• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Helsyeahs "Plunge" into Water-Cooling ***PICS***

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Helsyeah

Made the Cookies Crumble
Joined
Jan 27, 2004
Location
ID
What follows is a summary of my recent venture into water-cooling. First a little background on why I decided to jump into water-cooling in the first place:

Oh, I should warn you that I tend to be on the long-winded side :D. While I dont have as many pics as I would like, they should tell the story pretty well by themselves (in case you guys get bored with all the text :cool:).


So all was good and happy in the world of air cooling until I joined the ranks of Dual Core Opteron owners and the small furnace that these dual cores can become at elevated voltages and clocks. At first I thought I would be satisfied with air-cooling, I invested in a Scythe Ninja and felt confident that I could achieve a respectable overclock and still keep my load temps under control.

I received a 0551UPMW Opty 165 from Monarch and was able to get a decent 2.7 GHZ w/ 1.425v and a 48 deg C dual prime load temp:

Click for full size​


On a cool morning I could also do 2.8 w/ 1.525v and a 53 deg C load temp. After initial testing I did some burning-in and saw a marginal decrease in the amount of voltage it took to hold stable @ 2.7 but I only was able to shave off ~.0125 v on average. (Side note, all three of the dual core opty's that I have tested have shown some, but very limited benefits from burn-in for a week... but this is topic for a different forum :)). My testing seemed to indicate that I topped out @ about 2.85 with 1.525v with load temps pushing 56 C.

After messing around with burn-in I actually took a week and played some games with my new and shinny computer (see 1st rig in sig for more specs). I was very happy with my setup and had gotten my tweaking fix satisfied...

But being the addiction that tweaking computer hardware is, I soon resumed my quest for pushing my components to the bleeding edge. Of course there had to be a bit of drama involved, and so the later weeks included the following series of events:

  • more burn-in
  • removal of the the IHS from the 0551UPMW 165 in an attempt to lower temps
  • chipping of a SMD cap on the 0551 and by all indications partially crippling it, OC maxed out @ 2.3 and couldnt use any voltage options besides auto in bios
  • completely freaking out and in a panic ordering a new Opty 170
  • researching @ XS and deciding to completely REMOVE the chipped cap right after ordering the replacement 170
  • result: completely reviving the 165, all was good again :shrug:
  • receiving/testing the 170 (turned out to be a 0550VPMW) and finding it to clock about the same as the 165, but to run WAY hotter, maxed out @ 2.8 w/ 1.5v and a scortching 57 C load temp.
  • not being willing to attempt another IHS removal I sold the 170 on the forums here
  • finding some additional play money, I bought a new 165 from monarch, taking advantage of the coupon deals again
  • receiving said 165 got lucky with a 0608MPMW, one of the better clocking steppings available now
  • getting ballsy again and removed IHS from 0608, went off without a hitch, just had to be more careful
  • initial testing showed that I could get to 2.8 w/ 1.4 v but was limited by temps at 2.9 and above

After playing with the 170 and finding it to be even more of a furnace than my 0551 165 I started seriously considering the jump to water cooling. Then when I receive my 0608 165 and topped out @ 2.8 due to temps, and since no amount of reseating and reapplying AS5 was helping lower the temps, I knew something had to be done. The 0608 stepping seemed to be consistently hitting 2.95-3.1ghz on water. (and water cooling is fun right?)

Play money becoming available again and driven by the need to tweak something new (say for example, a water cooling setup) I started the gears in motion to switch from air cooling. But there was a small problem. I was hesitant to initially jump right in because from all the research I had done indicated watercooling would require more maintanance than your typical air setup, and had the potential of growing new and interesting life-forms in the loop. In addition, for the last several years I have been trying to focus on making my computer as low-maintainance as possible. While I have no problem with various biology experiments, for some odd reason I dont like the idea of conducting them in my computer and switching to a higher maintainance setup didnt sound exactly like what I might want to do.

So, back to more researching/digging, I knew that various additives and such would help avoid growing stuff in my loop, but what could I do to lower the potential need for increased maintainance?

Then, after more research at the Procooling forums, I stumbled across this thread where using opaque tubing could potentially extend the period between needing to flush coolants, and the chance of developing various life-forms growing in the loop.

Good enough for me, I finally took the plunge, no pun intended, and ordered up my water cooling components.
 
