New Pump CSP750 lauded for low noise

HammerSandwich · Post by **HammerSandwich** » Sat Jul 31, 2004 8:25 pm

frankwest wrote:When I was just running a CPU block I tried the 120mm ,with a similar shroud , the temps were better with the pair of 80mm.

Sorry for the delay; I've been away.

Were all fans running at 12V? What temperatures do you have?

frankwest · Post by **frankwest** » Mon Aug 02, 2004 3:37 pm

Yes all the fans @ 12 volts , right now the CPU is 34.1 with the temp sensor and sandra reports 38C. Its about 75F in the house , under load it goes up 3-4C. With the 120mm it idled a couple of degrees higher. When I got home from work today it was 82F inside and I was around 37C at idle.

Edward Ng · Post by **Edward Ng** » Mon Aug 02, 2004 5:38 pm

My pumps finally shipped out today via USPS from DTek in The Republik.

With luck, I'll have this weekend to play with liquid cooling for the first time!

Gooserider · Post by **Gooserider** » Tue Aug 03, 2004 9:52 pm

Just as a comment, the stuff I've seen on parallel pumps suggests they are bad news unless they are in completely independent loops (In which case it is debatable whenter they should be considered parallel) The problem gets worse the closer the pump's I/O fittings are to each other, with the worst case being split -> pumps -> junction. The problem occurs when one pump fails, and the water from the good pump's output flows backwards through the dead pump to the good pump's intake, giving an effective short circuit that bypasses all the cooling hardware. Putting a one way valve on each output will prevent this, but add excessive restriction. Even when both pumps are working, if their ouputs and flow restrictions aren't perfectly balanced, they will still end up 'fighting ' each other with the difference.

Series avoids the problem - a dead pump reduces the flow, but otherwise causes no problems, the pumps don't fight, but rather help each other so that you get a potential max flow at least as great as the weakest pump, and a potential max head equal to a bit less than the sum of the pumps in the chain. Given that higher head will give greater flow, theory suggests the series configuration should be better overall.

Gooserider

Edward Ng · Post by **Edward Ng** » Wed Aug 04, 2004 12:22 pm

Thanks, Gooserider; I will take that into consideration...

My pumps have come in, and much to my non-amusement, they are 1/2" ID, not OD, so I just placed an order for four 1/2"<->3/8" ID conversion couplers.

Along with them, I ordered two more 3/8" ID y-splitters as well as two antikink coil kits for 1/2" OD...

I've been trying to figure out which of the following configurations to try out:

1) Single straight loop, both pumps in serial, all blocks in serial; zero y-splitters.

2) The two pumps in parallel, with a y-splitter before them and a y-splitter to recombine immediately after, and then all blocks in serial.

3) Two pumps in parallel, with a y-splitter before them, but instead of recombining immediately after, one pump running straight through to the CPU block, and the second pump running through the GPU and then northbridge block, and then the two paths recombining after the blocks with a y-splitter before entering the heater core.

4) The two pumps in serial, then a y-splitter, then the CPU block off one branch from the splitter and the other two blocks in serial off the second branch from the splitter; rejoining the two paths with a y-splitter right before the heater core.

5) The two pumps in parallel, with a y-splitter before the pumps, but then a y-splitter after the pumps to recombine, before yet another y-splitter to break apart the flow one way to the CPU block and another way to the two remaining blocks, then recombining with the last y-splitter before the heater core.

At most I'd like to try three of them, but in all honesty, it's a daunting enough task just thinking about trying two of them. I'm extremely limited in time I can dedicate to this level of fudging around with fluids (particularly considering this is my first ever water cooling attempt, period; damn good thing I'm using Fluid XP+), so if anybody could please chime in and let me know which two or three combinations they'd be most interested in seeing, please state so.

I'm not going to try more than three setups, no way; I can see at least one that's already quite redundant, and a second one that probably isn't worth the trouble either, considering the other three, but that's just me, and I've yet to even try this, so maybe I'm wrong, but that's why I'm asking to see which two or three setups you guys & gals would most prefer to see compared. Best would be if you guys could please narrow it down to two, but if that's not possible, then please let me know which three from above to try.

Just indicate the scenario number; if there's a setup that I've missed that you think would be worth trying, please state it.

