• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

What pump to use?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Chill Dude

New Member
Joined
May 10, 2006
Hey folks,

I'm setting up a new wc system and there's some confusion on my part regarding the appropriate pump power. I realize head height is the more important spec, but this relates to flow through the blocks. On the one hand, I read that the higher the flow rate the better, because you need to have the coolest water in the blocks at all times. OTOH, I read that you can't use TOO high a flow rate, because that wouldn't allow the water enough time to "wick away" the heat from the blocks. :confused:

Here's what will initially be loading the pump:
1 Swiftech Apogee
1 Swiftech MCW60
1 CoolRad-12T
An estimated 3.5 feet of tubing (masterkleer 1/2" ID)
All above using 1/2" barbs; pump may require reducers, however, depending on model.

Here's what will probably be added later (or sooner):
1 Swiftech MCW30
Additional tubing for this block.
Yeah, I know... I should install ALL the parts I will ever use right from the start, so I don't have to break it down and spill fluid, etc. But, my Zalman 47J sink is working nicely on the NB for now.

Here's what could be added later if I go overboard:
Water temp sending unit
Water pressure sending unit :rolleyes:

So tell me, what sorta pump power should I be looking for which will be adequate for future loads w/o being too strong for what I already have?
I've read that the typical block presents the equivalent of 18" line resistance. Is this spec correct, and are my blocks be considered "typical" nowadays? Also, about how many inches would my CoolRad present? I need to know for the purpose of calculating final GPM from pump head specs.

Also, I'd welcome opinions on which pumps are quietest, as noise reduction is my primary objective for this Chieftec Dragon case.

Lastly, are all made-for-pc fluids created equally? That is, do some have higher viscosity which in turn requires a more powerful pump to achieve the same flow rates? I ask because I'm considering using a non-conductive formula which includes the lubricating, anti-algae, etc., additives.

Thanks.
 
I personally chose the Swiftec MCP350. It is small and easy to mount just about anywhere. It is fairly quiet but does vibrate a bit (like most pumps would I suspect), I have mine isolated in a closed-cell foam pocket to kill the vibration and all that it makes for noise is a very low hum.

Also it is possible to buy a modded top or mod the top yourself to increase the flow rate to be comparable to the MCP655 (140 lph modded vs 75ish unmodded i beleive, but dont quote me on those numbers)

Edit: I would also think it would have enough head for the components you spec'd...
 
popular pumps dc pumps:
D5
mcp350
aquaxtreme 50rz

Since you have so much stuff I'd go with a D5, but also a bigger rad. If you want to cool your CPU, GPU, and NB without loud fans I'd want atleast a 240mm x 120mm. Just my opinion though.

In terms of noise, mcp350 is the quietest. Only pump that rivals it in noise is the eheim 1048. but that is a very weak pump and brand new its about the same price.


When looking at pump specs try to find a pump with specs closest to this:

cathar said:
I still hold great confidence in saying that the "ideal" water-cooling pump, that is a realistic and feasibly built possibility (based on the MCP600), would be:

6-8mH2O (20-26') of peak pressure head
12-13LPM (3.2-3.4 US GPM) peak flow rate
25-30W nominal power draw (when in operation)
12VDC
Separated Magnetic Armature design (~16-20W of heat added to the water)
Centrifugal pump
from here:
http://forums.procooling.com/vbb/showthread.php?t=10825

Currently I don't think there is a pump that meets all those specs. MD20rz is close but consumes and dumps more heat into the loop. D5 and mcp600 consume less but about half of the peak head. C-systems has a pump design that does actually exceed the specs. They are just a manufacturer and no distributers have ordered any to sell so I don't think c-systems has made any past prototypes.

Cathar has some good graphs on flow of many popular pumps. The 20rz's graph is pretty close to ideal.

If you decide to experiment and search for a new pump by all means please post your results :D

As for the non-conductive fluids, they are mostly a warm blanket signifying nothing. They all become conductive after some period of time. Lots of people have had spills here and no damage, yeah yeah you can point at me. Just use good clamps,fill/bleed your loop, and leak test it for a day or two outside of your box.
 
The blocks only allow a certain amount of water through them because of how big the inlets/outlets are, so if its max GPH is 150 with say a pump with 10 feet of head, then only 150 GPH is going to be allowed through, so there's no such thing as having a pump that goes too fast, mine is 675 GPH @ 1 foot lift and has a head height of 13.5 feet.
 