The Project:

I had basically three goals for going to water cooling:

  • Knock my load temps down to the 40ish deg C.
  • Hopefully get to 3.0 ghz with my 0608mpmw.
  • Move to quiet 90 and 120 mm fans and bring the whole computer down to a very quiet hum if possible and still maintain good cooling during a full load.

Below is the list of components that I decided to use to make my setup.

Heatercore: Fedco 2-150

First I already had a Fedco 2-150 heater core that I had bought with the intent to eventually go to water cooling on my old 2.4c.

It is 6.5"x7.75"x2" and has 5/8" inlet and outlet.

2-150_399022_Heator_Core.jpg

This paticular core can be found cheapest at Napa for $23:

http://www.napaonline.com/MasterPag...HC&PartNumber=6603000&Description=Heater+Core

When I picked mine up two years ago it was cheapest at Autozone for a nice $16 investment.

Case: Lian-Li PC-201B

I also wanted to get a new case that would give me lots of room for my new project. I decided on a Lian-Li PC-201B:

11-112-109-08.JPG


http://www.newegg.com/Product/Product.asp?Item=N82E16811112109

For the most part this case appears to be almost identical to the v2000/2100 series but does not have wheels and has square corners vs the rounded & rolled corners found on the v2000/2100 cases.

The space for the lower hard drive racks was the perfect size for my heater core to be mounted vertically and the PC-201B would still fit under my desk, while the v2000/2100 series were just a little too tall (although in hind-sight, removing the wheels might have solved that problem).

Pump: Swiftec MCP350

By far the two most popular pumps seem to be the Swiftec MCP655 & MCP350. I went with the MCP350 for the simple reason that most seem to consider it to be quieter than the MCP655, and could later be modded to provide flow rates comparable to the MCP655.

Cpu Block: Swiftec Storm

For the time being I plan on having only a cpu block in my system. So with this thought in mind I decided to go with the Swiftec Storm. The Storm seemed to offer very good performance at the price of being fairly restrictive to flow. At this stage though, since I will only have the CPU block, I thought that the high restriction wouldnt be a big deal. Now if i decide to add a GPU block later the amount of restriction in the system may become an issue, but I won't worry about crossing that bridge until I get there.

Tubing: Tygon R-3400 (Black)

Since I wanted to use opaque tubing in hopes of adding another line of defense against little nasties growing in my loop and such, I dug around and stumbled across references to R-3400 Tygon tubing (Black). Browsing @ McMaster finally turned it up.

10 feet seemed like it would be plenty to account for any screwups I might make and still have some left over.

Tygon R-3400 black 1/2" ID 3/4" OD tubing: 10 feet
McMaster P/N 5255K26 on page 100.

Misc Hardware & Fittings:

While I was getting the tubing from McMaster I thought it would be a good idea to stock up on all the fittings and hose clamps I would need. I picked up Stainless worm hose clamps and both a Nylon and Brass set of fittings in case I wanted to go with the all-metal look on the loop. The whole mess included:

  • 1/2" Tee's, brass & nylon
  • 1/2" to 1/2" barbed couplings, brass & nylon
  • 1/2" to 3/8" reduction couplings, nylon-intended for connection to MCP350 3/8" inlet and outlet
  • 1/2" Tubing to 1/2" NPT (brass) female pipe thread coupling (improvised fill-port)
  • 1/2" NPT brass hex plug for fill port
  • 1/2" tubing to 1/2" NPT Pipe thread Tee (straight 1/2" tubing barbs with the Tee being a male 1/2" pipe thread (nylon, couldnt find brass). Intended as an easy drain port.
  • 1/2" NPT Female brass pipe cap for drain port

The fittings & clamps total was a bit on the expensive side (~$40-50) due to brass fittings.
 
Initial Build WorkLog:

I ordered the pump and storm from the www.heatsinkfactory.com along with 4 Panflow 92mm medium speed fans. I chose 92mm fans because the heatercore was only tall enough for two 92mm fans. With 4 I figured I could do a push/pull configuration with a custom shroud. (I ended up doing something completely different, read on)

The case was ordered from Newegg. Everything came in and it was off to the races.

The first thing I did was to set up the entire loop and test it out in the bathroom (I'm renting, and had absolutly NO say on the pink paint scheme...) Mostly I was concerned about the pump working right and that none of the other components had any weird leaks.

Click for full size​

All worked well, water seemed to move how and where it was supposed to once I got the pump properly primed.

So, on to the case, fitting components up and making any mods necessary.

