oh and in case you missed my clock speeds go check out my project log on it.. I clocked like a mother before being stopped essentially by my mobo. I also hit my idle temperature on air when overclocked on water.

[Konrad]

Ah ... I've never done waterloop cooling. Scared of something like this happening, of course.

This is catastrophic failure - how often does it really happen? How often does non-catastrophic failure occur? Was this watercooling system always maintained properly or was maintenance somewhat neglected?

Basically - has it happened to anybody else? How often? Why?

[Snowman]

This is not a catastrophic failure, nothing ever got hot enough for my computer to shut down as I have it set up to do if temperatures go over a certain threshold. As far as being properly maintained that may be my fault if the wireless nic is dead because my loop needed drained and flushed and refilled with distilled water as i had a precipitate in mine. Don't start fretting about how catastrophic this is until the actual damage has been surmised. The wireless nic was free and we carry namebrand dual antenna wireless n usb sticks at work.

[Konrad]

lol, maybe "catastrophic" wasn't the right choice of wording.

I've never gotten my feet wet, so to speak, with liquid cooling. I'm still of the mindset that water plus electronics is an utterly bad combination.

[mDust]

Ah ... I've never done waterloop cooling. Scared of something like this happening, of course.
Basically - has it happened to anybody else? How often? Why?

When I used barbs I used to get leaks every now and then. Finally, I got sick of it happening all the time and put rubber cement around the base of the barb before inserting the hose and clamping it on. That worked just fine...until I had to disassemble the loop. That's when I bought compression fittings. They cost more than twice as much as the barbs but have never leaked. It's been more than six months.
As for the severity of the leaks, only the one I mentioned in the previous post was catastrophic. That one cost a lot... The only other damage was a tiny wireless N card that was part of the mobo...it fried, but the only time I had ever used it was to test it out just after I bought it. No biggie. The rest of the leaks did absolutely no damage. Did components get wet? Yeah. But they survived just fine. I'm not a lucky person, but my PC components are.:)
I think people are afraid of liquid cooling because they read horror stories online about worst case scenarios. You don't ever hear about all the minor leaks that don't do any damage because they're not exciting or post-worthy. Also, you don't see the other 95% of water cooling users posting about how their system has no leaks and never has. As long as you hook everything up properly and use quality parts, the risk is tiny and manageable.

[diluzio91]

i can second part of what mdust said. i originally used barb fittings, and i ended up redoing my loop because of the leaks i encountered. my leak was pretty minor, and it was probably from using non swiftech barbs in a swiftech raditor. explination below

dd radiator fitting mount point

_........._
||.......||

swiftech rad fitting mount point

_.............._
| |............| |
..|-|........|-|

if that makes sense... i just doubled the orings and it sealed fine. also, compression fittings are 100000% worth it IMHO. easier to work with, easier for pretty much everything. it helps mine where $.71 each too :D

[Konrad]

Your illustrations are horribly disfigured (um, your html is showing...) but your example is understood, diluzio.

Paying "more than twice as much" for $.71 fittings doesn't sound like a bad investment on a $1-2K computer. They're basically OEM-spec O-rings, yes? Rubber cement sounds goot - does it react/dissolve when exposed to the usual antifreeze mix?

Although perhaps not strictly necessary, but good for paranoid peace of mind, I think conformal sealant (on the electronics) would work. I expect it would act as a bit of an insulator and thus counterproductively heat the components a bit, would this heat buildup be at all significant?

I've got a couple of ancient P4/3.2s I can blow up, they can even overclock quite warm, I think I'll give watercooling a try (after xmas).

[mDust]

Please, please tell me where you got compression fittings for less than a dollar! I paid about $5/ea for mine.


Rubber cement sounds goot - does it react/dissolve when exposed to the usual antifreeze mix?

Although perhaps not strictly necessary, but good for paranoid peace of mind, I think conformal sealant (on the electronics) would work. I expect it would act as a bit of an insulator and thus counterproductively heat the components a bit, would this heat buildup be at all significant?

I put the rubber cement behind the barb so contact with the water would be minimal at best. Why rubber cement? Because there was a jar of it sitting on my desk.

|..X...|<-----water flow was here
|.___.|
|/___\|
|x|..|x|
|x|..|x|<-----rubber cement was where these little x's are.
-------

Also, you should not have any antifreeze of any kind in your loop. Distilled water and a biocide is all you need. There are no pros to putting automotive antifreeze in your loop, but there are cons...such as reduced thermal transfer to and from the water.

[Konrad]

No worries - I'll do more research and start asking the right questions once I can actually dedicate any serious time to my first watercool.

You can use
blocks to better show text diagrams.

Because all the characters/spaces inside them have the same width.

