Sign in with
Sign up | Sign in
Your question

Something's wrong with my CPU block (Storm) *56k beware at bottom*

Last response: in Overclocking
Share
December 6, 2007 10:44:44 PM

Hi all.

Ok, there's definitely a problem with my water loop. Flow is res>pump>rad>cpu>gpu>gpu>res. The water flowing into my CPU block, is filling the lines, and has quite a bit of pressure, so I'm 99% sure my radiator isn't slowing water flow down.

However, the water coming directly out of my CPU block, is at a dribble. (going to try to post pictures soon)

So I'm pretty much 99% sure that the problem with my loop is that my CPU block is slowing the flow down too much.

The thing I'm confused with is, how do i fix it?

I'm using Fluid XP+ Ultra (which may be a *tad* bit thicker than water) and my CPU block is Swiftech storm R2. Pump is Swiftech MCP655.

Also, i did open the block and clean it best i could. When i assembled it back together, all jests where open.

Is it possible that maybe i screwed the barbs into the block too tight, and that maybe the barb is too close to internal parts, restricting flow?

I really don't think that my pump is too weak to do this loop.

Should i buy a different waterblock? And if so, what kind?

Thanks for your help, and i will try to post pictures soon.

--------------------------------------------------------
EDIT: To make a long thread short, my storm is impeding my water flow by 3+ GPM. Turning my MCP655 into about 1GPM. If anyone has any clues, let me know!
--------------------------------------------------------
December 7, 2007 12:51:16 AM

It's your cooling loop setup that's the problem.

For purposes of definition, “Head” refers to the height of a vertical column of water. This is the maximum height that a pump can sustain any semblance of flow rate before it loses its capabilities. For purposes of an example we'll use a pump rated at 317gph with an imaginery "head" of 36 inches. At 0 inches of height you will have maximum flow rate and the pressure will be zero. Pressure is a measure of resistance to flow. Thus, at its initial discharge, at 0 height, the pump experiences its least resistance and generates its fullest flow. As the height in the cooling loop increases, the resistance to flow increases and the flow rate decreases. Earlier we said that our pump had a "head" of 36 inches. The closer the pump gets to its "36 inch" height, the less flow is generated.

So, at 0 height we have 0 pressure and 317gph. At 36 inches we have full pressure and no flow.

The one component that is responisble for the greatest restriction to both flowrate and head is always going to be the rad. This is because of the fittings and the twisting nature of the tubing internally. The CPU should always be the FIRST component that the coolant hits right out of the pump so that it can take advantage of the best flowrate and head.

You should put the rad at the end of the loop BEFORE the coolant goes back to the reservoir:

Reservoir - pump - CPU - GPU - GPU - rad - back to reservoir

Set your loop up like that and you'll see better results since the CPU and GPUs will have the benefit of the best flowrate and head right out of the pump with no hinderance.

However, I do need to point out that having one rad in a cooling loop that involves more than one major component is forcing the latter components to deal with not just their own heat but the additional heat fromt he previous components. That last GPU waterblock is having to deal with the heat from the CPU, a GPU and it's own heat.
December 7, 2007 1:07:48 AM

Head really refers to open loops. In a closed loop it's more like the torque.

In the closed loop the pump doesn't have to push water up as the weight of the water on the "downward" slope equalizes it.

Also I'm probably wrong on this one, but I'm pretty sure in a closed loop flows will also be roughly the same everywhere in the loop while pressures will decrease (lowest at the inlet of the pump, highest at the outlet of the pump).

I agree with most of what you said to a point. It's a trade off. Some pumps generate heat and then the CPU will have to deal with that heat but get water at a higher pressure. Or it can be after the rad and get less pressure but cooler water (even though all the water will be within 3-5 C anywhere in the loop really).

Also in your example... do you really think if you ran tubing 4 ft off the top of a pump with 36 inches of head that at you would have no pressure at the outlet of the pump (0 height) and full pressure at the top? Pressure will decrease as it leaves gets further from the pump. Edit: Or that once the water reaches the max head it will continue to flow 317 GPM at the pump?

Also.. do you think the pump actually pumps 317 GMP out of it in a closed loop yet somehow, somewhere, you lose flow even though it's a closed system?

