Pump / reservoir advice

DeeFrag

Supreme [H]ardness
Joined
Jan 14, 2005
Messages
8,054
Hi,

I've finally decided to go watercooling for my cpu only, after years of Vantec Tornado torture on a SLK-900.

I need a bit of advise on my pump and reservoir. I was looking at this pump:
http://www.wholesalepumps.com/Product.cfm?DID=7&PID=631

and using an empty Absolut vodka bottle as my reservoir. I think I can drill out 2 holes, one near the bottom and another maybe a third of the way up the bottle and use silicone/RTV sealant to hold the in and out barbs.

Is that pump reliable/quiet? and would a 750mL bottle of Absolut be big enough for that pump? or would I need to get to work on a handle of liquor right now? :D

Also, is there a very large difference in performance between the DD RBX and TDX cpu blocks?

Other components I pretty much decided on, 0.5" Tygon tubing, modified bonneville heater core, 2 x 120mm fans, everything will be mounted in a separate box with the hoses passing through a blank pci slot.

I might eventually put the watercooling box in a minifridge with 2 holes drilled in the side of the fridge for the hoses and sealed around the hoses with Mono Foam or silicone.
 
That pump has very low flow for a 1/2" system. Better suited for a 1/4" system. Any particular reason you want a submersable pump?

The RBX has a little better performance than the TDX but is more of a pain to hook up.

Make sure you buy more tubing than you think you'll need. You really don't need to use Tygon. The Clear Flex is is very good tubing and much cheaper.

The mini fridge idea has been floating around for quite some time and bottom line is that the fridge wouldn't be able to handle the heat. It would be too much of a load on it.
 
The pump can be used as a submersible or inline.
How is 250 GPH low flow? It doesn't produce a whole lot of pressure but there are other models in the product line. The 350 GPH pump seems to be able to far outperform an MCP650 in both flow and head for less than half price. I read a number of posts where people are using the Pondmaster pumps, there weren't any comments about the pump however.

I'll probably get the DDC from Dangerden, but I'm also open to check out some alternatives on the cheap. I don't care too much about price, I care about price/performance.

I've read the RBX is better than the TDX, I'm looking for some hard numbers to quantify the difference.

I've also read most of the posts regarding the fridge idea and it seems nobody has actually done it to see if it would overwhelm the fridge. I think with a large enough reservoir, the amount of heat brought in from one cpu shouldn't keep the compressor running all the time. It's something I'm willing to try eventually.
 
I think you need a pump with more head. 5ft of head just is not good. tdx vs rbx is a tought thing to compare but if you are going with a pump like you picked I would jut go tdx. rbx shines more with a higher head (therby higher flow) pump and a #4 insert. procooling.com has some good info.
Also the mini fridge has been done. I think in each case I read about the system overwhelms the fridge. the compressors on them can't keep up with the load.
 
actually shumph you have it backwards, it is the TDX that only outperforms the RBX by a slight margin when you add in a pump with good head pressure and the #4 or #5 insert. Otherwise the RBX takes the slight performance lead. Overall there isn't much difference between the two.

As for the pump, that pump isn't very good. The head pressure is bad, you really dont want to look at the maximum rated flow of the pump, because that will mean nothing to you in your system. If you get the pondmaster pump I would definitely go for the 350gph version, it has much better head pressure.

Also, I concur with Shumph on the mini fridge. People who come up with these ideas dont realize that refridgerators are not designed to handle a constant heat source....your eggs or juice dont pump out 100W + of heat now do they? If you place a watercooling system, or radiator, or whatnot in a minifridge, the thing will just fail.
 
Thanks for the info on the DD waterblocks.

With the pumps, I know not to look just at max flow since the flowrate is dependent on the pressure it has to overcome. Just wanted some opinions on the pumps whether they are on par performance-wise with the $70+ pumps sold online for watercooling systems.

So the 350GPH or 515GPH pump might be a reasonable pump, for water cooling 1 cpu block, for less than $50?

With the fridge thing, I'll probably do some thermodynamics calculations first. There's got to be a way to get it to work with an optimal reservoir size and shape which would use the specifc heat of water and the deltaT to your advantage. Chemical engineering degree and 4 years of thermodynamics, fluid mechanics, and reactor design ought to finally come in handy.

