DIY Corsair H50 killer...?

[Billb2]

Its not a linear relationship though. C/W does change with flow rate, but the scale is not linear.

edit: I test radiators @ 1.5GPM since that is the point of diminishing returns. However, I think its time I hike up my boot straps and start doing variable flow rate testing on radiators... man that is going to take even longer. 😱

Been waiting for you to do that. Your testing shows which rad is "best", but can't be used to predict actual performance unless the proposed system is designed for same 1.5gpm you use.

And I liked the old website better, but I guess it just takes time to get used to change.

 

[Billb2]

Well pretend each time they go around they drop off a passenger.

No. The faster car only has enough time to drop off 1/2 of a passenger each lap.

You assume that, on each pass, no matter how much time the water spends in the rad, it gives up the same amount of heat. If that were true we'd all be using 50gpm pumps at 200psi and rumming our CPUs/GPUs at ambient.

 

[Billb2]

it seems to me that doubling the flow rate has loosely the same effect as using a rad with double the surface area. ...or am I missing something?

Yeah, you are. When you double the flow rate, you halve the time the water spends in the rad. It doesn't make any difference how many "feet" the water is in the rad, all that matters is time.

That being said, Skinnee's going to show us the effect of turbulance.

 

[skinnee]

Been waiting for you to do that. Your testing shows which rad is "best", but can't be used to predict actual performance unless the proposed system is designed for same 1.5gpm you use.

And I liked the old website better, but I guess it just takes time to get used to change.

I don't even like the "best" thing, each radiator has its sweet spot for fan speeds and we'll see the same thing with flow rates. Not sure if I am going to test each fan speed with variable flow... the test time would just kill me.

Website is going to take another design revision or two before I'm happy with it. Had to change from the static HTML though, the maintenance was killing me. Thanks for the compliments though, cheers.

 

[aigomorla]

No. The faster car only has enough time to drop off 1/2 of a passenger each lap.

You assume that, on each pass, no matter how much time the water spends in the rad, it gives up the same amount of heat. If that were true we'd all be using 50gpm pumps at 200psi and rumming our CPUs/GPUs at ambient.

because the car is only dropping off 1/2 its load, water temp is thereby lower then if u had it on a lower flow system.

Bill, must i pull the EQ for you?

300W is the capacity of water for 1C @ around 1gpm.
350W is the capacity of water for 1C @ around 1.5gpm
400W is the capacity of water for 1C @ around 1.75-2gpm.

Its math bro... you cant argue with physics..

Your limitations will be at the waterblock + rad + ambients.

 

[smackababy]

I just want to express my remorse that my employer has decided it appropriate to block skinneelabs...

 

[skinnee]

really?

 

[aigomorla]

really?

all the PORN... errr... sexy photo's of water blocks... you have..

To an average IT person, its like girls gone wild.. :hmm:

 

[v8envy]

No. The faster car only has enough time to drop off 1/2 of a passenger each lap.

You assume that, on each pass, no matter how much time the water spends in the rad, it gives up the same amount of heat. If that were true we'd all be using 50gpm pumps at 200psi and rumming our CPUs/GPUs at ambient.

You know what, you may be on to something here. The obvious optimization to a ridiculously high flow & pressure system is a really large, winding automotive radiator. Granted that would never handle anything near 20 never mind 200 psi (it's a huge amount of si), but high flow would not be a problem.

I still want to use an automotive radiator as a side panel of my case. Some day I'll yoink a 1973 GMC truck radiator from a junk yard and realize my WC dreams.

 

[smackababy]

really?

It only tells me access denied due to local policy and that the website is catagorized as "none"...

But, they block so many sites, its a wonder they haven't blocked the entire internet. If I google something, 90% of the results are blocked when I click on them.

 

[Schmide]

No. The faster car only has enough time to drop off 1/2 of a passenger each lap.

You assume that, on each pass, no matter how much time the water spends in the rad, it gives up the same amount of heat. If that were true we'd all be using 50gpm pumps at 200psi and rumming our CPUs/GPUs at ambient.

I'm sure there is a point where the curve levels out, but it never reverses its self. (i.e. there is no point where a lower flow system will out perform an high flow system) That being said the effect of greater flow has a marginal effect on cooling. Think single digit percent differences for avg flow to really good flow, certainly really slow flow system will drop in performance exponentially.

You may want to read

Cathar's effect of tubing sizes
Swiftech MCR320 Review
Martins Radiator Estimators

 

[nitromullet]

So, I've been just lurking this thread for the past few days and doing my research. Here are all the parts I've picked out for the moment, and will probably pull the trigger on this. If anyone with more experience than me (pretty much everyone here) has advice on a better value/price component, let me know.

Main components:

XSPC RX120 Extreme Performance Radiator - $55.45
Enzotech Sapphire CPU Water Block Rev. A w/ 1366 bracket -$39.48
Swiftech MCP655 12 VDC Pump Liquid Cooling System - Retail - 74.99

Accessories (~$20):

Danger Den Brass Tubing Plug
Polypropylene "T" Fitting 1/2"
Water Cooling Funnel
Swiftech Tubing 7/16th inch (ID) X 5/8 inch (OD) x 10ft
Swiftech HydrX Coolant

After shipping and a $10 off newegg coupon code the total comes to: $203.18

I decided to go with a T line only instead of a T line + res (as mentioned in the newbies guide) mainly because I'm trying to keep costs down, and I really can't find a decent affordable res.