-Ed

PS I can probably do some nice pictures of the blocks and pumps for this (alone as well as mounted), if you guys would like it (of course I'll get full setup pictures); I apologize that I can't get a shot of the heater core all by itself, but I already installed it to my system and believe me, the mounting process was quite a handful (took me over two hours to get the entire heatercore/fan/shroud/radbox assembly together and mounted!).

Spod · Post by **Spod** » Wed Aug 04, 2004 12:59 pm

1 and 4 for me, perhaps with 2 or 3 to compare to. I expect that 1 will win for overall bestness, but it would be good to have that confirmed.

Gooserider · Post by **Gooserider** » Wed Aug 04, 2004 1:55 pm

Before putting everything together, I would reccomend doing a bunch of 'bucket tests' to evaluate flow rate. The basic rule is that the faster the flow through the blocks, the more cooling you get, so doing flow rate tests to find arrangement that gives the best flow rate for a given bunch of components should give the best cooling.

Flow meters are expensive, most folks don't have them, and they impose enough restrictions themselves to mess up the results. Less expensive and far simpler is to set up a series of tests where one times how long it takes to fill a bucket of known volume with the flow through a given configuration, and the results are accurate enough to tell you what's going on.

A kitchen sink or bathtub makes a good water source (use tap water, corrosion and contamination aren't going to be problems for the amount of time we are talking about) and get rid of as much electronics and electrics as possible - this tends to be a sloppy process and electricity and water don't play well together...

The easiest setup is to suck from the source with the pump(s), then go through the radiator, and the desired block setup into the measuring bucket. If doing a branched setup, it is best to measure each branch seperately (let the other branches dump back into the source) as well as the total. If you know the volume of the test bucket it is a simple math equation to convert the time it takes to fill the bucket into an LPM or GPM flow rate.

A few other notes...
1. Ideally the hoses should be about the same length as you'll use in the system, but this isn't real critical, just keep them short.
2. Try to keep the height difference between the pump intake and the hose outlet as low as possible, more importantly keep the difference the same for all tests.
3. Minimize the amount of distance the pump has to suck in water from, and remember you may have to prime the pump for each test. Putting a shutoff ball valve in the line between the rad and the block setup will keep you from loosing the prime.
4. The optimum result is to maximize the flow through the CPU block, since it is the greatest heat source. Next flow should be the GPU, then the low heat sources like the NB, Hard drives, PSU, etc. What some folks do is get a dual outlet WB on the CPU, with all water going to it, then one outlet going to a different branch with the other blocks on it.

Gooserider

HammerSandwich · Post by **HammerSandwich** » Thu Aug 05, 2004 12:30 pm

I vote for 1 & 3, with a new scenario for baseline:

0) Single pump with all blocks in series

Edward Ng · Post by **Edward Ng** » Tue Aug 10, 2004 12:04 pm

Well, some bad news, folks...

The Maze4 does not fit 6800 GT...the holes do not line up!

But, there's good news, too...

I do not know if it's fate or luck, but I placed my order on a Danger Den NV-68 block for my card, hoping to compare the Maze4 to it; it ships in tomorrow, so all is not lost. It just means I won't be able to compare the two for you guys (well, since it doesn't fit, not need to, even if I rigged the thing to somehow make it stay on).

First test will be Scenario 0: single pump, all blocks in serial, zero splitters. I'm still waiting for more suggestions as to alternate scenarios you guys want to see in the upcoming circuit comparisons.

-Ed

Sklug · Post by **Sklug** » Tue Aug 10, 2004 3:31 pm

What disturbs me about the whole nv6800 waterblock situation is that the waterblock will likely make my choice of next gen video card for me. I already have a Maze4GPU for my Radeon 9800 Pro, which works just fine on an X800XT. I had been considering getting a 6800GT, but not at an extra $125!!! for that waterblock. If anyone ever sees another 1/2" compatible waterblock that will work on a 6800GT let me know. I know there's some german one for $200.. I'm taking one in the $50 price range. Otherwise, I'll just wait until the X800XT's come down in price.

Steve

TheDarkHacker · Post by **TheDarkHacker** » Tue Aug 10, 2004 3:38 pm

you could always just use one of the swiftech gpu blocks and put ram sinks on the memory. i am almost certain it will fit

Edward Ng · Post by **Edward Ng** » Tue Aug 10, 2004 4:05 pm

Sklug, here is a series of e-mail exchanges I had with the Silverprop people:

My first e-mail:

From: Edward Ng
To: [email protected]
Subject: Fusion and 6800GT...