"Head" isn't a pressure rating or flow rating. It's kinda like Torque. How much work can it do. In a closed loop the resistance is the only factor which will effect flow. Head is the "torque" that pushes against this. Pressure/Flow Curves are a bit more important for closed loop applications.
December 7, 2007 2:23:22 AM

Ok, but do you honestly think that because i have my radiator inbetween my pump and my CPU waterblock, that my flow magically changes from awesome flowrate, to nearly a dribble?

Because here's the thing, even with my radiator hooked up inbetween them, i should be getting decent flowrates with a 655. My CPU is idleing at 50 degrees C.

If people can get good results with a low GPH/head pump with the storm, i should be getting decent flow with mine.

This is the situation i'm trying to fix because I'm about to go and buy a apogee GTX with overnight.
December 7, 2007 2:37:17 AM

icewolf69 said:
Ok, but do you honestly think that because i have my radiator inbetween my pump and my CPU waterblock, that my flow magically changes from awesome flowrate, to nearly a dribble?

Because here's the thing, even with my radiator hooked up inbetween them, i should be getting decent flowrates with a 655. My CPU is idleing at 50 degrees C.

If people can get good results with a low GPH/head pump with the storm, i should be getting decent flow with mine.

This is the situation i'm trying to fix because I'm about to go and buy a apogee GTX with overnight.


Flow doesn't change in a closed loop system.

You should be getting at LEAST 1.0-1.5 GPM around the entire system (not great, but I said atleast).

Take the Pump/CPU Block out of the case. Run tubing from a bucket to the pump then the CPU block then back to the bucket. Compare flow with and without the CPU Block. If you have a lot of flow without it then it's definately crazy restrictive (this is a ghetto pressure drop test).

Try rebuilding it again, maybe something is blocking the nozzles or anything in there.. Redo the o-rings, try anything while it is outside the case.
December 7, 2007 3:45:58 AM

The pump was imaginery but the example was correct. Once the loop achieves the state of max head it will have 0 flow as it has reached the highest the pump can push the coolant.

I didn't say tha head was a pressure of flow rating. I just pointed out its relationship to the flowrate.

icewolf, what kind of gpu waterblocks are you using? Fullbody or just something like the the Swiftech MCW60? How about your rad?

The Apogee is more restrictive to flow than the Storm btw

December 7, 2007 4:08:56 AM

using BFG GPU waterblocks, they came pre-installed on the 8800's. radiator is a thermochill pa120.3

I just don't understand how someone can have the exact same rig as mine, even down to the fluid, and have everything running perfectly....

Is it possible that my pump is broken? can it be broken and still push water, just not very much; or is it one of those, if it's broken it's not working at all deals.

Just now, i drained the loop (again) and cleaned and double checked the storm block, i didn't see anything unusual
December 7, 2007 4:19:57 AM

Well, there are two versions of the MCP655 - one with a dial for the flowrate and one that just has one flowrate. You can tell the difference by checking on the back of the pump if you have a tiny red dial that you can manipulate
December 7, 2007 9:52:31 PM

Yeah i have a flow rate dial. It's been turned to 5 for all these tests


I'm about to drain the loop again, as me cleaning the CPU block a second time didn't help flow rate at all. I'll hook just the pump and CPU block up and check flow rates then. Assuming that the CPU water block is too restrictive (even without the radiator and GPU's) should i just try a different block? Someone said the Apogee GTX was just as restrictive if not more than the storm, but i find that hard to believe.

Also, since i have the water cooled edition GTX's, is there any way/where to buy just the fan and heat sinks for them? Assuming i wanted to convert the water cooled into air cooled so that i can just take them out of the loop?

I mean, i have a Silverstone TJ-09, and i see absolutely no possible way of fitting in an extra pump and radiator and reservoir just for my video cards.

Also, i got a picture of my tubes coming out of my waterblock. I know it's blurry, camera zoomed in on the wrong part, but, you can still see what i'm talking about:

December 7, 2007 11:00:22 PM

The Apogee is more restrictive than the Storm. There are a few heatsink/fan combos that will attach to your 8800GTX's that will expel the hot air out the back.

If you wanted to add a second pump you could get the MCP350 or 355 and attach it to a drivebay reservoir so that it would only take up one 5.25 drivebay. The MCP355 is every bot as powerful as the MCP655 - just not as silent.