I did find this pump: OCZ Hydropulse Water Pump 800L that works with this res: EK-Reservoir Combo DCP 4.0. I'm sure this pump would be fine for my current plans, but my concern is that it won't have as much growth potential as the MCP655. My plan is to probably pick up a Primochill Typhoon III (when they are actually in stock) to pair up with the MCP655, and build up the loop with a bigger rad and more blocks.

 

[tno]

I think this thread just turned into a physics class. I'm sure of it. I even got dizzy, drew pictures of robots killing dinosaurs in my notebook and then took a nap. Honestly though this is a great thread and is giving me a lot of room for thought, being that I'm looking to switch to a non H50/Dominator set-up. So, to the OP, great question, saves me the post, can't wait to see how this works out for you.

 

[nitromullet]

I think this thread just turned into a physics class. I'm sure of it. I even got dizzy, drew pictures of robots killing dinosaurs in my notebook and then took a nap. Honestly though this is a great thread and is giving me a lot of room for thought, being that I'm looking to switch to a non H50/Dominator set-up. So, to the OP, great question, saves me the post, can't wait to see how this works out for you.

I went ahead and ordered the parts I listed. I'll definitely update with experiences/results.

I did find a few other kits that would probably be closer to an "H50 killer" based on cost. Of course, all are still more expensive.

PTS CoolKit Lite - Rev.3 - $144.99
Danger Den Intro Kit - Version 2 with T-Line - $163.47 (MC-TDX block for 1366)

A big part of my criteria was the single 120mm rad because that is all I can accommodate without cutting up my case. I didn't want to leave any scars should I decide against this in the long run.

 

[Yellowbeard]

After all is said and done, I wish you could do a head to head comparison on the same system. Beating the H50 with your setup should not be too hard but, at almost 3X the cost, I don't think "killer" or "destroy" will really apply.

Have fun.

 

[nitromullet]

After all is said and done, I wish you could do a head to head comparison on the same system. Beating the H50 with your setup should not be too hard but, at almost 3X the cost, I don't think "killer" or "destroy" will really apply.

Have fun.

Well, there are always alternatives... There's the $144 kit from Petra's which is more than the H50, and the $50 Masscool 7WA002L which is less... I'm sure each solution has it's advantages and disadvantages.

edit: just ran Prime 95 for a little over an hour with my TRUE (push/pull ~1550rpm), i7 920 @ 3.66GHz vcore 1.250v, HT enabled, ambient 20C, max cpu core temp 79C, usually hovered between 76-78C on cores 0 and 1, 74-76C on cores 2 and 3.

edit 2: Tried Prime 95 at 4GHz (20 x 200, vcore 1.375) with the TRUE, and was seeing temps as high as 94C on core 1. I'm not really comfortable with those temps, so I didn't run this setup very long.

...will update when I get the wc loop installed.

 

[GAM3RIG]

@ aigomorla- take a look @ this:

isn't that sexy MODDED WC H50? 😉

 

[aigomorla]

@ aigomorla- take a look @ this:
http://img15.imageshack.us/img15/995/img1175rp720x5401.jpg
isn't that sexy MODDED WC H50? 😉

:X

no comment..

 

[Rubycon]

@ aigomorla- take a look @ this:

isn't that sexy MODDED WC H50? 😉

How does that tiny pump push water effectively through all that?

If you're going through all that trouble perhaps you should start from the ground up with the right stuff. 😉

 

[aigomorla]

H5O Killer here:

😛

 

[Schmide]

@ aigomorla- take a look @ this:
isn't that sexy MODDED WC H50? 😉

Is that yours? Because you know that radiator is aluminum and you seem to have very little if any ethylene glycol in there. Without a corrosion inhibitor, three months?

 

[Rubycon]

Where's the rest of it? The block costs more. 😛

I want to see the results of an H50 with its radiator in a bucket of ice water.

You know what I just may try it since I have an H50 sitting in the box here and plenty of ice (and buckets!). This should be fun and not cause a disaster if something goes really awry. 😉

What is the point some of you may be asking? Well simple! It would demonstrate that increasing capacity of the radiator would result in lower cpu temperatures under high load. Of course a larger radiator means more fluid and this does not simulate that. I suppose it would simulate someone putting the radiator outside when its really cold out hehe. You get the idea though.

 

[aigomorla]

yeah but at the cost of eatting flow...

gives you an idea how flow works...

You see why we all advise in your flow being around 1gpm ~ 4-5 Lpm.

 

[Rubycon]

You know the H50 being a packaged solution is just begging for experimentation. How about a thicker 120mm radiator? Double 120 radiator?

What else?

What liquid is in the H50? Will it freeze if the radiator is quenched in a Dewar filled with dry ice and acetone? 😀

 

[Yellowbeard]

Propylene Glycol and H20 but, I don't know the ratio or freezing temp. Also, we've seen some pretty cool mods so far, no pun intended. Lot's of people are improving the tubing and going to UV reactive coolant, etc etc.