Page 1 of 1

Waterblock Suggestions?

Posted: Tue May 11, 2004 3:30 am
by mdaniel
I'm looking at starting a water cooling rig and I would like some input on choosing a waterblock. I know the RBC, WW, etc. are the best, but I'm not interested in using the louder pump that they require. I'm probably going to use a Hydor L20, a BIP w/ Enermax adjustable 120 and a res. The only thing I want to WC is the cpu. I've read mixed reviews on pretty much everything. I want one that will support 1/2' tubing. It looks like I've narrowed it down to a DD Maze3, DD Maze4 and the Zalman. Any thoughts or other suggestions would be nice. I'm not intersted in GPU cooling, as the VGA Arctic Silencer is very quiet & effective on my 9700. I just want to be able to mildly OC and lose the loud CPU HSF that is the loudest aprt of my rig.

Thanks

Posted: Tue May 11, 2004 5:19 am
by chylld
the zalman waterblock is apparently pretty good value for money. not a top performer, but respectable, nonetheless.

if your budget will stretch to a high performance waterblock however, i recommend you still get one, e.g. a D-Tek WW or a Dangerden RBX. even with a quiet pump, they'll still outperform the cheaper offerings by a significant margin.

another good waterblock is the Asetek Antarctica. in fact, i suggest you get an asetek waterchill kit as it seems to fit your needs nicely, and contains many of the components you already want to get.

happy hunting!

Posted: Tue May 11, 2004 8:22 am
by afrost
I'm running an RBX with an eheim 1048 and am pretty happy with it. With the RBX you can use the different nozzles to adjust for your flow rate....but the trial and error aspect of it would be a pain. I put in the jet nozzle and haven't tried any others yet. I read a review where someone tried the jet nozzle and got better temps than the stock nozzle so that's what I went with.

1048 has less flow and head rating than an L20 I think.

the reason why i didn't go with a WW is because they weren't in stock when i wanted a block.

Posted: Tue May 11, 2004 7:18 pm
by bobkoure
afrost wrote: 1048 has less flow and head rating than an L20 I think.
There's a site in Germany that did actual flow testing (imagine that!) pump spreadsheet at watercoolplanet.de
I'f you're not a German speaker/reader (I'm not but have systran) I've put a translated version here: Pump Data
I've also deleted pumps we don't see in the states and uninteresting-to-me data like size, cost in Euros, where to buy it in Europe.
AFAICT, the "result" columns are measured flow through 2 meters of tubing of the specified ID at 0 head (exit at same altitude as inlet) I could easily be wrong on this one as I haven't found the "pump testing procedures" (or however you say that in German) page.
Remember that these are European (so 230V) pumps being tested - except for the HPPS which should be the same Europe or States.

Posted: Wed May 12, 2004 10:40 am
by mdaniel
OK...I'm pretty decided upon the Hydor L20 as my pump. It should be enough to handle a res, a rad and a cpu block only. I like the small size of it to fit in my 3700AMB and I like the low noise it produces.

I'm confused on the waterblock though. Everything I've read has suggested that the RBX & WW type of blocks wont work well with anything less than an Eheim 1250 or something equal to it. Does anyone know how the results of one of these compare to something like a Maze4 when used with a small pump?

Posted: Wed May 12, 2004 11:42 am
by afrost
you will mostly hear that 1250 thing from overclockers who only care about performance and not noise. not that a 1250 is loud but i think a WW or RBX will be fine with the L20. It will kick the crap out of aircooling anyway. If you want something better you will have to wait for Cathars P2, and who knows when that will be out if ever.

Posted: Wed May 12, 2004 12:43 pm
by mdaniel
I just found some new info on new versions of the Hydor pumps now on sale at DD. I just picked up the L25 for $38 from DD.
Depending on your system needs, you may want to try one of our new pumps; L25 or the L35. The new series of pumps feature new impellers and lower wattages. The new L25 ($39.95) uses 10W of power with 265 GPH output. The L25 body is smaller than the L30, but the hose connectors are the same. The new L35 ($59.95) uses 23W with 450 GPH output in the same body and hose connector size. I have tested both pumps and found them to be quiet and powerful. Both pumps will be hitting the PC cooling market very soon.
That ups the flow rate from 185 to 265GPH. That's all I need. I think I'll hold of on the waterblock until DD releases its new TDX, which is supposed to be like an RDX, but with two fittings. DD indicates a new product will be anounced on Friday.

