Radiator Sandwich / Stacked Radiators / "SLI-Radiators" Test results

[HESmelaugh]

@Eddie:
Yes, I used one set of fans between the rads. No other setups tested yet.

@Naekuh:
Are you going to sandwich 'em?
Anyhoo, the I think of the whole radiator stacking as something more for those among us who don't have a money-tree in the back yard (or a back yard, for that matter)... :-)
Just kidding.

[STEvil]

Too bad the radiators dont have holes for fittings on top so you could toss the 90's out.

Interesting testing

[NaeKuh]

@Eddie:
@Naekuh:
Are you going to sandwich 'em?

yup. double stacked san ace core.

[sirheck]

I KNEW IT

Cant wait for the Swiftech version.

[Factotum]

wow, HESmelaugh killer PA120.3?
nice test. i have mcr320 and mcr220, i need to try to do the same

[Martinm210]

The other test you could try is running the radiators in parallel instead of series. This is essentially how the double thickness rads run and why they generally have lower pressure drop.

Running them in series should perform better with the higher flow rate, but parallel might be an option if restriction was a concern.

Running two separate rads should be worth roughly half the previous water/air delta or a 50% gain, but 30% gain for only 3 fans is a pretty healthy gain.

[Utnorris]

What actually would be interesting is to see if using shrouds has any effect, i.e. Rad - Shroud - Fan - Shroud - Rad. I wonder if the shrouds would focus the air flow better in this scenario.

[karl_eller]

Wow, those are fairly impressive results. It'll be good to see how the Rad > Fan > Rad (or Rad > fan > Rad > fan) setup goes compared to two separate radiators, but this certainly looks hopeful for a lot of people who can't fit two 120x3 radiators in a case...

Thanks a lot for the tests, and you're giving me ideas *credit card commits suicide*

Eller

[CedricFP]

I'm planning on stacking 2 120.1 rads.

[moogle]

My thoughts exactly. Not exactly fair testing two rads against one alone.

I'm confused, how is it unfair? Obviously in a test to see which single rad cools better you wouldn't do that but this one is just to show you can get the same cooling as PA120.3 for cheaper right?

[HESmelaugh]

Ok, I added testing the rads in paralell to the list of things that need testing.

I'm also thinking of using a different testing method. I would like to have a heat source inside the water (in a reservoir). It should be possible to get a more constant heat output and more precise results running something else than a cpu and gpu right?

[The-Fox]

I was surprised myself that the sandwich beat the PA. Just to make sure, I did two test-runs each.
And I have to admit that I really don't have enough data to be able to understand how or why the sandwich is so effective.

The reason is pretty obvious.
The second rad adds more surface area for cooling and despite getting hotter air at the intake with lower CFM, it's still able to dissipate more heat than a single rad.

Btw, I had this idea in my mind too some time ago, but the wiser choice for this sandwich setup will be the XSPC rad, the R120-T (V)


with the horizontal ports which has less impact on flow and easier for setup, can have you choose between running two different loops, one loop in parallel or one in serial choices are fun

[Martinm210]

Ok, I added testing the rads in paralell to the list of things that need testing.

No need to test if you're not interested, it's a tremendous amount of work for maybe no difference at all...just a thought. I'd probably be more interested in comparing the same with more fans, something like a PA with 6 fans and the sandwich with 6 fans or something like that. Whatever you're willing and have time/interest to do.

I'm also thinking of using a different testing method. I would like to have a heat source inside the water (in a reservoir). It should be possible to get a more constant heat output and more precise results running something else than a cpu and gpu right?

You can use the water heaters like I was doing if you want, but you'll have to take the heater apart and solder the thermostat solid. There is a little metal spring that pulls the contact away when the temperature is high. I just solder that contact together and slide it all back together again. The biggest pain is getting the thing apart.

for a triple rad I would use at least a 300W heater, two for a total of 600W is even better if you're trying to split hairs.

The hardest part of the radiator testing for me was waiting for equilibrium and ensuring a constant ambient. It doesn't matter what the ambient is, but it needs to be holding constant. If it's rising or lowering, you're delta will be off a bit. Unfortunately it usually takes a couple of hours of testing before you've warmed up the testing area to a point where you might have a fixed ambient.

Also lot of air in sensors helps, you probably can't have too many and they all need to be in the same exact place for each test because you will get radiant heat from the radiator. Probably best if your air in sensors can be several inches away from the radiator if possible.

Finally the fan RPM is another pain. As the fan warms up the RPM will increase. You have to let the fan warm up for about 10 minutes at the desired RPM or constantly monitor and adjust.

Rad testing is one of the more difficult ones I've found, it's a PITA. Oh yeah one more, don't change the air flow in your room. A small think like a fan running in the room will change your results I noticed this when I got my air conditioner and used the fan to circulate air in the room better. I had to retest anything I wanted to use for a comparison.

[sirheck]

Yet another reason you are highly regarded Martin.