MCP350/DDC 1/3 heat dump of mcp655, most quiet between mcp655 and aquaexteme.

Moded MCP/DDC same spec as unmodded one, same noise level, double flow. Depending on how restrictive loop is it might perform better or worse than mcp655.

Aquaextreme 50z - 1/2 heat dump of mcp655/d5 though runs little noise. The noise is more anoying than mcp655/d5

DDC Plus - 1/2 heat dump of mcp655/d5. Noise level from reading is similar to mcp655/d5 at setting 3 of 5. This stronger version of mcp350/ddc.

Modded DDC Plus- I don't know that one.

MCP655/d5- Higher heat dump than ddc/mcp350 and aquaextreme. It might seem quiet depending hardware configuration. Also depending on fan speed ran. Many user say it is quiet but the opinion is subjective. From my own experience mcp350 is much more quiet than mcp655
 
Thanks for the tips guys!

I checked out the models u recommended and it does seem the MCP350 offers the best bang for the buck in a low-noise, low-heat unit. Size really isn't a concern, as there's tons of room left in my Dragon case. This model surely packs an impressive punch from such a tiny body, though.

Since the 350 apparently does better in higher-restriction systems, I wonder whether 7/16" ID tubing on 1/2" barbs -- or even down to a 3/8" scheme -- would better serve me by providing a higher backpressure for the pump. Or, would the smaller tubing (particularly 3/8") cut point-to-point volume such that the end result would be pretty much the same as using 1/2" ID tubing on 1/2" barbs at a somewhat reduced flow rate?

The consensus of the thermodynamic and mechanical engineering crowds seems to be that for best heat exchange, a high flow rate through the block and a low flow rate through the radiator is best. I wonder if the "sweet spot" could be obtained with a modded 350 given the components I already have in my possession and by merely using the right tubing diameter and lengths.

Since the 1/2" modded inlet has such a drastic improvement on performance, wouldn't it stand to reason that even larger inlets (3/4, 1" -- if the cap would allow for 1") would be progressively superior? Or would the point of diminishing returns be reached at anything beyond 1/2"?

Is the low noise produced by the 350 more of a soft buzzing or hum? I've read about the MCP655 (D5) and the D4 producing a hissing and/or high-pitch whine. This is entirely unacceptable. I have tinnitus and hissing, whistling, whining or other high-pitch noises in a certain frequency range, such as from motors or singing flyback xformers drive me nuts. It can even precipitate episodes of high-freq pulsatile tinnitus (trust me, ur glad u don't know what that is), so it can make using an air-cooled computer pure torture, especially with extended periods of use. Low-decibel levels aren't much less irritating, nor is distant proximity. It's the frequencies I'm sensitive to, so if I detect them even at long range and low levels, it causes distress.
By contrast, low-pitched hums (as from many pumps) don't seem to be nearly as irritating to me, provided they aren't high-decibel. I'll probably have to ditch this Western Digital HD in favor of a Seagate due to the WD's bearing whine.

Am I correct in assuming that virtually any aquarium (powerhead) or pond/waterfall pump is going to be noisier than these DC pumps? Hard to believe, as they have the same number of moving parts -- one impeller. Very few aquarium pumps have noise level specs on the manu sites, so one has to buy these not sight unseen, but rather sound unheard.

Swiftech spec's the 350 at 8.3W nominal @ 12V. I was under the apparently mistaken impression that the DDC was identical to the 350, but the DDC is spec'd at 10W (older) and 18W (Rev2). There was apparently a problem with the 10W not starting up properly; upping to 18W supposedly fixed this. Does this mean I should buy an 18W DDC because any MCP350 will be 8.3W and might have the stalling prob? If there's an 18W Rev2 of the 350, Swiftech isn't mentioning it on their site. So just which is better, higher wattage DDC to ensure against stalling while adding more heat to the loop or lower wattage 350 and pray it doesn't stall on me? Or, does the 350 mod help resolve the stalling issue?