Have you been able to perform any fitment tests with the Fusion GPU block on nVIDIA GeForce 6800 GT cards? I would like to purchase a Fusion but am worried about fitment on my new card, since it is a freshly released product. Thanks in advance for your help!

Edward Ng
SPCR Reviewer
[email protected]

Their first reply:

From: Jeremy (Silverprop) [mailto:[email protected]]
To: 'Edward Ng'
Subject: RE: Fusion and 6800GT...

Hi Edward,

It should fit without any problems as our clips are adjustable.

You can use 1 or 2 FX clip kits to mount either 2 or 4 holes.

Cheers,
Jeremy Poon
BSc (Hons) BCom Melb
_______________________
Silverprop Australia Pty Ltd
www.silverprop.com

Level 1, 55 Commercial Road
South Yarra, VIC 3141
AUSTRALIA

My reply to that:

From: Edward Ng [mailto:[email protected]]
To: [email protected]
Subject: RE: Fusion and 6800GT...

Which fitment would work best? Standard (ATi)? Or GF/FX?

Edward Ng
SPCR Reviewer

Their final reply:

Hi Edward,

You need the GF4/FX clip kit to mount with the 6800.

Cheers,
Jeremy Poon
BSc (Hons) BCom Melb
_______________________
Silverprop Australia Pty Ltd
www.silverprop.com

Level 1, 55 Commercial Road
South Yarra, VIC 3141
AUSTRALIA

-Ed

EDIT: Btw; if you check Swiftech's web site, you will see that their MCW50 block can be installed, but you must buy the MCW50-NV4045 kit, specifically!

Sklug · Post by **Sklug** » Tue Aug 10, 2004 4:34 pm

I'll check out silverprop.. the swiftech however, are 3/8" OD fittings. My system uses 1/2" ID tubing. I'd hate to add the restriction of adapters. My Maze4GPU works great on my Radeon.. annoying I have to replace it. But until ATI comes out with a 16 pipeline part for less than $500, the 6800 GT seems the way to go.

Steve

HammerSandwich · Post by **HammerSandwich** » Tue Aug 10, 2004 9:36 pm

Edward Ng wrote:First test will be Scenario 0: single pump, all blocks in serial, zero splitters.

Excellent! Thank you, sir.

Edward Ng · Post by **Edward Ng** » Tue Aug 10, 2004 9:43 pm

You're welcome.

EDIT: PS: Pics of the new NV-68 block I received.

Gooserider · Post by **Gooserider** » Wed Aug 11, 2004 6:30 pm

Ed, looking back over your list of proposed setups I seem to not see the one I think would possibly be the optimal for you....

1. Both pumps in series going to:
2. The radiator then to:
2. the CPU block which feeds into:
3 A "Y" splitter so that:
4a. Branch 1 goes to GPU block and:
4b. Branch 2 goes to all other blocks in series:
5. Recombine the two blocks at either a res or the pump intake (preferably the res)

Gooserider

Edward Ng · Post by **Edward Ng** » Wed Aug 11, 2004 6:47 pm

You know, I just realized something very, very important...

It would take a great deal of typing, so I'll not discuss it here (leave it all for the article), but needless to say, I'm currently already running Scenario 0, and have decided to run Scenarios 1 and 3 to compare.

My reasons will be indicated in the write-up. Please give me time to do this, as it's a lot of work (anybody else think they'd try this?!?) and will take a whole lot of analysis on my part.

Also realize that it's very difficult to get a water cooling system to reach absolute stabilization when your ambient temperature fluctuates greatly during the period of heating (the body of water can take double-digit hours to reach a particular point of stabilization for a given amount of heat produced; my room is hot when I'm gone because it's NY summer and it's much cooler when I'm here due to air conditioning, both conditions of which will occur during the duration of any of the tests).

I'm going to do my best, guys; that's all I can promise.

-Ed

PS I repeat, as it is, this is my very first time water cooling. Ever. So keep that in mind...

DryFire · Post by **DryFire** » Wed Aug 11, 2004 7:47 pm

Ed, sorry if i missed it in your previous posts but how noisy are the pumps?

Edward Ng · Post by **Edward Ng** » Wed Aug 11, 2004 8:04 pm

Nothing to be missed; I haven't mentioned the noise level yet...