The tubing isn't full of coolant...mmmmmmmmm
Well, that's not right.....

Two things come to mind here:

one, you have a leak somewhere. However, I seem to doubt this as you would see a stream of little tiny bubbles in the tubing and a big puddle at the bottom of your case - which I am assuming you don't.

The other thing you might consider (and it does make sense as far as hindering your cooling loop) you have air trapped somewhere, perhaps in the rad. Every few months I take my rig apart and give it a little cleaning - vacuum the fan filters, perhaps change out a few tubes, etc. I have a secondary PSU that I have set to run when I flip the switch on it (in other words, it doesn't have to be connected tot he MB. I do this because I like to set up my cooling loop before I put ALL my hardware back in and I like to runt he pump as I add coolant. I realize that air gets trapped in my rads so I sometimes have to tip[ my case while the cooling loop is running so that the trapped air flows to the reservoir.

I think trapped air is your strongest culprit as to why your cooling loop is behaving like it is and it may just be in your rad
December 7, 2007 11:27:58 PM

Ok, well i just finished running the pump and the CPU block, both at the same time and independant.

Now, the pump by itself is running like a champ, i'm pretty impressed with it.

When i hook the storm block up to it, it's definitely not coming to a dribble, BUT, it is going quite slow. Now i know the flow rate would be enough to cool the cpu efficiently, but i'm assuming, that with the flow cut of the CPU Block, and then adding in the radiator plus the 2 GPU blocks, it's just limiting the flow that much more. (i think so at least).

This is to be expected, but i wish there was some way of measuring how much exactly it's putting out. I'm going to try and find a gal jug and time how long it takes for it to fill up.
December 7, 2007 11:46:10 PM

That's the best way to "ghetto" measure it! Let it fill the gallon and time it. Not hard math to figure out the GPM after you know that.

1.5-2 GPM = Pretty Good.
2+ GPM is excellent.

The D5 on setting 5 + a Swiftech Storm and 5 ft of 1/2" ID tubing should get about 2 GPM of flow through it.

With an Apogee GTX you should get 2.2-2.3 GPM. A GT would get about the same.

Believe it or not 1+ GPM and you should be ok. Less than 1 isn't good.
December 8, 2007 12:09:44 AM

Ok, it's filling pretty much exactly 1GPM with the storm block connected.

What are your thoughts on this?

EDIT: i'm also going to try to remove the storm and check GPM with nothing but the pump. theoretically, the 655 should be putting out 5.28 GPM right? (317 GPH / 60)

Is it possible for a pump to be broken, and only work half-assed; but work none-the-less?
December 8, 2007 12:12:31 AM

icewolf69 said:
Ok, it's filling pretty much exactly 1GPM with the storm block connected.

What are your thoughts on this?


What size tubing?
December 8, 2007 12:14:16 AM

1/2" ID, 5/8 OD
December 8, 2007 12:16:29 AM

Was the water flowing "basically" horizontally with no big ups and downs in the loop? (It is an open loop, so now head pressure does matter a bit unless the up is equalize with down).
December 8, 2007 12:19:03 AM

Cnu: Yes, basically it's fully horizontal.
December 8, 2007 12:21:13 AM

Run the pump with about 5-10 ft of tubing, let me know how much tubing you run off the top of the pump and on the intake of the pump and how many GPM you get.
December 8, 2007 12:31:29 AM

ok the 1GPM value was given with the following:

17" bucket to pump
32" pump to block
40" block to bucket

equals 89".....~7.42 Feet
December 8, 2007 12:38:05 AM

icewolf69 said:
ok the 1GPM value was given with the following:

17" bucket to pump
32" pump to block
40" block to bucket

equals 89".....~7.42 Feet


With 8 ft of tubing the MCP655 at setting 5 should pump 3.75ish GPM.

With the Swiftech Storm in the same 8ft loop should be about 2 GPM.

I ask without the block so you can check the pump.
December 8, 2007 12:40:49 AM

ok, i'll disconnect the block and check the pump
December 8, 2007 12:51:28 AM

ok, this one is a bit harder because the water's moving so fast.

It filled up the gallon in about 15 sec. Minus 2-3 seconds for the time it took me to pull the hose out of the bucket, slam it into the jug, then hit the start button on my watch.