Thanks for your attention and input. It's helping a lot! :thup:
 
Chill Dude said:
a high flow rate through the block and a low flow rate through the radiator is best. :
Rubbish! Think of it this way...if you double or halve the flow rate, in one hours time, the water will be spending the same amount of time in the block(s)/Rad. In one hour, traveling through the loop it will spend either twice the time, half as often or half the time, twice as often! Flow rate changes have no effect on how long the water is in the block(s)/Rad. The only way to get more or less time in the block(s)/Rad is to increace/decreace the total amount of water in the system. You can look at this this way: with more/less water each molicule will go through the block(s)/Rad less/more often per hour. But the amount of water in the system has no effect on cooling efficiency since this is an equilibrium situation. The water is going to get to an equilibrium temperature no matter how much of it there is. Higher water content will effect how long it takes to reach this equilibriun, but not how much heat is transfered once equilibrium temperature is reached.

Flow rate does affect the efficiency of the heat transfer though, it affects heat transfer right at the water/copper boundry. The more turbulance the better the transfer (with a big change in efficiency at the transition from laminar to non laminar flow). More turbulance right at this boundry is what is important and generally, higher flow rater means more turbulance. The idea of getting the "old, hot" water out of the way and replacing it with "new, cool" water applies to this surface boundry layer only. How fast the water is moving just above this boundry layer is of little import.
 
Last edited:
I kind of figured you would want slower flow through the rad. Give the water a little extra time to cool down under the fan.

Say I had a room with a huge box fan set up in the middle (like the radiator). Ive been out in the hot sun in the garden and backyard (kind of like our hot CPU and GPU blocks). Now if I run through the room quickly and back out to the garden and to the backyard come through the circle again quickly, I'm just haulin *** past the fan. True I will pass by that fan alot but there is no way that fan is gonna cool me off that good. But if I kind of slow down when I come in the room and just walk by the fan instead of flying past, Im going to cool down a little better.
 
Speedmonkay said:
I kind of figured you would want slower flow through the rad. Give the water a little extra time to cool down under the fan.

Say I had a room with a huge box fan set up in the middle (like the radiator). Ive been out in the hot sun in the garden and backyard (kind of like our hot CPU and GPU blocks). Now if I run through the room quickly and back out to the garden and to the backyard come through the circle again quickly, I'm just haulin *** past the fan. True I will pass by that fan alot but there is no way that fan is gonna cool me off that good. But if I kind of slow down when I come in the room and just walk by the fan instead of flying past, Im going to cool down a little better.
Unlearn this now please. It is wrong. Period.
 
Firstly, Chill Dude, welcome :) :welcome: Good to have ya here.
For rad specs, look here. A lot of folks swear by 7/16 ID tubing because that matches the ID of 1/2 OD fittings and causes less resistance.

Since the 1/2" modded inlet has such a drastic improvement on performance, wouldn't it stand to reason that even larger inlets (3/4, 1" -- if the cap would allow for 1") would be progressively superior? Or would the point of diminishing returns be reached at anything beyond 1/2"?
Flow may improve but head would surely suffer. By how much will only be seen when/if someone trys it.

Good luck and post some pics for us when you get it afloat?


Speedmonkay said:
I kind of figured you would want slower flow through the rad. Give the water a little extra time to cool down under the fan.

Say I had a room with a huge box fan set up in the middle (like the radiator). Ive been out in the hot sun in the garden and backyard (kind of like our hot CPU and GPU blocks). Now if I run through the room quickly and back out to the garden and to the backyard come through the circle again quickly, I'm just haulin *** past the fan. True I will pass by that fan alot but there is no way that fan is gonna cool me off that good. But if I kind of slow down when I come in the room and just walk by the fan instead of flying past, Im going to cool down a little better.


I'll bite. Using your hypothetical analogy: The rad isn't a 'room with a fan'. It is a long, fanned hallway wherein you spend much more time in front of the 'fan'. By saying you are just running past the fan isn't exactly accurate.
 
Dice said:
I'll bite. Using your hypothetical analogy: The rad isn't a 'room with a fan'. It is a long, fanned hallway wherein you spend much more time in front of the 'fan'. By saying you are just running past the fan isn't exactly accurate.

Ah ok I gotcha .... more like a little mouse running through a extremely long zig-zaggy maze under a giant box fan.

I reread Billb's post a couple more times and understand it a little more now. Sorry I do better understanding some stuff with cheap analogies.
 