If you allow the pump to touch hard metal or hard mount directly, it will rattle or buzz. If you have it suspended or sitting right on foam, and none of the tubing directly attached to the pump touches any hard surfaces, inside a well damped case? Not likely audible outside of a couple feet from the machine.

This is my experience with a single pump so far; second pump isn't even in yet, as Scenario 0 has only one pump.

-Ed

EDIT: PS, this is preliminary experience; no serious listening tests. In fact, my air conditioner's been on this whole time.

DryFire · Post by **DryFire** » Wed Aug 11, 2004 9:44 pm

So I'm guessing nothing to be heard over a hd or fan coming from the pump, even in a undampend case?

I how this pump proves reliable as i hope to move to water cooling for my next computer.

Edward Ng · Post by **Edward Ng** » Wed Aug 11, 2004 9:53 pm

No, I can't detect the pump over my SpinPoint SP1614N.

The case is a Tsunami Dream.

HammerSandwich · Post by **HammerSandwich** » Fri Aug 13, 2004 9:49 am

Cancel the review, Ed! Your pumps are now obsolete: http://www.c-systems.ca/.

More info from the manufacturer

Edward Ng · Post by **Edward Ng** » Fri Aug 13, 2004 10:17 am

That is too damned funny!

I still think my circuit comparison will bare relevance (i.e. even with the Mark II, the effects of different pump arrangements on overall cooling capacity will still be largely the same); let me know if you guys are interested. If not, I'm just going to do up my circuit the way I believe will perform best and move on with my life!

-Ed

EDIT: Btw, I sent an e-mail to Dtek, asking when they'll have Mark II available for purchase; I wonder if I can sell off my pair of original ones and purchase a pair of Mark II's; I really like this little pump!

Actually, if anyone's interested in my pair of Mark I's, I am willing to sell them for $50 shipped, for the pair, once Mark II is available (assuming Mark II costs the same as Mark I). One is still brand new (even has the stupid seal over the inlet) and the other one is used. If it's a long time before Mark II comes out, or they cost a lot more, then both my pumps may have gone to use before I sell them (but the bright side is, you know they work and you know who tested them

).

HammerSandwich · Post by **HammerSandwich** » Fri Aug 13, 2004 10:49 am

Definitely continue with the review.

Dibs on the pumps for $50, but we'll have to figure out the variables (i.e. timing, price on Mark 2, etc.).

DryFire · Post by **DryFire** » Fri Aug 13, 2004 10:52 am

edit: give them to hammer.

Edward Ng · Post by **Edward Ng** » Sat Aug 14, 2004 7:04 am

HammerSandwich, you're a member of the Pro-Cooling forums, too, no?

Anyhow, I'll let you know as soon as I have the Mark II's in hand (do not know when that may be though, as Dtek has yet to reply).

Getting back to this circuit analysis...

What sort of test do you guys & gals want for me to use to load the system? Also, would you guys prefer that I test only retail speeds, or also for maximum overclock? By retail speeds I mean I can test at speeds available for purchase (6800 at GT and Ultra speeds and Barton at desktop 2500+ and 3200+ speeds) and maximum overclock means maximum overclock. The problems is that when I try to do maximum o/c load test on the CPU alone using CPUBurn, the MOSFETs on my mainboard shoot past 100C (AN7 reports the temperature of the PWM modules)

!!! This is why I believe I may stick to just the retail speeds for the circuit analysis.

-Ed

EDIT: PS, I have a set of Microcool MOSFET sinks and Southsink coming in; let's hope it helps!

GlassMan · Post by **GlassMan** » Sat Aug 14, 2004 4:48 pm

Gooserider,

I considered using this type of setup but it has a few problems. I was considering how to best use the two outlets on the RBX.

#1 Cuts the flow to both circuits.
#2 If the resistance of both circuits isn't carefully balanced, the side with high restance (probably the GPU) will see minimal flow.

Chipset blocks are usually low resistance, so the third block adds minimal pressure drop

The water temperature rises very little (<1C) through a block,so you don't get an increase in Delta T. You're better off getting all the water to go through all the blocks.

HammerSandwich · Post by **HammerSandwich** » Sun Aug 15, 2004 11:47 am

Edward Ng wrote:HammerSandwich, you're a member of the Pro-Cooling forums, too, no?

