Current Project: Dual Phase Change - DH800

[ciscokid454]

Ok, this is a preliminary post.
I've been spenting time piecing things together. With my schedule i'm not afforded the luxury of setting things up all at once sit down. This is also and major undertaking retrofitting the custom WB that is being used as the waterchiller.
The principle for this project was to take a pair of old Kryotech SuperG's and mount them to cool a dual Xeon system running on an Iwill DH800.
First off, the components.
Motherboard:Iwill DH800
Cpu's: Pair of 2.4/533/512
Pair of 1.6/400/512 LV
Pair of 2.8/800/1024 ES
Memory: Don't know, i've got two sets of twinx pc3200 and a set of twinx pc4000.
Video: Either 9700pro or 5900u256, Most likely the 9700pro as it already has the WB mounted and i'm being lazy with it.
Storage: Lsi Megaraid, either the 320-2 or enterprie 1600I also have the 150-6 i might try using in here.
Box is going to be a gaming rig, as i've got the x5dpe-g2 with 8Gb of ram for a server and the Ic7-max3 as a gaming machine as well, (max3 will probably go away if this is succesful)
Ok, no i run into a problem.
I'm walking through the final construction phase of the installation.
Mounting the hardware and planning the tube routing.
The case is codegen from newegg.

Closeup of the front.

Side with the two kryo cooling heads and the first problem i'm finding(where/how to fit them)

Back of the unit showing the old PC Power & Cooling ATX600 Full tower psu i'm going to use for the hdd's.

MOBO side, showing the BIX2 that i'm probably going to remove, don't think i need it.

Closeup of mobo side where kryo tubing is coming into the bottom of the case to the other chasis side.

CPU/mobo closeup

Dual cooling heads closeup

Current DD Dual bay res showing where it might go.

Same, from inside showing where tubing run will be

The main problem is mounting the dual cooling heads to each other, and I must apologize, I forgot to get a pic of the WB chiller. I'll get one of it tomorrow, and post it.
I'm rethinking the placement of the coolant loop.
My initial thought was to have the cooling heads on the drive side with the res, now looking at the placement, space and tube routing, i'm thinking of mounting it all in the mobo side, and getting a 3.5" bas res from DD. It would make tubing alot easier and functional, but give me a tighter space to work in.
Damned if I do and damned if I don't on this one.
The dual cooling heads mount flush with a waterblock I made for a previous system, where i used one of these kryotechs to cool an Abit NF7-S 2.0 up to 250fsb.
I'm planning on cooling the NB and GPU inline with the cpu's.
Cpu's will use y adapters to split the fresh cool water straight to them first in parallel(spelling).
then to the NB, followed by the GPU, then back to the pump, up to the res and back into the chiller.
Anyway, that's it for the moment, if anyone has any idea's or thoughts, please let me know, and Merry Christmas everyone.

 

[KoZLop]

Expensive. Look's like a good project soo far!

 

[ciscokid454]

Expensive. Look's like a good project soo far!

only expensive part is the case, mobo and new cpu's atm, as i've got most of the hardware from other machines.
Hdd's are coming from machine in sig. I picked up the kryo's cheap off of ebay.
The two cooling heads are going to mount up to the chiller WB, which again i don't have a pic atm.
This was just a preliminary hardware fitting for the case and the kryo's.
I'm going to take the case back off, and attempt to mount the kryo's to the chiller sometime tomorrow hopefully.
They basically sandwich the chiller block and are bolted to it.
Don't really know what to expect with this system.
Just hoping for a decent overclock. I'm more than likely going to use the 2.8's as they are emt64/800fsb chips and should be the best performers(plus the 64bit will be nice).
Who knows, the 1.6's might blow me away, more than likely though the 2.8's are going in and staying.
Once I get it all situated, up and running i'll of course post pics/stats in here for you all.
BTW, on the kryo's they are r404 native, being superg's.

