new watercooling setup; best of the best

[BillA]

Bleh...my computer crashed and I lost the post.
(Joe did you do this to me ? joke eh ?)

And in conclusion .......
Kevin, please read the first two posts in this thread. Lisch asked a question and Joe answered. I called Joe on his waterblock appraisal and he got hot.
My posts have been (generally) on topic and certainly less personally oriented than Joe's.

I enjoy an analytical and technically rigorous discussion because it FORCES one to ask why things perform as they do, and how might they be made to perform differently. (Yes, my background is materials engineering, product development, testing, and marketing.)

Marketing, the art of telling the customer what they want to hear. Here is an unattributed quote, and my response, from the major non-participant to this thread:
"I read the posts, but I think that practical experimentation may dispute the theoretical conclusions, as has happened so many times in our little niche<g>"
"no argument
we have speculation, calculation, and demonstration;
and they are not always the same"

But it takes GOOD testing to be able to distinguish between marketing and performance, a concept that is apparently too taxing for some. Of course one can always select products by popularity.

This is not my website/forum, its Joe's; so I'll cease my unpopular carping in the interest of tranquility.
Over and OUT
be cool

[lisch]

sigh

i checked on those dangerden rad's there sold out to! actually those would fit in my system perfectly, even for so than the little black ones because of the in/out ports facing forward..

[Kevin]

If you aren't going to be using pelts, OCWC's silver tower should work fine.

[Freakyfrank]

he guys

thanx for bringin this thread above..

almost made a 3mm thick colplate for my celeryII 600 @ 900 +? (shame shame )

but now i've read the whole threat (in 15 minutes.. record??) i'm gonna make myself a 10mm thick coldplate.. COPPER!

3mm = 1/8"
10mm = 3/8"


i got some tech data here..

u are also going for the 2 TEC's parallel?


celery II 600 1.5v

@600 1.5v 18 W
@900 1.5v 27 W
@1008 1.5v 30,5 W
@900 2v 36 W
@1008 2v 40,5 W
@1116 2v 44,8 W
@1200 2v 48,1 W

we'll take the worst case (not that i'm hitting 1200Mhz )

serie 1 TEC / 3 TEC’s parralel :

layer type Watt endW Temp DeltaT

105 mm2 CPU 48 W 48W
156W TEC PEL 117 165W -54C 28C
3x 156W TEC PEL 351 W 516W -13C 49C
317,25 cm2 Radiator 24C

1 - ( 48 / 117 ) * 70 = 41C delta
1 - ( 165 / 351 ) * 70 = 37C delta


serie 1 TEC / 2 TEC’s parralel :

layer type Watt TusW Temp DeltaT

105 mm2 CPU 48 W 48W
156W TEC PEL 117 165W -37C 28C
2x 156W TEC PEL 234 W 399W 4C 49C
317,25 cm2 Radiator 24C

1 - ( 48 / 117 ) * 70 = 41C delta
1 - ( 165 / 234 ) * 70 = 20C delta


2 TEC’s parralel :

laag Type Watt TusW Temp DeltaT

105 mm2 CPU 48 W 48W
2x 156W TEC PEL 234W 282W -31,6C 28C
317,25 cm2 Radiator 24C

1 - ( 48 / 234 ) * 70 = 55,6 C delta

1 TEC:

laag Type Watt TusW Temp DeltaT

105 mm2 CPU 48 W 48W
2x 156W TEC PEL 117W 165W -17C 28C
317,25 cm2 Radiator 24C

1 - ( 48 / 117 ) * 70 = 41C delta

and off course 2 BIG-ASS papst 127mm diameter 230v.. total of 300 m3/h or 174 CFM at only 37 DB!


------------------
OC'ing is my middlename

[Freakyfrank]

DAMN

where is my nice layout?

well here it is in short:

ambient temp = 24C
TECS = 156 W (75% = 117W)

serie 1 TEC / 3 TEC’s parralel :
cpu die = -54C

serie 1 TEC / 2 TEC’s parralel :
cpu die = -37C

2 TEC’s parralel :
cpu die = -31,6C

1 TEC:
cpu die = -17C

U wanna know why? look at data above!

btw for the serie 1 / 4 parallel the CPU die would be -62C af FULL load and i will get a 5% discount.. LOL if i can make the money i'll get 5 TEC's and two 1300 L/h pumps (2x 288gph)
(and its still cheaper than a kryotec)

but two 156W tec's will do it? or just one ?
what do u guys think?

im running my celeryII 600 @ 900 just with a golden orb (but on 2V) at 40C full load
1008 or 1035 or 1080 would be nice..


PS well is this the longest thread or what




------------------
OC'ing is my middlename

[H2gO Flasher]

dam thats the best read I've had in ages
- its about time questions were asked

[Joe]

it will all be answered soon enuf

Swiftech is sending one of their blocks and it will be benched against OCWC's blocks ( Cu and Ag all which cost 50% of what the swiftech one does)

it will be run on a Duron core with a 130watt Pelt.

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[lisch]

finally! a head to head review.. now thats more like it

[VegetaSS4]

this is gonna be sweet-ass good.

[Joe]

Talked to Gabe at Swift for quite a while tonight on the phone when he called me.

This will be one helluva comparison!

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[BillA]

May I suggest that you follow Joe Citarella's procedure http://www.overclockers.com/articles300/ of embedding the thermocouple in the baseplate.

This is also the Intel etc. procedure.

Glad to hear you're talking directly with Gabe, info is always better "from the horse's mouth".