PS. I read repeatedly on this forum that the Pondmaster pumps are identical to the Danner Mags. (Danner Mag 2 = Pondmaster 250, Danner Mag 3 = Pondmaster 350)
 
danner mag2 is an underpowered pump, as is the pondmaster 250, danner mag3 is a pretty good pump, as was the pondmaster 350 (by the specs at least, I have no experience with it) That is why I recommended you get the 350.

And with the fridge, there is no way around it. Changing the size of the resevoir and whatnot doesn't change the fact that you are pumping 100W of heat into your system, which then has to be cooled by the fridge. Unless you use radiators outside of the fridge to cool the water to ambient first, and then use the fridge to chill the water a bit lower, it simply wont work.

And if you chill the water below ambient, then you have to worry about condensation and what not, by the time all is said and done, you have a lot of hassle and extra money spent, for very little performance gain.
 
DeeFrag said:
With the fridge thing, I'll probably do some thermodynamics calculations first. There's got to be a way to get it to work with an optimal reservoir size and shape which would use the specifc heat of water and the deltaT to your advantage. Chemical engineering degree and 4 years of thermodynamics, fluid mechanics, and reactor design ought to finally come in handy.


What's the difference how big your reservoir is? You're still pumping the same amount of heat into and out of the fridge.
 
If you have a very large reservoir, the rate of temperature change will not be as high as with a small reservoir, so the compressor will not have to work constantly as it will only have to work to cool down the reservoir, then it can turn off while heat is added and the temperature increases again.


Kinda technical stuff:
An extreme example, if you take 1 liter of cool water and add 1 liter of boiling water, the result is 2 liters of very warm water. If you take 1000 liters of cool water and add 1 liter of boiling water, the result is 1001 liters of basically cool water. The amount of heat added to each system remains the same, however the resulting change in temperature is very different because of the ability of water to absorb heat, ie. specific heat.

Very technical stuff:
A large reservoir would allow a greater length of time to warm up but it would also require a greater amount of time to cool down. This shouldn't be a problem as long as the refrigerator cooling system has greater BTUs of cooling energy than the cpu has of heating energy. This requires an energy balance equation which would take into account the shape of the reservoir, since different shapes transfer heat at different rates, as well as the size of the reservoir, the heat transfer coefficients of the: air in the refrigerator, the interface between the exterior surface of the reservoir to the air, the actual wall of the reservoir, the interface between the inner wall of the reservoir to the water inside, and the water. This is when stuff gets kinda complicated ;)

In the end, as long as the fridge has the cooling power to overcome the heat output by the cpu, when the reservoir reaches a temperature which triggers the compressor to kick "on," it will be able to actually drop the temperature in the reservoir to a temp. which allows the compressor to turn off and wait for the reservoir to heat up again.

Conden-what? I live in the desert now where the dew point sits very low. For example, in July when outside air temp. is 94F, the outside dew point is around 50F. Right now, the temp gets to around 70F in the day which makes the dew point about 40F. Indoors are usually much drier, I don't have to worry much about condensation unless I get around freezing temps.
 
DeeFrag said:
If you have a very large reservoir, the rate of temperature change will not be as high as with a small reservoir, so the compressor will not have to work constantly as it will only have to work to cool down the reservoir, then it can turn off while heat is added and the temperature increases again.


Kinda technical stuff:
An extreme example, if you take 1 liter of cool water and add 1 liter of boiling water, the result is 2 liters of very warm water. If you take 1000 liters of cool water and add 1 liter of boiling water, the result is 1001 liters of basically cool water. The amount of heat added to each system remains the same, however the resulting change in temperature is very different because of the ability of water to absorb heat, ie. specific heat.

Very technical stuff:
A large reservoir would allow a greater length of time to warm up but it would also require a greater amount of time to cool down. This shouldn't be a problem as long as the refrigerator cooling system has greater BTUs of cooling energy than the cpu has of heating energy. This requires an energy balance equation which would take into account the shape of the reservoir, since different shapes transfer heat at different rates, as well as the size of the reservoir, the heat transfer coefficients of the: air in the refrigerator, the interface between the exterior surface of the reservoir to the air, the actual wall of the reservoir, the interface between the inner wall of the reservoir to the water inside, and the water. This is when stuff gets kinda complicated ;)

In the end, as long as the fridge has the cooling power to overcome the heat output by the cpu, when the reservoir reaches a temperature which triggers the compressor to kick "on," it will be able to actually drop the temperature in the reservoir to a temp. which allows the compressor to turn off and wait for the reservoir to heat up again.