 

[cornelious0_0]

This is something I wanna follow through on and see how it turns out, but I just find it kinda funny how you hit enter after every sentence in that last post......was there an actual reason?

 

[ciscokid454]

This is something I wanna follow through on and see how it turns out, but I just find it kinda funny how you hit enter after every sentence in that last post......was there an actual reason?

N
T
I
K
O

 

[gclg2000]

Just looking at first hand so far.....you making a water chiller right????

And is that a radiator i see in those pics??? Why are you using a radiator that will only warm up the chilled liquid and be useless really. Your blowing "heat" accross it, since the liquid inside the lines is colder than the room air, your defeating the purpose.

And if your going to be using a chiller, you need to insulate if your going to pass the dewpoint of the room air, otherwise you will get alot of condensation build up.

How are you going to "chill" this liquid....with a fridge or something??? Or an evap from a A/C unit??

Correct me if i'm wrong here, but that is what it appears you to be doing/

 

[ciscokid454]

Just looking at first hand so far.....you making a water chiller right????
And is that a radiator i see in those pics??? Why are you using a radiator will only warm up the chilled liquid and be useless really. Your blowing "heat" accross it, since the liquid inside the lines is colder than the room air, your defeating the purpose.
And if your going to be using a chiller, you need to insulate if your going to pass the dewpoint of the room air, otherwise you will get alot of condensation build up.
How are you going to "chill" this liquid....with a fridge or something??? Or an evap from a A/C unit??
Correct me if i'm wrong here, but that is what it appears you to be doing/

radiator is in there in case i decide to switch to water and to test it, see what kind of difference i get with air, water and waterchiller.
Yes, i'm doing a waterchiller.
I'm using two old kryotech's, they were used inKryotech SuperG systems to get 700Mhz Athlons to do 1Ghz. Now they'll be chilling the water to see how cold can get it.
The radiator won't be in the finished Chiller System
I haven't even finished the current mods to the system yet to start applying the insulation to the lines. That will be done, i'm more worried about getting everything finished first, before doing that.
I have insulated the board itself, as well as the cpu's and sockets, and vid card, and mem slots and anywhere else condesation can happen.
Anything else i'm forgetting to mention?

 

[Zeratul1320]

you should not use an acrylic res. such as that bay one, assuming that is what it will be used for since the acrylic will crack at low temps.

 

[ciscokid454]

Ok, here's an update with pics:
This is both coolign units sitting side by side, case off, I've got to reapply some insulation around the one with bare tubing, I've also got a mounting issue with them both.

This is the chiller block, nothing special,both cooling heads mount flush up against it.
I've got a mounting problem with the cooling units and this block like i said, i've got to flip one of the mounting plates around on the cooling unit.

Differnet angle of the chiller block.

And yet another angle. yes, the block looks messy, i'm going to sand and polish it down some before mounting it. It's been through some fun past couple weeks getting retrofitted with these "all thread" rods.

The two power supplys i'm considering using. Ones a new PC Power & cooling 510, the other is an old 600W Full ATX from a PC Power & Cooling solid steel tower unit.

600 full tower and 510 showing how I'm wanting to to be seated, once everythings fitted.

These are the new mounting plates for the Codegen case and the old one for the 600. I'm trying to figure out the cleanest easiest way to attach and mount them both. (simplest thing would be to use a different psu than the 600:-p, I've never had issues with the unit though and want to use it.)

Original Mounting plate for the 600 Full tower with the 510 next to it.