[HESmelaugh]

Thanks for the info, Martin.
I was thinking about getting two heaters with 300W each. I still need to make a reservoir where I can submerge them, though. I already use multiple air in sensors and have a rig where their distance to the radiator is always fixed.

I probably won't be able to get results as precise as the ones we got used to from your testing, but an improvement over the setup I have now should be easily achievable.

Changing fan rpm is something I have also noticed but don't have a good way of controlling yet.

[HESmelaugh]

I just had an idea. Maybe there could be an easy alternative to double-threaded radiators that would allow us to connect any two radiators together.
I tried to illustrate with a quick sketch:



There is something similar available for connecting two GPU waterblocks. Example: Link
All that needs is another G1/4" thread in the middle and you've got a T-line for connecting two radiators in paralell. The radiator-threads could then simply be facing each other.
What do you think? Would this work?

[Whilhelm]

Yeah that would work great as long as both radiators matched. Running the two in parallel would probably change the temps very slightly as you wouldn't have one rad with slightly warmer water then the other.

What about increasing the spacing between the radiators taking an old fan and cutting out the fan and just use the frame. and then you could have a set up like this.
___________________
[_______Rad 1_______] Outlet Rad (Cooler water)
[_______Fan 1_______] *Possibly switch fan 1 and spacer depending on results.
[______Spacer_______]
[_______Rad 2_______] Inlet Rad (Warmer Water)
[_______Fan 2_______]

with the air flowing downward through Rad 1

I wonder how that would perform on a dual or even triple 120 rad setup.

[dionysio]

with the koolance sli connectors and the cube like -| |- only for an 38mm fan though. this is the neatest way to stack rads in parallel...

[billb]

If the flowrate is to your concern and you want to avoid elbow fittings, how about turning one of the radiators 180 degrees, so that the fittings sits opposite of one another.
Thats how I have my 2 rads in my TJ07.
Something like this:

How about just some shrouds.

[HESmelaugh]

Holy banana! Where are you going to fit that? ^^

Honestly though, if you've got enough space for that construction, you might as well equip both rads with fans with the airflow pointing away from the center. That way, both rads get fresh air.

[inCore]

Yes, this is true. If I were to stack radiators it would be somewhere like the lower compartment of a P18x series or perhaps TJ07.

Anyways, just popped into the thread to congratulate and thank you for your testing HESmelaugh, keep up the superb-ness.

[Boogerlad]

I just had an idea. Maybe there could be an easy alternative to double-threaded radiators that would allow us to connect any two radiators together.
I tried to illustrate with a quick sketch:



There is something similar available for connecting two GPU waterblocks. Example: Link
All that needs is another G1/4" thread in the middle and you've got a T-line for connecting two radiators in paralell. The radiator-threads could then simply be facing each other.
What do you think? Would this work?

Wouldn't that just be a T-line?

[Martinm210]

Thanks for the info, Martin.
I was thinking about getting two heaters with 300W each. I still need to make a reservoir where I can submerge them, though. I already use multiple air in sensors and have a rig where their distance to the radiator is always fixed.

I probably won't be able to get results as precise as the ones we got used to from your testing, but an improvement over the setup I have now should be easily achievable.

Changing fan rpm is something I have also noticed but don't have a good way of controlling yet.

Perfect!

There's no right or wrong way IMHO, just things to watch for and try to keep the same across comparative tests.

For Fan RPM, I've done it a couple of ways:
-Variable PSU and monitored RPM on the crystalfontz, could also be monitored on any MB fan header. You just need to splice off the yellow and black wires for the RPM to the fan header and use either the variable PSU or fan controller to tweak the voltage until the RPM meets the target. Plus or minus 20RPM or so is about as good as it gets.
-Crystalfontz percentage controlled.

Fan RPM is hard though, slight variations in household voltage will still make it go up an down. Something like a line voltage conditioner might be the ticket for fixing that little issue. In the end, I just decided to log the RPM.

Anyhow, just a few challenges I had.

I think it would also be interesting to try and measure heat dissipated the way the coolingmasters guys do with water in vs. water out and flow rate. That might be a means for getting around all the issues with ambient changing and such, but you'd need to really beef up the precision of the sensors to measure that small of a delta. I had thought about trying to rig about 5 sensors per side and average all 5 sensors to improve resolution. I think you probably need relative accurace down to .01 or so.

Anyhow, the stack rads is pretty neat, keep up the awesome work!

[HESmelaugh]

Wouldn't that just be a T-line?

Not exactly. All the T-lines I know have static threads, which makes it impossible to connect a radiator on either side. What we would need is a T-line with rotary threads.
In case you know anyone who makes such T-lines, please let me know!


Martin: I'll try warming the fans up for a few minutes and then logging the rpm. Might turn out to be the easiest solution. Though I'm afraid I will have to get out a soldering iron (or my purse) to modify/get new testing equipment at some point...


InCore: Thanks!

[Serpentarius]

there's been in my mind bout ... 2 MCR320 vs 1 PA120.3

2 MCR320 vs 1 PA120.3
(34cm + 34cm width) vs (62cm width)