High Idle?! Home made Block Ideas...
 

Hey all, it's pretty damn HOT in my room, but doesn't bother me one bit... but it DOES bother my rig! Usually temps range from 80F to a blistering 89F. All the while AsusProbe says I'm idling at a steady 40C, no matter WHAT the ambient is! I was down to around 26.5 one day and it STILL was high at 40C. So I decided that I was going to do some more modding to my current home built unit. See the pics Anyhow, after some serious modding and getting a new pump, heater core, tubes, the damn thing still idles a high 40-42C but max's around 45-48 on the hottest days (89F+ ambient!)... Getting a milled block like the maze series should drop idle, but the block I've got isn't all bad and it should preform much better than it's currently doing... Any advice would help! (Aside from getting a new block, at the moment, I'm strapped for cash, and I kinda like building and experimenting first)

I've also attached some pics of my ideas for a north bridge block also... THanks! http://www.geocities.com/v12v12v12v1...ing_Set_up.jpg http://www.geocities.com/v12v12v12v12/NBridge_block.jpg

I'll have to UL the pics, the URL doesn't seem to be working?

Here's the other one, sorry bout the mess!

Welp, looks like I'm not part of the "proforums-click" So I'll have to respond to myself: "V12|V12, I think you're best choice would be to get off yer lazy arse and get one of them fancy milled blocks that everyone's jumping on the bandwagon to get, then build yourself one of them artsy-fartsy-fancy clear res's, and then order a Maze3 (copper top) and then your temps should be okay. As for the rest of you're system, seem's okay, but it's nothing to bother worth replying to... Good luck, toodles!"

:)

Well... just reverse the fan on the heater core... They're best at pulling. Apart from this, let's see the real thing :)

Wow...self gratify much? Bitter......Party of one...This way please......

As far as your NB block, it looks OK. I would introduce as many pins as possible since that's the way that you seem to be wanting to go. ( I.e. shave down the Swiftech MCX462, or soemthing similar. You get my mental image.)

As far as the rest of your setup, go rad --> CPU --> pump. It's better that way. As far as the homemade block.......it takes only trial and error...........retail blocks are PRE-tested :D.....

Yea been there done that. It takes awile to find something that works well. It depends verry much on how well you grasp the cooling eficiency of the parts you are working with. I started with the same kinds of designs:

first this:
http://community.webshots.com/album/28051709IzZlTDxLyZ

then this with an alpha:
http://community.webshots.com/album/30556976fWLpIMxkEt

finally ended up with this:
http://community.webshots.com/album/35111553UecmAe

Just keep at it this has taken much time and tinkering.

Nice... Now that's what I'd call a tight fit! How well did those blocks perform? I think I'm going to try the 2nd North Bridge Idea first, it has the best flow potential... but will be the hardest to make... Gotta find a way to get more replies next time... this is pathetic! Thanks though :)

Well my eirlier ones peroformed ok but my later ones perform well. You can see here:

http://community.webshots.com/album/38150901opPARw

The second design looks good. Do you plan on using an available heatsink with fins to start from? Keep the pins only as tall as you would need to get the barbs to fit on the sides and make them go to the top of the block. If you leave a wide open space at the top of the block then most of the water will flow there because of the path of least resistance.

Yeah I'm thinking about going with design #2, for the fact that the water will be flowing in the direction that forces it through the pins, and I'm going to make them fit to the top of the copper end cap as much as possible... basically I'll just punch and drill the pin holes, then use some silver braze to meld them into the copper base plate... wish I had some thicker plates! 3mm ain't much, and stacking+brazing is a hassle!

I might order some metal stock online, but I've never seen any place that sells anything less than 3/4in thick...

Hope this works out, cause if so, then I'm going redo my current CPU block to the North Bridge Idea #2...

Thanks for the advice! :)

I would have replied more quickly, but I've been REALLY busy. Interesting topic, thanks V12! I don't really pay attention to who is in a clique (I really hate that concept, probably because I didn't make it into any of the neat cliques in my school days).

Anyway, glad to see you are experimenting with different block designs! Let me tell you what I chose to do, and then attempt to defend it in spite of what people commonly use.