I took both sets of hard drive racks out of the bottom to test fit-up with my heatorcore. I played with a couple different positions and arrangements for the rad, but in the end I decided on the configuration shown below. There was 1/4" of space between the top and bottom of the heatercore and the case. I used some closed-cell packaging foam to fill the gap.

Click for full size​

This would let me keep the rear-most set of hard drive racks and looked like it would still provide adequate air flow for the rad. The hard drive racks have serveral series of 7/8" holes. Luckily I was able to use one of these holes to run one of the tubes from the rad through (the second tube would miss the rack all together. The tube that missed the rack also perfectly lined up to be run through the cable opening in the case. The tube that ran through the hard drive rack would require a cutout though. I made a quick template and eye-balled it into about where I thought it would need to go:

Click for full size​

After getting the cutout sketched into place I turned my attention to deciding how do my fan setup. As was mentioned above I initially thought about doing a push-pull setup with two sets of 92mm fans and a shroud. Then while fitting up the rad I stubbled across a different option. In the front of the lower level of the case is a nice fat 120mm fan setup with a filter. This might just do the trick by itsself as an air inlet. Then playing around with the 92mm fans I discovered that arranging them in a 2x2 grid would almost perfectly cover the heatercore, and still would be narrow enough to fit inside the case. The 2x2 setup seemed like it would provide a good amount of air-flow mounted in the pull position as well. So off to my 3-d modeling software to create a mounting bracket for the fans:

Click for full size​

The bracket will be made from some 12ga aluminum that I have laying around (actually I grabbed from my dad's shop last time I went for a visit :)). Now, being somewhat impatient, and not having a press-break available to make the bracket myself, I gave the drawing to a guy at work that offered to do it, and immediately started working out a substitute bracket from some 1/4" black styrene stock I could get my hands on. I followed the same basic layout from the bracket that I designed to be made from the aluminum, and went to work with my roomates jigsaw and cordless drill.


A couple hours later, and taking the time to cutout the slot in the case for the tubing I ended up with the following arrangement (excuse the foot shot in the first pic):

Click for full size​

Click for full size​

I planned on sealing the space between the fans and the heatercore with some more of the packaging foam to create a psudo-shroud. This is shown a few pics down and actually works very well.

Things were looking good, and my girlfreind was over hanging out, so I took a break for the rest of the day, and well, the rest of the next day too... Eventually I got the aluminum mounting bracket from the guy at work, but the plastic one seems to work so well that I plan on waiting until I needed to completely tear apart the loop to add the aluminum one in (mostly because thats what it would take to add it.)

So, all was looking good, I hooked up the rest of the water loop, cut my tubing to what I thought I would need and checked fit-up with everything:

Click for full size​

Everything looked like it would work out great, so I exteneded my T-line and started filling the loop with distilled water + 1 drop per quart of iodine. I leak tested/blead the loop for 26ish odd hours and all seemed good to go, other than the occational air bubble that I could hear pass through the pump. (I later discovered that the T-Line is best located on the inlet to the pump to help speed the bleeding process up)

Click for full size​

Click for full size​

So the time came to bite the bullet and add all the hardware to the case and get things up and running for real. I bought a DFI Expert in the classifieds here as a backup board and dropped it in to run my initial testing. Initially I dropped my 0551 Opty in just to make sure that everything worked right (didnt want to hose my 0608 right off the bat). I made sure everything was plugged in except my hard drives, double checked the water level and hit the power button. And....
 
Last edited:
All was good!! I loaded up the bios and was greated with 23 C idle temps, SWEET!!!.

Click for full size​

So on to several days of testing:

  • Tested 0551 Opty initially to find max load temp on expert. 2.8 w/ 1.5 v loaded at 36 Deg C. Note that the Expert seems to read temps low by ~6-8 deg C. 2.9 was impossible on air, but watercooled it now looked to be very doable @ ~1.6v
  • Replaced the 0551 with the 0608 in the Expert.
  • 2.9 w/ 1.425 passed 2+ hrs of prime, load temp @ 35 C
  • 3.0 took 1.525 to start getting stable, but dual prime would consistently fail @ 2-3 hours, load temps only @ 38 C
  • Played with the HTT mult, dropped it to 2x and raisd the LDT voltage from 1.2 to 1.3, and got prime to be stable for ~4-5 hours, but still failing.
  • Guessing that the Expert has issues with hight HTT, played around and found that around 325ish, instabilities started cropping up.
  • Switched Expert for the DFI SLI-D that I had run before and had not noticed any problems @ high HTT.
  • Prime started doing great until PWM temps rose to 57ish C, then prime would fail.
  • Added 80mm fan right over PWM sinks and....