/|\
/\|/
--*--
/|\/
\|/

[mDust]

You can use
blocks to better show text diagrams.

Because all the characters/spaces inside them have the same width.

/|\
/\|/
--*--
/|\/
\|/

That's not fun or challenging!

[diluzio91]

@ Konrad

lol. oops. well anywho, i had a choice between a new radiator or cramming an extra oring on... i went with the oring.. but no tubing leaks which makes me way too happy. also, the compression fittings were normally something like $6-8 each, but i knew a guy (thanks again x88x)...

[x88x]

Paying "more than twice as much" for $.71 fittings doesn't sound like a bad investment on a $1-2K computer.

The $.71 was only because he got them at a massively reduced price. :whistler: Using DD's prices as a good market estimate, good barbed fittings normally go for about $2.50 each (http://www.dangerden.com/store/fb-fat-boy-g-1-4-fittings.html) and good compression fittings normally go for about $8 each (http://www.dangerden.com/store/compression-fitting-1-2-id-5-8-od.html).

EDIT:
Oh, also, since you brought it up, mDust. ;)

Also, you don't see the other 95% of water cooling users posting about how their system has no leaks and never has.
Aside from a couple leaks during testing (ie, the computer was not on), I have never had a leak in almost a year and a half of liquid cooling.

[Konrad]

So I'm figuring the count at 12 pieces (in + out @ pump, rad, reservoir, cpu, gpu, other gpu), plus 2 more for each item added to the loop, plus a bunch more if additional loops are involved. 12 pieces x $8 ($72) still isn't too bad (though $8.50 or $30 would be better, lol) ... I already spend about that much on fans and heatsinks.

Paying a small premium for proper fittings, tubing, blocks, etc seems worth it when weighed against the reduced risk of waterzorching your computer.

How much maintenance do these loops require? Flushing, scrubbing, tightening, inspecting, whatever ... a few hours per week? per month? per year? Biocide was mentioned - do these systems tend to clog up with algea? Should they be kept in the dark or intermittently blasted with x-rays? Do windows, CCFLs, LEDs, EL wire, and UV blacklights actually accelerate algea growth?

I'm starting to get into this idea. I've already considered ways to fab glass-encapsulated thermistors at the shop. I'm assuming glass reservoirs are best, don't want any crazed polymers.

[diluzio91]

im not an authority, but here's what i have observed in my short time watercooling.

rule 1. PLAN EVERYTHING.
rule 2. get a reservoir. (see my worklog if you wonder why) you can get by without one though
3 plan a way to drain.
4 leak test for a while, be sure before you commit to it working, i had a block that leaked no matter how much i tightened it, i had to take it apart to fix it.

as for the loop, i used distilled water, its way too cheap to not use it.
refill the reservoir as needed, the first week or so the level will drop alot as the bubbles work out. I've herd of people cleaning once a year or so. replacing water as needed. the rule of thumb for tubing is to change it as needed, i have herd of it lasting 6 months, i have a friend who has had the same tubing for 2 years. acrylic reservoirs are okay, i recommend bitspower as they use delrin for the parts you screw into, rather than acrylic. For biocide i just have a small silver coil (it looks nice and its pretty small) and you dont have to worry about algae growth. think hard about what you get for tubing size. 3/4 tubing can be stiff and hard to work with, but has better flow rate. I use 6mm tubing and havn't noticed any difference in any of my temps. Good Luck!!

[Kayin]

Actually, on the topic of biocide, I've got a toy that will make sure that nothing will ever grow in it ever. Also seems to have temps on par or so with water, and it looks hella cool.

CJ could tell you a bit more about it if he wanders through here, he's played with it as well. I should have some pics up soon with it in use...

Konrad, if you have some fun ideas kicking around, please share. Overkill is always best.

[Konrad]

A much wider choice of tubing materials is available if transparency isn't a requirement. Strategically immersed sensors would probably serve as well as visual indicators for detecting flow problems, perhaps even better if combined with automonitoring/alarm/shutoff circuits.

Ha, I'd go for nice big fat 1" tubing all around if I could. I expect my choices will realistically be largely dictated by what's available on the blocks, pumps, and rads ... I can fab (or have fabbed) coppers without problem, but there are still dimensional limits which might force thinner tubing.

Does mylar tape affect fittings? I don't know a lot about plumbing, but I've known plumbers (lol, and welders) who swear by mylar threading.