Conden-what? I live in the desert now where the dew point sits very low. For example, in July when outside air temp. is 94F, the outside dew point is around 50F. Right now, the temp gets to around 70F in the day which makes the dew point about 40F. Indoors are usually much drier, I don't have to worry much about condensation unless I get around freezing temps.

Thermodynamics, my friend. Not technical at all.

You are adding heat to the system at a rate of about 100 watts.
You are removing heat from the system at a rate of much less than 100 watts. In designing a mini-fridge, all one has to do is make sure that heat is pumped out of the system at a greater rate than it enters the system through the [very efficient] insulation. A milliwatt of heat removal would work if the thing was insulated perfectly.

Reservoir doesn't matter jack squat, other than in the case that the cooler is more powerful than the heater at idle, and less powerful at load. Which isn't really a safe situation anyhow, as gamers, folders, etc tend to load their machines for long periods of time.

There are several posts in this thread and dozens of threads out there about using mini-fridges, as it seems the perfect size/shape. All end with the fact that a mini-fridge Doesn't have the heat removal ability to work.

There is no case with a stable, long-term load where a reservoir would improve a heat transfer system that isn't, in the first place, able to remove the heat. (Without taking into account the heat transfer of the reservoir material itself, which is presumed to range from negligible to negative if the reservoir is cooled)
 
Squalish said:
You are adding heat to the system at a rate of about 100 watts.
You are removing heat from the system at a rate of much less than 100 watts. In designing a mini-fridge, all one has to do is make sure that heat is pumped out of the system at a greater rate than it enters the system through the [very efficient] insulation. A milliwatt of heat removal would work if the thing was insulated perfectly.

Most mini fridges actually have a refrigeration power of between 600 to 800 BTU/hr, some have compressors capable of 1000-1500BTU/hr. A cpu heat load of 100W converts to about 341 BTU/hr. Therefore the mini-fridge has the ability to cool down the reservoir since it is able to remove roughly twice the heat. I can see the compressor burning out if the 100W is added directly into the fridge without the temperature change being damped with a sufficient sized reservoir since the compressor would have to turn on and off very rapidly, maybe even run constantly.
 
it doesn't matter how big the resevoir is, there is no "dampening effect" you are still pumping out 100W into the fridge. The fridge may have a rated capacity higher than that, but it is not rated to CONTUOUSLY run with a heat load of 100W...stop trying to make it work, because it wont (well it will for a small period of time and then die)
 
No dampening effect you say? If you were to piss in the arctic ocean would the temp of the entire ocean suddenly rise 90 degrees since you added fluid which is 98.6DegF?

Thermodynamics lesson:
All materials have a specific heat. For water it is 4.1868 J/gDegC.
1 watt is the same as 1 J/s. So 100 W = 100 J/s

If there is 1 gram of water, it takes 4.1868 J of energy (heat) to change the temperature by 1 DegC. If there is 100 g of water it would take 418.68 J of energy to change the temp by 1 DegC.

Assume I have a 4 liter reservoir, that is 4000 mL = 4000 g of water.
Q = cmDT
To raise the temp of the reservoir by 5 DegC it would take:
Q = (4.1868)(4000)(5) = 83736 J
100 W heat load = 100 J/s added to the system.
t = 83736/100 = 837.36 s = 13.956 min

That's roughly 15 minutes to raise the temp. 5 degC in a 4 liter (more or less 1 gallon) reservoir. More water makes time go up, less water makes time go down. Greater change in temp makes time go up, less change in temp makes time go down.

Even if reservoir temp rises 5 deg. in 15 minutes, that temp change must be transmitted to the heat sensor in the fridge by radiation/convection which also takes some time.

This is all scratchpad physics with a lot of assumptions and estimations. I plan to put a lot more thought into the details before I actually try it out.

ps. I don't mean to flame, I mean to educate :)
 
DeeFrag said:
No dampening effect you say? If you were to piss in the arctic ocean would the temp of the entire ocean suddenly rise 90 degrees since you added fluid which is 98.6DegF?

Thermodynamics lesson:
All materials have a specific heat. For water it is 4.1868 J/gDegC.
1 watt is the same as 1 J/s. So 100 W = 100 J/s

If there is 1 gram of water, it takes 4.1868 J of energy (heat) to change the temperature by 1 DegC. If there is 100 g of water it would take 418.68 J of energy to change the temp by 1 DegC.