Click for full size​

BAM!!!, 3 ghz, dual prime stable baby!!

I spent another couple days tweaking my ram, getting that completely stable and all was good to go.

Sweet! mission accomplished, Load temps @ 40ish C, 3.0 ghz dual prime stable, and the computer was reduced to a fairly quiet hum, all three goals met...

But then thats not the end of the story quite yet. I spent some time tweaking my ram, found some minor instabilities that I ironed out, and then turned my attention to the GF for a few days....


Modifications WorkLog:

Of course I couldn't leave well enough alone :D.

After most of the air had finally (2 weeks later) worked it's way out of my loop I started noticing odd silver particles in my T-line. I checked things out, all seemed good, but the silver particles had me worried, and my water smelled like solder... Did som research and oops, found out that I really should have flushed my heater core with vingar or alcohol, which is where the solder "dust" for lack of a better word, was coming from.

On top of the dirty water, I really didn't like how it took 2 weeks to completely bleed the loop (and granted, my T-line definately needed to be higher in the loop, and located on the inlet to the pump). So stopping by Home Depot I picked up all the supplies to make a "toolless" PVC res, plans found here.

Also, to avoid any weird corrosion that might try to happen I picked up some Hyper-Lube Water additive and mixed it into my distilled mix @ about a 5% ratio ( initially thought I wouldnt really need any additives besides a bit of Iodine). Two lunch-breaks later and a full flush and refill of my loop and heres what she looks like now (only took 10 mins of tilting case to bleed the air out, gotta love that vs the 2 weeks it took before):

Click for full size​

Temps are still good, and for now the loop still looks clean, time will tell on that though.

I still have one more part of the project to finish and that is to make some kind of duct to bring fresh cool air to my PWM area on the mobo. I can either put an air inlet in a couple of my 5-1/4" bays and duct air right to the PWM area that way, or fashion a duct to bring air in from the inlet in the upper right hand corner of the side panel:

Click for full size​

Right now I'm leaning toward ducting from the side panel because I wouldnt need to fab up something nice-looking to mount a 120mm fan in the drive bays, and the ducting shouldnt be that crazy. I'll update once I get it figured out.



I do have to say that the whole water-cooling move has really been fairly painless. A bit of research, pre-planning seems to make things go really smoothly. I have yet (knock on wood) to have a leak, and everything really seems pretty easy to take care of. Overall I am extremely pleased with watercooling in general.


So, thoughts, questions, comments, what do you guys think?


Edit:

UPDATE: See watercooling the VID cards below!
 
Last edited:
lol, costs? too much :)

Heatercore - $16
Case 201B - $270
Swiftech Storm - $75
Swiftech MCP350 - $71
Panflo FBA09A12M1A - 4x$11 -> $44
Tygon R-3400 Tubing - 10ft -> $39
Brass & Nylon Fittings - $45ish (jeeze, brass is expensive)
Parts for Reseviour - $16

Total = ~$576 + shipping...

Ouch, now I know why I never sat down and calculated it all out before :eek:

But without the case it's a pretty reasonable $205ish for just the components.
 
Helsyeah said:
lol, costs? too much :)

Heatercore - $16
Case 201B - $270
Swiftech Storm - $75
Swiftech MCP350 - $71
Panflo FBA09A12M1A - 4x$11 -> $44
Tygon R-3400 Tubing - 10ft -> $39
Brass & Nylon Fittings - $45ish (jeeze, brass is expensive)
Parts for Reseviour - $16

Total = ~$576 + shipping...

Ouch, now I know why I never sat down and calculated it all out before :eek:

But without the case it's a pretty reasonable $205ish for just the components.
Bah, my pump costs more than your case :eek: Although, I didn't pay retail :D
 
nikhsub1 said:
Any plans to watercool the vid?? Do it. DO IT. DO IT NOW!!! Looks good man, nice clean job :D

Ahh man you do realize that your holding a full glass of Guinness :beer: in front of a extremely parched alcoholic by bringing up watercooling the vid cards...

lol... for the time being I'll be just running air cooling actually. I just did a step-up on my EVGA 7800 gt to an 7900 gt and picked up another 7900gt to run in SLI... we'll see about slapping water cooling on them though, god knows that I'm weak when it comes to these projects and they tend to start get out of hand :D.