I made a block which uses about 25 parallel cross-drilled channels (at about 0.097" inner diameter) across an aluminum block which is about 1" thick. The design was shamelessly stolen from a formerly active member of the water cooling community Xjinn. You can see a picture of his copper waterblock here: http://www.procooling.com/articles/a...t-team-wc3.jpg

The ideas behind this block are to create high surface area with multiple parallel channels like the heatercores use in dumping the heat to the air. Also, by making the individual channels narrow, you increase turbulence so that the boundary layer is very thin. The boundary layer is the layer of non-moving water that insulates the metal from the cooling water flow. By adding a large number of the parallel channels, you reduce the restriction on water flow which can kill overall system performance.

Now, why would this be better than using fins? Basically, I believe that it is better because the tall fins or pins cannot effectively transmit heat away from the waterblock base. Also, you get dead spots where water doesn't move if you use pins.

The reason I used aluminum? Basically because I got the raw material for free, and because aluminum is much easier to machine than copper.

The temps I get are around 48C at full load and 34C at idle. Ambient temps are about 27C, and I'm using a std heatercore with 2 120mm fans running fairly slowly, 3/8" tubing, and an Eheim 1250 pump. Interestingly, I get about 40C for normal apps using 100% of the cpu for hours. I can only get up to 46C to 48C if I use K7Burn for over 30min, which means that I could probably bring that down if I wanted to run my fans at full speed. Temperatures taken with the XP internal diode on a XP2100 running 13x133 (I haven't unlocked it yet).

Nods,, Brian when you do unlock that XP tell me if it is any more difficult then a 2000+ and earlier. I heard there was more to it.

v12, you can take a look at this (see attached image)

This is a top-performer block made by a german company (http://www.cooling-solutions.de/).
Looks like one of your designs :p

damn you gmat for showing all these blocks made by companies i cant order from!
:cry:

that is a nice looking block

Whoa, that's looking mighty nice... But, I'm not into that acrylic stuff... it scratches too easy, eventually stains and has a much higher random leak factor... as for whoever who was trying to convence me that The heat diffs between copper topped and acrylic are meaningless, obviously (no offense) hasn't had any material science classes or thermal dynamics, it's a FACT of physics that metal transfers etc. et.c et.c

Anyhow, nice design, I wish I had summer access to the AeroSpace machine shop here, but, I don't and I have to make due with some lame home copper tooling, which is like trying to go muddying with slick tires! Just takes too damn long and my dremel+ mini drill press kit is worthless, it's pretty much a fancy plastic miller.

I think what I'm going to try, is make an actual flow walling of pins brazed together. LIke the pic in Nbridge idea #2, where the pins in the middle separate the water in to 2 channels with staggered pins in the middle... Or course this sounds great and I can picture just how it's going to look, but the final creation will probably look like a mess and perform like one! My current homemade block is pretty hefty with the copper and all the cross cuts in the fins etc, but I've got some sort of mobo probe that is a POS and lies, as in reference to 40C idle@31.67 ambient... then a 40C idle@ 23.33C ambient...which is impossible! There has to be an idle drop if the ambient drops significantly.

Does anyone know where I could get one of them Diobe reader, SMbus thingys? I really need to get some accurate readings :mad:

***Damn, I should really cave in and buy a nice block, but I wanna build my own for some strange reason, and have tools and metal shavings scattered everywhere and then sweating over wtf the temps are lame! And then having to hunt around the physics dept looking for scraps of precious metal (copper!) I'm just a ghetto'd cooler to the core...that is unless someone want's to donate a nice, small, TORQUED OUT 10" drill press! Or a diode reader... :)

Check out pHaestus article on the diode reader.

Diode Reader
If you don't mind getting a new mobo (I know, V12, you probably don't) get an Epox 8K3A mobo. It reads the diode directly, so you have no need for an extra circuit. I think you can get the non-RAID version for about $85 or $90.

Unlocking the XP2100
I haven't done it yet, but I think you are right, Hallis. Unless you can hack the BIOS well enough to make it work, you have to reconfigure the L3 (don't quote me on the L3) bridges to make it look like an XP1800 to the mobo and THEN short out the L1 bridges. Oh for a decent BIOS so that I could use any multiplier that I wanted. With the above hack, you can't use multipliers higher than 12.5X because of BIOS limitations. Darn software engineers never can get it right.