Anyhow, I'll let you know as soon as I have the Mark II's in hand (do not know when that may be though, as Dtek has yet to reply).

I am and okay.

Edward Ng wrote:What sort of test do you guys & gals want for me to use to load the system?

This is a tough one. We know Russ got max load with CPU Burn, and Prime95 + 3dmark gave the same results as Prime95 on my box when sthayashi brought his Killawatt over. I need to check what the Germans do when they measure vidcard power draw.

Edward Ng wrote:Also, would you guys prefer that I test only retail speeds, or also for maximum overclock?

Max OC isn't really needed in an SPCR review, IMO. OTOH, if you're going to do the OCing anyway, you might as well report the results. Also, most SPCRers probably would be interested to know if WC can give a higher clock with less noise. Sound reasonable to everyone?

Edward Ng · Post by **Edward Ng** » Sun Aug 15, 2004 1:42 pm

Well I have seriously bad news...

I found some weird gunk growing in my system, so went out and bought some hydrogen peroxide and flushed the system a couple times with pure 3% H2O2, and then a combination of distilled water and a bit of 3% H2O2. After that, I redid the fill & bleed on my system, but I seem to hit catastrophic failure. It took me nearly an hour to deal with a leak where the tubing goes out from my system to the heatercore, and then after I fixed that, I have leaks in front of and behind my pump that I cannot solve (and I do not know how they cropped up). At this point I am really, incredibly frustrated and just don't know what to do.

If anybody's willing to spend time on the phone with me, I can provide my phone number or I can call you up (if you'd prefer not to incur long distance phone charges) or we could discuss the problems via instant message (definitely preferable; I use AIM, so just click the AIM button under my post). I honestly do not know what to do right now; the system is sitting there, negected, ever so slowly leaking away into a bowl. I really have no idea what to do.

-Ed

Btw the gunk came out completely; when I ran the hydrogen peroxide flush, it all came out as some really gross looking orange organic matter.

EDIT: Okay, here are some photos of the whole setup as it sits here waiting for extra parts to fix the damage I've done (I've not the patience to make the photos all pretty like, so you guys will have to settle for what I've got linked):

The whole setup,...

a straight angle shot,...

the heatercore, passthrough, fittings & clamp array to fix the leak,...
Yes, I did try the metal screw-tight clamps, and they didn't work. Only using these plastic clamps did I solve the leak; my CPU block had begun leaking and only the plastic clamps could solve that as well. It seems the metal clamps like to come loose over time (unscrew), which is a problem that the snap-tight plastic ones aren't prone to.

scenario 0 configuration...

and finally, the pump and all the cable-ties attempting to tame the leaks, to no avail (leaking at the 1/2"ID to 1/2"OD adapters).
I've got some 5/8"OD snap-in clamps on order; hopefully they'll do the trick for me just like the 1/2"OD ones did on the rest of my system. I may end up redoing all the tubing, fittings and clamps just to be safe, so I also ordered five more feet of 1/2"OD and a foot more of 1/2"ID.

GlassMan · Post by **GlassMan** » Mon Aug 16, 2004 5:41 pm

Ed, isn't it fun putting stuff together for the first time. It has distressed me that id and od are rarely specified.

If I could offer some advice, your adapter setup looks like it couldn't help causing problems now or in the near future. check danger den
http://www.dangerden.com/mall/Misc_Hard ... ttings.asp for the fittings (1/4 NPT) and/or the 1/2" id Tygon tubing and clamps http://www.dangerden.com/mall/Misc_Hard ... _clamp.asp for the runs in and out of the pump. They claim the 3/8 tubing will stretch over 1/2 od fittings. (soak in hot water to soften if to tight) so the 1/2 should fit no problem

Maybe if you tell Danger Den what you're doing they will send you the tubing for free. My 6800GT is looking for quiet cooling and the sooner you can show results the sooner I and others here will consider buying instead of just considering. Maybe they'll send you a TDX (my first choice, the 6000 2nd) to review as well for the series pump tests.

Reviewers that have had trouble with plastic fittings have been advised to use teflon tape around the threads to prevent leaks. The fittings will break if tightened to much. Always use teflon tape, it makes life easier,but make sure no excess hangs off the inside end as it will breakoff and cause a blockage.

Best of luck, I for one, am very interested in this project.