Edit:

Ah, I have been eye-balling the RD-20 or RD-30. Should I go and cool the vid then I really should jump the pump up to an RD-20 or RD-30, and while im at it, upgrade to a better rad, and and.... god someone stop me before I break out the C-card again... ;)
 
Looks good and clean. That black tygon is something I haven't seen before.

You know you are going to watercool those videocards so you might as well break down and do it when you get to it.

Watercooling and computer tinkering in general can definitely be addicting (in a good way that is). I have my main system torn apart right now changing pumps and some tubing routing myself.
 
It looks like you have the pump discharging right into the CPU water block? If so I would reroute that so the water coming out from the rad goes into the CPU water block instead. Nice work!
 
Thanks for the comments guys. I spent alot of time really trying to track down black tubing, and only found like a handful of references to the R-3400. A couple here, a couple at the Bit-Tech forums and one or two over at Pro-cooling.


Owenator said:
It looks like you have the pump discharging right into the CPU water block? If so I would reroute that so the water coming out from the rad goes into the CPU water block instead. Nice work!

Man, now I'm starting to rethink my whole routing... hmm... I could just add the GPU blocks to the loop as well, if I decide to reroute everything. My other 7900 gt will probably be here in the next week or so. I'll have to drain the loop anyway to reroute and adding those two extra blocks would be easy then, hmmmmmmm.... :D

Ah man... and I wanted to pay my C-card OFF this month... ;)
 
Owenator said:
It looks like you have the pump discharging right into the CPU water block? If so I would reroute that so the water coming out from the rad goes into the CPU water block instead. Nice work!
That's a myth, it doesn't matter if it comes from the pump or rad the temps will be the same.
 
thegreek said:
That's a myth, it doesn't matter if it comes from the pump or rad the temps will be the same.

It is true that the steady state temp of the system will be the same but there is still a lower temperature out of the rad then into the rad. If it comes out of the rad at one temp then is heated by the pump before going into the cpu the water temp will be higher than out of the rad. The total heat added is the same and the rad outlet inlet temps may not change but as the water flows through the system the temp does go up. Swiftech's instructions label the flow as out of the rad to the cpu then gpu then pump then rad. That said, in practice you might not be able to tell because the temp increase is very small from component to component.
 
Owenator said:
It is true that the steady state temp of the system will be the same but there is still a lower temperature out of the rad then into the rad. If it comes out of the rad at one temp then is heated by the pump before going into the cpu the water temp will be higher than out of the rad. The total heat added is the same and the rad outlet inlet temps may not change but as the water flows through the system the temp does go up. Swiftech's instructions label the flow as out of the rad to the cpu then gpu then pump then rad. That said, in practice you might not be able to tell because the temp increase is very small from component to component.

QFT,
The myth that I believe you were thinking about, thegreek, is that the flowrate is not constant (i.e. lower restrictions in selective places) but in any system with the same pressure drop and the same pump (closed system) the flow rate will be constant throughout the system... or something like that...
 
Last edited:
Owenator said:
It is true that the steady state temp of the system will be the same but there is still a lower temperature out of the rad then into the rad. If it comes out of the rad at one temp then is heated by the pump before going into the cpu the water temp will be higher than out of the rad. The total heat added is the same and the rad outlet inlet temps may not change but as the water flows through the system the temp does go up. Swiftech's instructions label the flow as out of the rad to the cpu then gpu then pump then rad. That said, in practice you might not be able to tell because the temp increase is very small from component to component.
The diff in temp between in and out of the rad in his loop is probably round .3C or so at most. However, the additional PRESSURE to the block being right after the pump may benefit the system more than the .3C temp drop and less pressure.
 
gulp35 said:
QFT,
The myth that I believe you were thinking about, thegreek, is that the flowrate is not constant (i.e. lower restrictions in selective places) but in any system with the same pressure drop and the same pump (closed system) the flow rate will be constant throughout the system... or something like that...
Flowrate is constant throughout the loop, PRESSURE is NOT.
 
nikhsub1 said:
The diff in temp between in and out of the rad in his loop is probably round .3C or so at most. However, the additional PRESSURE to the block being right after the pump may benefit the system more than the .3C temp drop and less pressure.

I agree that the temperature difference is not goin go to be great. It still exists but we are perhaps splitting hairs. Overall reducing the length of tubing and the per-component pressure drop along with using larger diameter tubing have a more dramatic effect.
 
Back