Posted: Wed May 12, 2004 1:59 pm
by pdf27

Posted: Wed May 12, 2004 2:04 pm
by chylld
interesting.

from the looks of it, it has a single clear top layer and a single copper base. the top layer seems to only have barb/screw holes and a plenum at the input, so that means all the complexity is in the copper base. with only a plenum to channel input water onto a copper base, i wouldn't have too high hopes for the performance of this waterblock.

all of the current high performance waterblocks have complex arrangements even before the water hits the copper. the cascade has microjets, the whitewater has a slot, etc.

it appears that dd and swiftech are really trying to bring down block production costs - their latest designs seem to have a huge emphasis on reduced milling time. but of course i could be wrong :)

Posted: Wed May 12, 2004 5:42 pm
by HammerSandwich
I just checked out the new Hydor pumps at Dangerden's site, since Hydor's site isn't particularly informative. Unfortunately, DD's head specs are weird at best. Assuming that the L25 really does deadhead at 1.35m, it is not much of a pump for PC watercooling. Even the Eheim 1048 gets 1.5m, while Swiftech's MCP600 provides 3.2m.

Low head will limit a system's actual flow, especially with the more restrictive blocks like the RBX, WW, and Cascade. With that said, these blocks are sufficiently better than a Maze4 or Swiftech 5002 to provide lower temperatures with much less flow. Real-world results are greatly system-dependent, so you need to learn the math involved or keep throwing money at the problem until you like the results. Chylld likely will agree... :)

For a low-flow system, I'd look to Swiftech's new MCW6000 block. I've yet to see any independent tests on it, but Swiftech's data looks very solid (and BillA ain't much of a BSer). The alternative is to wait for Cathar's P2, but - as afrost said - that's not coming out any time soon.

Posted: Wed May 12, 2004 9:53 pm
by afrost
that's weird, i thought i saw a graph on procooling that put the L20 and L30 above the 1048 tested with a real radiator. I might be on crack though....

Posted: Thu May 13, 2004 6:18 am
by mdaniel
Well, I just ordered one from FrozenCPU. They cost the same amount as the RDX, but I like the two barb design...whether it works better than the RDX or not, no one knows at the moment. My case will be cramped enough with two bards to not have to add a Y-splutter.

I'll let ya'll know what my temps are when I get it and have a chance to set everything up. Sorry, I won't be able to give a comparison to any other waterblocks, as this will be my first WC setup. It will consist of a Hydor L25, the new revision of the DD bay res, a BI Pro, an Enermax adjustable 120 on the rad and the new DD TDX. Maybe I'll even break out the digital camera for my computer for a first time. I'm just trying to get rid of my noisy cpu fan w/o giving up any performance. I figure I should have everything installed next weekend by the time everything arrives and I have a chance to leak test it.

Temps with my SLK900A on my 2500+ @ 3200 w/ and Enermax adjustable 92mm running at 2000RPM are 46C at idle and 55C at load. AS5 will be used and the room temp is usually around 72F.

Posted: Thu May 13, 2004 6:42 am
by afrost
if your setup ends up similar to mine you should see about 37 to 40C load temps depending on ambient and rad air flow. that should allow you to crank up that 2500 a couple of notches ;)

Posted: Sat May 15, 2004 6:16 pm
by mdaniel
Good Job FrozenCPU!

They sent me the wrong waterblock. They sent me the A64 version instead of the XP version...which means I will be waiting even longer to setup my first WC rig.

:oops:

Posted: Sun May 16, 2004 1:01 am
by pdf27
In case anyone hasn't seen yet, the TDX testing results are out. Performance is marginally worse at the same flow rate than an RBX, but the TDX has lower resistance (so higher flow rates) and doesn't have the mounting issues of the RBX.

Posted: Sun May 16, 2004 1:44 am
by chylld
bit of a disappointment, the tdx.. it would have been nice to see them make some real improvements to their waterblock, like little river, swiftech and asetek do

Posted: Sun May 16, 2004 4:29 am
by pdf27
I wouldn't go that far - in real life it'll probably beat the RBX. At any decent flow rates it looks like only the WW and Cascade will beat it at the moment - and the cascade is no longer in production. That isn't bad performance at all.

Posted: Sun May 16, 2004 4:33 am
by chylld
the cascade is no longer in production, but it's successor is well on the way. codenamed "storm" by cathar (temporarily) it's a design that pips even the SILVER cascade by a couple of degrees, yet takes a tenth of the time to manufacture compared to the whitewater.

think about it. :)

Posted: Sun May 16, 2004 10:44 am
by pdf27
Yeah, I've been reading the thread on ProCooling.
If it really takes 1/10th of the effort of a WW, then I'd have to say he's probably got the market sewn up for years!

Posted: Thu May 20, 2004 7:42 am
by HammerSandwich
HammerSandwich wrote:For a low-flow system, I'd look to Swiftech's new MCW6000 block. I've yet to see any independent tests on it...
Overclockers has just tested the MCW6000. Cooling is very good, but the pressure drop is quite high. Using this block may result in a low-flow system!

Posted: Thu May 20, 2004 2:48 pm
by chylld
that's a huge disappointment. i hope the mcw6002 is significantly less restrictive, although that seems highly unlikely...