New mounting plate on top of old, with both psu's sitting beneath.(tight fit) :-(

The sheet metal i've seen is not as thick as the metal for the psu mounting plate is.
So i'm going to figure out how to make them all fit together.
Anyone got any idea's on the psu thing?
Oh, back to the original problem i was talking about.
The two cooling heads, are basically reversed of one another. The original mounting plates/enlocures are made out of some type of insulation material. Easiest thing I can think of is to cut the one mounting plate so that i can remove it, turn it 180 degree's and then reinstall it. The chiller should then bolt right back up to both cooling heads with no problem.I'll get some more pics of this tomorrow. Hoping to have it all together tomorrow for the most part. Don't think i'll have a solution for the psu mounting plate that quickly.
EDIT: I forgot, i'm also trying to keep the original look of the kryotechs, so i'm modifying the front facia to join together. *(pic tomorrow...sorry) So looking at the front you'll see two kryotech facia plates, and side base plates. I want to join the two plastic facia pieces together, not sure of the best bonding material for that though. (brother suggested jbweld). Each piece is white, so i'll be giving them all a nice coat of black paint before it's all said and done as well
Cisco

 

[ciscokid454]

you should not use an acrylic res. such as that bay one, assuming that is what it will be used for since the acrylic will crack at low temps.

The Acrylic tops on the danger dens will be a problem then as well won't they.
*sigh* I was just reading an article/thread about this the other night.

 

[gclg2000]

as i look on even more. Why are you even doing this water chiller.. Those are 2 Vapor coolers ready to go. Why don't you just slap those right onto the cpu. Mod them somehow so they will fit or attach new evap's and recharge the loop(s).

 

[Zeratul1320]

that and now that I look at it the way you plan on chilling your water probably wont work well either. I saw an article on a similar idea, let me find the link, http://www.procooling.com/articles/html/kev_explores_pelt_water_chille.php

You might want to attach a really big heatsink on each head and put the heatsinks in a large res.http://www.xtremesystems.org/forums/showthread.php?t=18408

I found this guide for water chilling. It suggests putting the cold part in a res. You might be able to do this too, make a steel or aluminium res, insulate it a lot and stick it in your case. Either way, you should insulate everything that comes in contact with the cold water, tubes and all. I personally have not done this, in fact i have not water-cooled at all, but i am researching it now for future reference, i just don't want you to be disappointed if you don't get what you want out of the project.

 

[ciscokid454]

hmm, more and more bullet holes on this project.
I'm doing this to see what i can do with what i currently have. Without going extreme on modding the units themselves. If I wanted the highest overclock, i surely could diect mount them onto the cpu's and go from there. My test on this project was to see how cool i could have the whole system running, not just the cpu's. Last time i used one of these and the chiller block, I had my nf7-s 2.0 and 2500+ running 250fsb. That's not to say i won't change this. Just running an idea through and see where it goes.
I've never seen one done like this, if someone can find something even remotely like it, then please post it. The link to the chiller using the pelt's is very interesting.
I made the chiller block with several paths to slow water down, passing through it, and i'm using a slow pump for everything, and extened tubing runs. Also have a 600gph eheim that i might test with and see if running water through the loop very fast makes any effect. Nothings even setup yet or even been leak tested, so i'll post more as I complete it, and we'll see how things go
Cisco

 

[VO_Nuclear]

The project looks very good and promising, very innovative.

I'm not sure how well the design will work though, as a water-chiller. If you want to stick with a water-chiller design, why not mod the Kryos to act as true water-chillers in a insulated tank? Both of them combined could provide very low temps, and I think in the end it will be much easier.

Either way, great job and good luck!

 

[ciscokid454]

The project looks very good and promising, very innovative.
I'm not sure how well the design will work though, as a water-chiller. If you want to stick with a water-chiller design, why not mod the Kryos to act as true water-chillers in a insulated tank? Both of them combined could provide very low temps, and I think in the end it will be much easier.
Either way, great job and good luck!

TY, there are a ton of modding options avail to me if I had the funds available.
First thing I can see is having better cooling mounts put on, with more flexible tubing.
Cascading both of them together as well. ATM, my local cooling "experts" wouldn't have a clue of what or how to go about anything like that. If someone around DFW know's anyone that can do anything like this and can do it inexpensively, then fine hook me up.
Until then, i'm following through with my original theory to see what I can do with this. It may fail miserably. I'm going to see what it does though first.