Speedmonkay said:
Ah ok I gotcha .... more like a little mouse running through a extremely long zig-zaggy maze under a giant box fan.

I reread Billb's post a couple more times and understand it a little more now. Sorry I do better understanding some stuff with cheap analogies.

No need to apologize, analogies are excellent learning tools. :beer:
 
Dice said:
Firstly, Chill Dude, welcome :) :welcome: Good to have ya here.
For rad specs, look here.

Thanks, this is a great forum.
I actually had read that pdf prior to getting my rad. That's one reason I ended up with the CoolRad. It seemed to have a good price/performance ratio -- an especially good price, considering the $30 includes the fan shroud. Anyone looking to buy one should beware, however. CoolingWorks' internal quality control is pretty sad. I had to send the first unit back because construction was very shoddy -- rust and paint bubbling/missing, tubes completely soldered shut (!!) and solder blobs rattling in the bottom cap. So that $30 turned into $38 ($8 shipping back to dealer), which is still a good deal if it works as advertised. The second unit appears in far better shape. Credit to the supplier for actually opening it up and visually inspecting it, rather than simply tossing it in a box and sending it to me. Should I mention the supplier's name? hmm.. Suffice it to say, I don't blame them for the rad's probs and was overall impressed with their customer SerViCe. :)
A more thorough evaluation to be performed once I get the new pump.

Dice said:
Good luck and post some pics for us when you get it afloat?

Thanks. I'm modding my black Dragon case in a Borg theme and planned on fabricating a custom reservoir and using green dye -- think Borg maturation chamber. :cool:
For a flow indicator, I thought about anchoring a Borgified infant model with clear (or translucent green) monofilament line in several directions -- if the infant is restless, there be flow here! And who wouldn't be restless with all those Borg implants in 'em? :D
Has someone does this already? I hope not. Just when ya thought u had an original idea. :bang head
 
billb said:
Rubbish! Think of it this way...if you double or halve the flow rate, in one hours time, the water will be spending the same amount of time in the block(s)/Rad. In one hour, traveling through the loop it will spend either twice the time, half as often or half the time, twice as often! Flow rate changes have no effect on how long the water is in the block(s)/Rad.

Ok, but please continue to resist shooting the messenger. :)
I was only relaying what I had read in lengthy debates in some engineering groups. The discussion was among persons with alleged advanced degrees in the relevant fields. I am not among them, as electronics is my forte.
They did mention numerous times how counterintuitive their assertions were, but stood by them.
Your explanation is a logical one and some did argue in favor of it. It was the majority of these supposed engineers who contended the opposite. And I should have said majority earlier instead of consensus. And, btw, that was only the majority in those particular threads, not necessarily the majority view among the discipline as a whole. For all I know, it could have been the same people using different names across different groups/forums in support of their favorite arguments.

Some of it was rather math intensive, and as I was next to REM state at the time, I didn't bother trying to absorb all of that. I just went by the final conclusions they drew.
Maybe they were daydreaming of some sort of pulsed, inline, ancillary pump assist configuration where u could change rate & pressure anywhere in the loop on demand, coupled with accessory sinked reservoirs cooled by additional fans, etc etc. Who knows. I know electronics, not water -- though I do know the two don't usually mix well. :D

At any rate (pardon pun), I appreciate your effort and have assimilated your tutorial.
 
Chill Dude said:
Should I mention the supplier's name? hmm.. Suffice it to say, I don't blame them for the rad's probs and was overall impressed with their customer SerViCe. :)
A more thorough evaluation to be performed once I get the new pump.


By all means comment on your experiences with a merchant. That is good stuff to know. There is ever a subsection in which to do it. Good to hear that they fixed the problem when it was brought to their attention. That, in itself, is the definitive measure of a business's customer support practices.

At any rate (pardon pun), I appreciate your effort and have assimilated your tutorial.

Well, you had better have. Resistance is futile.:cool:
 
Nature has a funny way of making theory not work so well in real life. They may be correct about low flow being better in things like Nuclear power plants (just a guess here) but in a PC water cooling system, Head pressure and flow are king and the more the better.
 
Chill Dude said:
Ok, but please continue to resist shooting the messenger. :) ...
At any rate (pardon pun), I appreciate your effort and have assimilated your tutorial.
No criticism intended, it's a common nOOb mistake. Glad to help!
 
Back