Assume I have a 4 liter reservoir, that is 4000 mL = 4000 g of water.
Q = cmDT
To raise the temp of the reservoir by 5 DegC it would take:
Q = (4.1868)(4000)(5) = 83736 J
100 W heat load = 100 J/s added to the system.
t = 83736/100 = 837.36 s = 13.956 min

That's roughly 15 minutes to raise the temp. 5 degC in a 4 liter (more or less 1 gallon) reservoir. More water makes time go up, less water makes time go down. Greater change in temp makes time go up, less change in temp makes time go down.

Even if reservoir temp rises 5 deg. in 15 minutes, that temp change must be transmitted to the heat sensor in the fridge by radiation/convection which also takes some time.

This is all scratchpad physics with a lot of assumptions and estimations. I plan to put a lot more thought into the details before I actually try it out.

ps. I don't mean to flame, I mean to educate :)

Nothing you stated here applies directly. You're talking about adding heat to a syatem, whereas your loop will have heat coming in and going out of the system. It needs to reach equilibrium. Regardless of what thermal profile is created in your system (THIS will change with reservoir design, as you change thermal resistance of various parts of the loop), you will ALWAYS have the same heat flux through it under equilibrium conditions.
What happens when my water loop has a CPU putting out 100W of power but my radiator can only disperse 90W? Water temp increases until there is a large anough dT to make my radiator disperse that extra 10W. I can increase my reservoir size, but that will not change this fact. Yes, it will dampen the blow (it will take longer to achieve that dT) but in the end, heat flux is 100W.
 
zer0signal667 said:
Nothing you stated here applies directly. You're talking about adding heat to a syatem, whereas your loop will have heat coming in and going out of the system. It needs to reach equilibrium. Regardless of what thermal profile is created in your system (THIS will change with reservoir design, as you change thermal resistance of various parts of the loop), you will ALWAYS have the same heat flux through it under equilibrium conditions.
What happens when my water loop has a CPU putting out 100W of power but my radiator can only disperse 90W? Water temp increases until there is a large anough dT to make my radiator disperse that extra 10W. I can increase my reservoir size, but that will not change this fact. Yes, it will dampen the blow (it will take longer to achieve that dT) but in the end, heat flux is 100W.

I already stated that a 100W heat load is equal to 341 BTU/hr and most mini fridges can cool 600-800 BTU/hr. This means, whenever the compressor kicks on, it will be able to cool the 341 BTU/hr constantly coming in and still have another 400 BTU/hr cooling ability which will cool the reservoir down to the point where the compressor can turn off and allow the reservoir to heat up again.

The reservoir will always have a constant heat input of the 100 W, but it will also have an intermittent heat input of -200W which means the reservoir will act like a system with a net heat input of +100W then -100W. This translates to heating and cooling cycles.
 
Ok for the last time, mini fridges are not designed to run 24/7, it will burn out.

Second point, even if you were to do this you wouldn't get much of a performance gain at all....why you ask? Do a little experiment for me, take a bottle of warm water slightly a bit above ambient, stick it in the refrigerator and see how long it takes to cool down to the temperature of the other liquids in the fridge. The point of this? The fridge will not cool the water quick enough anyways to make a noticeable impact on the loop.

And before you keep talking your gobbledygook and ignoring our point that NOTHING you do will change the fact that your water is getting 100W pumped into it constantly, take a gander over at the water chiller section of www.xtremesystems.org. There are a few discussions there about using mini-fridges as a chiller, the conclusion in all of them (even by people who have done this before, and people who have done it with the fridge packed with a huge resevoir)? Well the conclusion was that the mini-fridge did not give any real big performance gains, and it required cooling of the fridges compressor to keep it from dieing.

Simply put, mini-fridge = not good for water cooling.
 
Everyone has their own opinions.

I have mine. I'm not forcing anyone to go try out my ideas, I merely said I plan to try this out in the future. If you don't agree with my ideas, that's fine. Either ignore it or present why you think it won't work with science instead of so-and-so said blah. I'm running out of things to waste money on so I figure I'll start on watercooling :D

And that gobbledygook is also known as physics.
 
DeeFrag said:
I already stated that a 100W heat load is equal to 341 BTU/hr and most mini fridges can cool 600-800 BTU/hr. This means, whenever the compressor kicks on, it will be able to cool the 341 BTU/hr constantly coming in and still have another 400 BTU/hr cooling ability which will cool the reservoir down to the point where the compressor can turn off and allow the reservoir to heat up again.

This is fine by me, I'm just trying to turn you away from the idea that reservoir design is going to change the workload on your phase change system.
 
Back
Top