Posted: Sun May 23, 2004 9:11 am
by GeekMoses
For further knowledge of the subject, the D-Tek Customs White Water http://www.dtekcustoms.com is a good cpu block as well. I use it an like it. =p

-GM

Posted: Wed May 26, 2004 2:28 am
by Seal
I think alot of you have misinterpreted the other watercooling websites significantly on the front of high pressure waterblocks+low pressure water pumps.

Before the age of high pressure waterblocks came about, the talk was all about making your waterblock as low restrictive as possible. People found that with low restrictive waterblocks if the pump was too large, temperatures would get worse.
... Let me explain, in a low restriction waterblock, water can flow very freely through it. If cooler water is running through a hot waterblock, lower (but not too low) flow rates would be BETTER. This is because rather than the water rushing past, it would go through the block slowly and thus would allow more time for heat to transfer from the copper to the liquid before being pushed out for the next lot of water. In a high flowrate+low restriction setup, water would just constantly loop and when there are significant temperature changes it would take longer for the high flowrate system to adapt to the difference. This theory can easily be explained by physics' specific heat capacity of copper and water being quite high.

Nowadays the story is quite different, high restriction waterblocks are now the norm and with highly restrictive waterblocks a different method is used to cool the waterblock. High restriction waterblocks like RBX, maze4, WW etc... all use methods to restrict the water flowing, forcing it to fire through small surface areas at very very high speeds usually by impingement methods. ALL of these waterblocks are designed for HIGH HEAD PUMPS with HIGH FLOWRATE. I cant explain the physics of blocks like the WW because i simply admit i dont understand it to that degree although its well documented by figures such as cathar over at procooling. It ISNT hard to see otherwise with a basic knowledge of watercooling that a LOW flow pump with a HIGH restriction waterblock just is NOT going to work.

This brings me to my setup... I wanted a high performance QUIET setup and have a fair bit of watercooling experience under me. I have built high performance non quiet watercooling systems for others with blocks like the WW and eheim 1250 and a gpu block, and to my surprise it performed worse than my low restriction, quiet setup. This was because of the gpu block being a little restrictive, the cpu did not get as high pressure as it required and so temperatures significantly suffered. Me on the other hand, i have a eheim 1048, and a low restriction loop (wb is atlantis rev3 - not made anymore) and with a 1/2 decent radiator, my temperatures as many on this forum would agree with me are pretty amazing. My 120mm radiator fan is on 5v and with an ambient of 22 deg c my full load temperature is 42 deg c. (and thats with a crappy evercool 120mm fan - im getting a papst soon that will startup at around 3v, im planning to run it on 3.5-4v)

Posted: Wed May 26, 2004 3:41 am
by chylld
seal, forgive me, but i have to disagree with many of the points that you raised in your post.

higher flow rates are ALWAYS desirable. proof of this is the fact that every paper i've read on heat transfer indicates that the nusselt number (heat transfer) increases with the reynolds number (flow rate). you say that water rushes past the surfaces of the waterblock, that is true, but there is also a lot more water going over a given surface of the waterblock for a given time, compared to a low flow rate; so this balances out. furthermore, higher water velocities induce greater turbulence / larger stagnant regions which therefore increases heat transfer.

the WW works by impingement onto the bottoms of the 'valleys' in the middle of the waterblock. the resulting stagnation region spreads naturally along the valleys thus widening the stagnation region and providing high heat transfer in the area directly above the cpu die. a degree of secondary jet impingement occurs as well but is not as profound as on other blocks, e.g. the cascade and, to a lesser extent, the rbx/tdx.

a high restriction waterblock will often work just fine in a low flow setup. just because there is a lot of restriction in the waterblock doesn't mean it kills performance altogether, and in most cases you will find that a lower flow rate through a high restriction waterblock will outperform a higher flow rate through a low restriction waterblock. the only instance where a high restriction waterblock would suffer is when it causes the flow rate to slow to a near-trickle, and with modern pumps this is rarely the case.

the results you've seen from your friends' systems are valid however they do not support the conclusion of inadequate pressure resulting in lower performance. there are simply too many variables, the most important ones being the composition of the watercooling loop itself, and also the temperature measuring mechanism which varies drastically and unreliably from motherboard to motherboard.

Posted: Wed May 26, 2004 7:05 pm
by HammerSandwich
Seal wrote:People found that with low restrictive waterblocks if the pump was too large, temperatures would get worse... If cooler water is running through a hot waterblock, lower (but not too low) flow rates would be BETTER.
As chylld said, this just ain't so. Heat transfer is directly related to the temperature difference, so water loses cooling ability as it warms. If you still doubt this, think about aircooling: has anyone here seen a system that ran cooler at a lower fan speed? When a WC system gets hotter with more pumping, that's because the pump itself is adding heat to the water.

Posted: Thu May 27, 2004 1:41 am
by Seal
hmm good piont