 

[ciscokid454]

just a small update:
I'm waiting on some new parts from danger den, specifically the copper tops, and a new resevoir, that is supposed to not break due to temperature or anything else for that matter.
I've got teh chiller block mounting taken care of. I'm setting up the tubing runs in preparation for the new res. I don't feel comfortable testing it with the acrylic tops now, so this is getting backburner'd until this weekend more than likely.
On a side note, i'm going to be modding my lian li pc71b with a prometia machII and a 3.2e I recently acquired. I'll finish up the pics of the current config and post them this week.

 

[zer0signal667]

I made the chiller block with several paths to slow water down, passing through it, and i'm using a slow pump for everything, and extened tubing runs. Also have a 600gph eheim that i might test with and see if running water through the loop very fast makes any effect. Nothings even setup yet or even been leak tested, so i'll post more as I complete it, and we'll see how things go
Cisco

Why are you trying to slow down flow? Higher flowrates are always better for heat transfer. Rather than slowing flow (and heat transfer) you should focus on increasing turbulent flow (and heat transfer) inside the chiller.

 

[ciscokid454]

Just how i'm setting it up for the moment.
I've read alot of threads and articles discussing flow rate, and i'm satisfied for the moment, keeping the flow rate as it is with the current pump i'm using.
More importantly, the ehiem is rather large and even though the case is a dual bay, it's still not that accomodating compared to all the things i'm going to cram in here.

 

[zer0signal667]

Just how i'm setting it up for the moment.
I've read alot of threads and articles discussing flow rate, and i'm satisfied for the moment, keeping the flow rate as it is with the current pump i'm using.
More importantly, the ehiem is rather large and even though the case is a dual bay, it's still not that accomodating compared to all the things i'm going to cram in here.

Ok, but you said you were trying to slow the water down, which is counterproductive for efficiency of heat transfer. Perhaps in the future when you begin a tweaking phase of the project you might take this into consideration. I know how you feel just wanting to get everything up and running first.

 

[ciscokid454]

Ok, but you said you were trying to slow the water down, which is counterproductive for efficiency of heat transfer. Perhaps in the future when you begin a tweaking phase of the project you might take this into consideration. I know how you feel just wanting to get everything up and running first.

actually, i've done thorough testing with both the pumps I currently have for use, with the water chiller block and only one of the kryo units. There was negligable difference, between a high flowing system and a slower flowing system. I'm not even talking a 1/2 degree of difference between the two.I've taken alot of things into consideration, but if I find that during the "tweaking" stage as you call it, that it's not running as efficiently as I think it should i'll try the ehiem agian, and see what happens. Until then however, i'm still waiting on Cooltechnica to deliver my damn "uncrackable" resevoir and DD fill taps. Sucks.
I order from DD and cooltechnica at the same time on monday, DD already delivered my parts this morning. I could have bought everything from them, saved on shipping, and had everything here this morning, but I decided to save a couple bucks and ordered from both, and i'm paying the price in the long run now. I'm really getting dissapointed with cooltechnica. I had some delay issues with DD for awhile where they would take up to two weeks to get product out, In the past six months however, they've shipped out same day or next day with product, depending on order time and holiday's.
Oh well back to you.

 

[zer0signal667]

actually, i've done thorough testing with both the pumps I currently have for use, with the water chiller block and only one of the kryo units. There was negligable difference, between a high flowing system and a slower flowing system. I'm not even talking a 1/2 degree of difference between the two.

That's all I needed to hear Sounds like you've done your homework

 

[ciscokid454]

I know everyone's a better tech than the next person, and that everyone else's way is better. I understand that completely, but yes, i've done my homework on all of this, except the acrylic. I just literally read about acrylic cracking only a couple of hours after originally starting this thread. I'm really looking forward to Cooltechnica shipping my parts out to me, so I can finally get this underway. I'll get some pics tomorrow hopefully of things going together. I was going to start off with the 1.6 lv xeons, but i'm far too lazy irl to take things apart agian unless there's a problem, so i'm just going to stick the 2.8 es's in the board and go from there. Don't want to get anyone else's hopes up.