Look forward to the results. What about a DangerDen block also ??

be cool

[Joe]

I am working on a few methods of temp reading, one of which is on the BACK of the cpu.

I am talking to someone right now bout designing a temp sensor that will work on the back of a socket CPU even when its in the socket.

OC'rs temp Reading procedure is good for reading the water block temp over the core, but I am also concerned about the core temp under the core and don't fancy the weak Mobo Probes.

Also I need to find someone with a higher quality drill press and such to drill the holes for me as I don't like the idea of destroying the cooling ability of the block by warping the base plate from drilling it. ( small ripple where the drill went down, if its done too fast or wobbles while being drilled).

The temp reading is definitely critical and will be the subject of much planning.

(edit: OMG my spelling was awful! I shouldn't be posting while talking to people at work I guess
------------------
C-ya
Joe - Owner/Editor www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[This message has been edited by Joe (edited 01-09-2001).]

[BillA]

I saw a writeup where a flat thermocouple was put on the backside of the CPU and the leads were very thin lacquered armature winding wire and simply run out through the socket pins.

This would cause the CPU to sit at a bit of an angle but shouldn't affect much.
The writer said it worked well.
(Sorry, I don't remember which forum.)

be cool

[Joe]

and yeh, there will be quite a few blocks I think. I figure if I am going to go through all the work to do a show down between OCWC and Swift, might as well get a load of other blocks and basicly give a complete look at the cooling abilities of most blocks available on the market. ( alotta work... I figure a month or so of prepare'n and testing..)

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[Joe]

Originally posted by BillA:
I saw a writeup where a flat thermocouple was put on the backside of the CPU and the leads were very thin lacquered armature winding wire and simply run out through the socket pins.

This would cause the CPU to sit at a bit of an angle but shouldn't affect much.
The writer said it worked well.
(Sorry, I don't remember which forum.)

be cool

Thats one of the ideas but I dont plan on having the CPU sit on an angle at all and I plan on it being used as an anti-condensation device as well.



------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[BillA]

Have you priced a thermal test die such as the CPU mfgrs use ?

Such would GREATLY simplify your setup.
(eh ?, no computer ??)
After all, the only goal here is to be able to calculate the (relative) C/Ws of the various waterblocks (and HSFs if desired as well).

be cool

[This message has been edited by BillA (edited 01-09-2001).]

[Joe]

That's one of the topics me and Gabe talked about last night. There are different groups of people in this industry.

1. That look at how far it allows them to OC and don't really care much how it gets there
2. purely thermal geeks who only want to know how much heat you can move and don't really care about the CPU under it, just how cold the CPU gets
3. System designers that are just working the engineering of how the system should flow and work to move heat.

I am a bit of 2 and 3 and am not really in the first group at all. But the majority of people on the web ARE in the first group. So bypassing the need for a REAL cpu under it would sorta kill the interest in those folks. but using a CPU adds a very delicate, and unstable variable to the mix.

I have looked into a "Test Die" and from what I have seen, to get one to match Tbirds core size and heat output it would cost over 1500$ for it and the test equipment. Since procooling is a not-for-profit site (unlike some people think that we rake it in.. just FYI, procooling does not recieve 1$ from anyone, its a hobby I do for fun ) , we simply don't have the funds for that. its most cost effective to just use a real core, and be careful

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[BillA]

Sounds reasonable in all respects.
Look forward to the results.

be cool

[Freakyfrank]

mmmh
how do u know that the cpu is creating the same amount of heat each time?? by playing 1/2 hour UT .. well you gotto do it anyway so..hehe
i've seen somebody using a 50w resistor.. that was putting lots of heat out..
u can't measure the CPU.. but the heatsource is VERY constant and the coldplate on the bottom and resistor can be measured...
just a thought

I'm waiting!! HOPE ITS FRIDAY!! hehe
1 min to go here..


------------------
OC'ing is my middlename

[asymmetric]

Originally posted by BillA:

The heat removal capability of the coolant is (among other things) a function of the temperature difference between it and the object being cooled. By initially applying the coolant, when it is at its coldest, to the center of the waterblock (which is presumed to be the hottest area), the heat transfer is optimized for that specific area of the waterblock.

Ok, maybe I'll be shooting myself in the foot here, being pretty new to all this, but I think this paragraph summed up the entire thing, and perfectly illustrated why (IMHO) either of the coolers would be a fine choice.

On the one hand, we have an aluminium waterblock, with the water doing -most- of it's action near the center.

On the other hand, we have a silver waterblock, with the water acting more equally over the entire surface.

If you spend just a moment thinking about that comparison, and going back to some other data that was presented on the thermal transfer properties of these two materials.. I don't know about you, but something jumped right out at me.

Because of it's lower thermal conductivity, the delta (center to outside edge) on the aluminum block will be higher than that of the silver block, by a great deal possibly.

With that in mind, to me anyway, it makes perfect sense to concentrate more on the location where heat is being applied to an aluminium block/plate. With a silver block or plate, because of it's greater thermal conductivity, the delta will be smaller, and you will have to cool the entire surface effectively.

I think the material used is directly related to the kind of fluid circulation pattern you will aim for. For example, using a centrally focused approach on a silver block would probably be a bad idea, because the silver would be more evenly heated (and thus need more even cooling) than the aluminium does.

Hope this made some sense to someone besides me.. it is 5am..

[Cablepro]

actually, the bottom of the swiftech block is copper...

[Freakyfrank]

Joe..

can u also review a modified 486-heatsink
(thin and thick based)
and a direct waterflow on the hotside of the pelt?

------------------
OC'ing is my middlename