Doing some hours (or days) of dry and wet runs is good advice, thanx ;)

[Konrad]

Ideas? I've never done it before. I've been part of a group that did a complete midel-immersion rig, with a couple of peltiers and a modified aquarium pump to circulate the cooling. I've got a P4 rigged with midel-filled partial-vacuum heatpipes mazed through the heatsink, tremendously efficient (better than the commercial heatpipes we tested against, years ago, plus as a bonus won't kill electronics if it leaks; though still not as much heat transfer as watercooling, so says the math). But no, I've never actually done a liquid cooling rig. You're supposed to be the guys with the ideas <g>

[Kayin]

1" would not offer any gains. Honestly, over 3/8" with today's more efficient designs is kind of a waste.

If you want some custom stuff, though, let me see if I still have the 3d files for some of my designs, you can have all kinds of fun with them. I designed a block with interchangeable cores so you could test different materials without removing the mount from the board. It's not the most necessary thing, but it's really fun.

As for fittings, almost everything uses the G1/4" standard, or 1/4" BSPP with integral o-ring groove. PTFE or mylar tape can actually cause leaks in G1/4" systems by preventing the o-ring from sealing.

[x88x]

I only use plumbers tape on one of my fittings (because it leaks without it...elbow joint at the very bottom of my ~2L loop...), but depending on who you ask, the opinion I've heard most is not to use it unless you really need to, for fear of pieces breaking off and clogging rads or the pump (since the threads are so short on most fittings, you have to cut the tape, so unless you cut it just right...).

For fittings, my current system is and always has used barb fittings, and with the exception of one spot where the tubing got stretched out from being removed too often, I've never had a problem.

For tubing size, like diluzio said, you can use whatever you want as long as you can get fittings, but really you're not going to see much of a difference most of the time. I forget who, but I remember someone here went from 3/4" ID to 1/4" ID a while ago, and their temps didn't change at all.

As for cleaning, I flush new rads before I put them in the loop and clean blocks when I lap them, but other than that I don't touch any of it. The liquid I use (Feser 1 blue premix) has an antifungal (or something along those lines) additive, but if you're using distilled water, yes, it is recommended to use an antifungal additive or a silver coil/strip/etc.

[mDust]

So I'm figuring the count at 12 pieces (in + out @ pump, rad, reservoir, cpu, gpu, other gpu), plus 2 more for each item added to the loop, plus a bunch more if additional loops are involved. 12 pieces x $8 ($72) still isn't too bad (though $8.50 or $30 would be better, lol) ... I already spend about that much on fans and heatsinks.

Paying a small premium for proper fittings, tubing, blocks, etc seems worth it when weighed against the reduced risk of waterzorching your computer.

How much maintenance do these loops require? Flushing, scrubbing, tightening, inspecting, whatever ... a few hours per week? per month? per year? Biocide was mentioned - do these systems tend to clog up with algea? Should they be kept in the dark or intermittently blasted with x-rays? Do windows, CCFLs, LEDs, EL wire, and UV blacklights actually accelerate algea growth?

I'm starting to get into this idea. I've already considered ways to fab glass-encapsulated thermistors at the shop. I'm assuming glass reservoirs are best, don't want any crazed polymers.

There is an initial financial premium to start water cooling, however, everything other than the blocks can be reused in any system in the future. Your loop requires an initial testing each time you disconnect it and reset it up. Bacteria and many types of algae don't need light to survive and will love your warm water, so a chemical or silver biocide is necessary. If everything is setup properly you shouldn't have to mess with it but maybe once or twice a year to drain and refill it. To be honest, I've left mine alone for much longer than a year due to mixing in a bit of super concentrated colloidal silver. It kills anything that even thinks of growing in there.

Anywho, how's the repair work on your loop going, Snowman?

[diluzio91]

For tubing size, like diluzio said, you can use whatever you want as long as you can get fittings, but really you're not going to see much of a difference most of the time. I forget who, but I remember someone here went from 3/4" ID to 1/4" ID a while ago, and their temps didn't change at all.

lol. i was one of those. i actually got more of a temp drop by reorganizing my loop TBH. -3 on the gpu and only +1 on the cpu works for me

[Twigsoffury]

You're brushing up against fun, but you aren't quite having it...:)
Imagine what you could do with water!

WHAT I CAN'T HEAR YOU OVER HOW INCREDIBLY BAD A@# AND LOUD MY FANS ARE

5_Zp70eC670

(that is actually my computer)


FYI copper is naturally anti-microbial, Anti-fungal

http://www.copper.org/antimicrobial/homepage.html
http://www.antimicrobialcopper.com/

Staph infection medications have copper oxide in them, As well as roofing shingles, copper in shingles prevents that mildew look when it ages. Alot of hospital equipment and surfaces (like door handles and push plates) are also copper if you notice.

a copper Oxide in your lines mine actually increase cooling while preventing microbes from developing inside your lines. i'm sure you would need a agitator in your res though.