 

[zer0signal667]

I know everyone's a better tech than the next person, and that everyone else's way is better. I understand that completely, but yes, i've done my homework on all of this, .

I wasn't saying anything like that, simply stating the facts since often people have theirs backwards or upside down. I can't remember how many times I've heard people state for a fact that slower flowing water is best in the radiator vs faster flow through the waterblock. I always do my best to dispel that myth, so that's why I chimed in.

 

[ciscokid454]

I wasn't saying anything like that, simply stating the facts since often people have theirs backwards or upside down. I can't remember how many times I've heard people state for a fact that slower flowing water is best in the radiator vs faster flow through the waterblock. I always do my best to dispel that myth, so that's why I chimed in.

It's ok, I know you weren't or didn't. I'm always reading forum posts and people always chime in on things they "know" about more than anyone else.
That probably sounds like i'm insinuating you were. Not gonna get into symantecs:-p
I read several articles and threads before i even started this project, covering condensation, protecting your system and what not. The fast is, before this project I never have worried about it. If something dies due to my negligence, then that's my fault, no one else's and i'll handle it on my own accord. I am no newb to overclocking/modding/supercooling disasters, I'm pretty sure it's not on here anymore, and i probably don't have the pics, but I toasted a brand new VP6 when it first came out, by ripping the NB off with the ASE'd chipset cooler I on it.
I went right back out, bought a new VP6, and learned the proper way of applying and removing Artic Silver Expoxy. Years later when Abit was allowing all mobo's to be rma'd for $25 if out of warranty, no matter what condition they were in, I jumped on it and RMA'd 5 boards I had managed to kill over the years.:-p
But enough ranting, my whole theory behind this is that by allowing the NB to be cooled below ambient with the cpu's and gpu, you'll have a more stable system and possibly higher overclock. Plain and simple. I'm sure that seems totally idiotic to some, but that's my plan. I've been running this same theory for a number of years with good succes.
Dvd266u-rn's runnign 1.13/1.26/1.4 taullatins @ 160fsb+, MPX2's running any variety of XP past 160fsb. An NF7-S 2.0 running 250FSB+ 1:1.
Those stats mean nothing really, as this system is as yet untested until I get the last few pieces in and can finish the buildoff. I'm hoping that by monday everythings finished and fully up and running. That all depends on the parts as well. If I get everything in the morning, i'll sure as hell start piecing it all together and bleed testing the watercooling.
Once the watercoolings leak tested, I can start testing the chiller itself. I'm sure i'll run into several first run glitches, knowing my luck, but that's the fun of it all.

 

[ciscokid454]

sorry for the delay on pics and an update. the max3/prometeia mod went slow and painful, but it's up and running now, and i'm torture testing it now, for stability, so i'll start overclocking it tomorrow, while i'm starting the build on this project.

 

[ciscokid454]

My sincerest apologies for all those interested in this thread work calls sometimes, that and added problems with getting proper parts for this project.
Ok, i've got it all installed, mounted, connected, blead and leak tested.
water temp with both units running is 32f under no load, system not on at all.
Everything was iced over. I've run into another issue with the power switch, (power switch on the Board, not the physical power button) where the system won't power on immediately, once pressed. I've been attempting to get a resolution from iwill about it, but no luck as of yet.
I believe under normal heat load this machine will be fine, once I work out all the bugs.
Sorry, no new pics atm.

 

[ciscokid454]

ok, after many quirky issues, a few problematic power supplies and many restless nights.
She's alive. I'm currently testing it with one kryo on, see how that handles everything.
More later.

 

[ciscokid454]

original dh800 crapped out on me shortly after testing it, sent it back to iwill and went through many months of fighting with them to get a working board.
I finally found a company that still sources 1.2 rev boards, deinstalled and reinstalled everything this past thursday..now i'm running into x64 errors stating that the es chips aren't x64 chips....
working with iwill again to see what they can do, can't even flash the bios..