So about a G per 12-13 sec.
December 8, 2007 12:55:35 AM

icewolf69 said:
ok, this one is a bit harder because the water's moving so fast.

It filled up the gallon in about 15 sec. Minus 2-3 seconds for the time it took me to pull the hose out of the bucket, slam it into the jug, then hit the start button on my watch.

So about a G per 12-13 sec.


Pump is definately fine. For the size of your loop I'd say you should (as you said before) move to a less restrictive CPU block (unless you still think you might look at it and it's not 100% open inside).

Everything else checks out.

As was said, when you put the loop back together make sure you definately get all the air out, that will help as well!

Also that's a theory and inflated number on your GPM for a regular pump. With tubing 3.75-4 is normal. Only other thing is put another pump into the same loop to help flow.
December 8, 2007 12:58:45 AM

well what other block would be less restrictive? The Dtek Fuzion?

and do you think some thing's wrong with my storm block? i mean, even if my pump's pumping at 4GPM instead of 5, a 3GPM cut going through the storm seems a bit excessive don't you think? especially when people have similar setups as mine and are doing ok.

December 8, 2007 1:02:47 AM

Something just doesn't seem right with the block. Shouldn't be that restrictive.

Apogee GT/GTX or D-Tek Fuzion (With no Nozzle).
December 8, 2007 1:05:29 AM

man, i just took the storm apart, and everything looks clean. i mean, i don't really know what could possibly be the problem with it. I guess i'll try and call swiftech in the morning and see what they say.

I'm going to post pictures of my storm to see if you or anyone else notices any problems.

And HUGE thanks to you cnu for your help and patience ^^
December 8, 2007 1:16:42 AM

OK, here's the pics, so if any of you storm experts see anything, let me know!

Copper bottom piece:



Jet Streams:



Jet Streams (reverse side)



Top plastic piece:




This case is not closed yet!!! Still need someone to solve
December 9, 2007 12:20:42 PM

Hi,

The D-Tek Fuzion block blows away the Swiftech blocks on flow restriction... Why not just upgrade to an Fuzion?? They don't appear to be too expensive!! I am seriously considering upgrading my dual Apogee (in series on dual Opteron) as they do really restrict flow (dual DD C5 mag pumps couldn't fill the pipe between them @10V - 1/2" ID tubing).

and da proof :pt1cable: 
http://www.xtremesystems.org/forums/showthread.php?t=131297


Bob

icewolf69 > P.S. On the pictures you posted... THG Forums aren't to bright on scaling!!
I would change them to a smaller specified height/width with 4x the DPI!!
December 9, 2007 5:15:55 PM

Hey bob, yeah, my Fuzion is already ordered with overnight

I contacted Swiftech, they weren't too helpful since it was the weekend, but they think they will RMA it. I'll get an RMA number on mon and send it back. I guess i'll end up keeping it as a spare or selling it on ebay or something.

But i do have high hopes for my new fuzion, i think i'm going to bow it, and depending on flow rates, put in a nozzle.

On another note for those wondering, i tested flow rate of my radiator (Thermochill PA120.3) and my SLi 8800's (BFG version with waterblocks) and the flow rates is over 2GPM.

Basically i'll just match the flowrate on my fuzion with everything else. If the nozzles restrict them too much i just won't put them in.

on the pictures: do i resize them here in the forum? or wherever i uploaded the pictures?
December 9, 2007 5:22:55 PM

Fuzion with bow/nozzles can get pretty restrictive.

Here are some pressure drop tests on different nozzles.
December 9, 2007 6:04:28 PM

Yeah I've seen that article cnumartyr, but i couldn't remember how to get back to it. I was hoping to put in the 6.3mm nozzle, but maybe now i won't be able to.
December 9, 2007 7:58:53 PM

icewolf69

A jpg (etc.) image has a certain width and height in pixels but also DPI information. I had to Photoshop a couple of images for a thread I started on THG as it literally takes the size of an image (DPI * width in pixels , DPI * height in pixels). You need to make the actual dimensions smaller (higher DPI, or same DPI and less pixels). Hope that makes sense!!

Good choice with the block - the Fuzion looks good!! :hello: 

Bob
!