Dunno just a idea.

[Konrad]

So ... wouldn't a small concentration of isopropyl alcohol function as a toxin and (given the water volume) not be a fire hazard?

[Kayin]

Breaks a lot of the plastics used in mainline watercooling. Really bad for them.

[x88x]

Yup, plexi is not happy with isopropyl. :( Cracks all over the place.

[Twigsoffury]

Yup, plexi is not happy with isopropyl. :( Cracks all over the place.

what about polycarbonate?

[Snowman]

what about polycarbonate?

HATES ISO ask my last pair of lenses.

[Kayin]

Normally OK, depends on the formulation.

[Konrad]

Hey, isopropyl was just a thought - nobody ever does it, I figured there must be some reason why. Largely because everybody likes transparent plastics, duh, shoulda known.

Rubber, metal, and glass seem like the winners for me. Heavier, basically same cost (fab all the important parts at the shop anyhow), no thermoplastic weakness.

I'm not familiar with the pumps ... has anyone done takeaparts? The question is: which parts (and which material types) inside the pump are directly exposed to the cooling liquid? I can't think of any possible non-contact approachs, short of maybe a liquid metal and magnetic fields (which would probably be not good for high-frequency logics).

[x88x]

The most common pumps these days are Laing DDC pumps. In those, only the pump body and the impeller touch the fluid. There's an exploded view on the manufacturer's website:
http://lainginc.itt.com/LG-pump-DDC-Series.asp

[Konrad]

Squick, ABS plastic impeller. Without CAD specs, so metal fabbing this part would be a bit more involved than I really like (ie, not CNC and beyond my tool skills, therefore involving a real machinist).

I'm thinking more along the lines of what wikipedia calls induction motor (http://en.wikipedia.org/wiki/Induction_motor) liquid ring (http://en.wikipedia.org/wiki/Liquid_ring) pump circulating an ionic liquid (http://en.wikipedia.org/wiki/Ionic_liquids) (that is, a water-solute electrolyte salt) under partial vacuum (to change boiling point thresholds). Automotive oil pumps (of which there are plenty of CAD specs in shop) can be adapted to this, I think. Two identical units wouldn't cost more than an afternoon of my time and a couple pounds of metal. Plus assorted rubber rings, grommets, and electrical components.

[x88x]

Sure, maybe it's possible, but...why? What would be the benefit?

[Konrad]

Oh, the usual benefits of DIY:

Complete control over component selection, materials, tolerances, and costs. Plus the "uniqueness" factor; the elixir of modding, simply to do something different or perhaps to improve the "out of the box" stuff that everybody else uses.

The only "benefit" I see to purchasing a COTS unit is convenience. Higher prices and fixed tradeoffs (restrictions) come with those units ... maybe these restrictions were engineered into them for very good reasons, maybe they're just cheap or arbitrary. Won't ever know without studying them or building your own.

[Kayin]

Well, as far as weight alone, some of those blocks approach the limits of what gravity can deal with by themselves, adding more weight would be rather ill advised. Hence why I've been looking into such as the thermocapacitive plastics and such...

[Konrad]

Blocks would all be machined copper. I might even buy them stock if they're inexpensive.

I was thinking more about the pump, res, plumbing, and fittings. There are numerous good rubber-thermoplastic options available (and a wider range of "safe" liquids to use in them) if you don't require transparency. Why not use the same boring black and durable industrial pipes and fittings that are used in your freezer?

[x88x]

and a wider range of "safe" liquids to use in them

The problem with using different liquids is that until you start messing around with exotic stuff (I'm looking at you, Kayin :P ), the thermal transfer and capacity isn't going to beat something that's at least mostly distilled water. There are plenty of reasons besides cost why it's the most commonly used fluid.

[Konrad]

That is assuming normal pressure. With a little carefully calibrated vacuum you can make liquids vapourize at any particular temp you desire. I think the trick, for maximum liquid cooling, is to suck to just above the point where you're getting phase-changes (because you don't want to mess with mixed liquid/gas systems and stupidly loud compressors). Experimentation would be in order. Industrial fittings can be far more robust (leak and maintenance free) than wimpy modmarket gear.

[Kayin]

Well, certain block design also need to be clear because with the fluids we run, some have a chance to gunk up and stop working.

You're right about off-the-shelf stuff, but the mainstream stuff was not just tossed out with a moment's thought.

Perhaps experimentation IS in order...

[Konrad]

Blocks can seize up? Crap.

Crap.

[x88x]

Blocks can seize up? Crap.

Crap.

Some additives can break down, lack of some additives can cause buildup, mixing certain metals can cause corrosion. As long as you steer clear of all of those situations, you should be fine.