Might be worse off after adding custom loop

[Arras]

Decided to spend my day off yesterday replacing the Corsair AIO cooler with a custom loop, something I hadn't done in a while.  Overall, things went ok.  Found one leak where I didn't cut the tubing (soft tubing 3/8 ID, 5/8 OD) straight, fixed that and thought I was good to put everything back together and start up the system.

Started to get CPU over temperature warnings, so I took the panels off and made sure the pump was running - it sounded like it was, also checked the CPU water block and made sure it was tight (more on this in a moment)

 

Brought up some diagnostics and noticed that at idle, my temps were pretty high - hovering between 80-90c.  Not awesome.  Played WoW for an hour or so, temps pretty much stayed in that range during the whole session, figured it was stable enough to get through to the weekend when I could take a closer look at things.

Wife is teleworking today and said after a reboot, she got another CPU over temp warning.  She shut down and let it sit for a few minutes, it's booted back up.  Clearly this is less than ideal 

 

My guess is that either the CPU block is either 1) not seated very well or 2) a cheap hunk of scrap metal or possibly the pump is just not powerful enough or running at full capacity to adequately cool off the system.

 

Here's my setup:

 

EKWB EK-CoolStream SE 240 Radiator - Mounted on top of system (fans venting up)

 

EKWB EK-CoolStream PE 120 Radiator - Mounted at the front (fans drawing air in)

3 x Noctua NF-F12 PWM 120mm Fan on the rads

BXQINLENX Professional Universal CPU Water Cooling Block

EKWB EK-RES X3 150 Water Cooling Reservoir - Mounted at the back of the case

EK-XTOP SPC-60 PWM - Plexi (incl. pump) - Mounted at the front of the case, near the hard drive cage

SilverStone PWM Fan Hub System Cables, Black (CPF04) - The pump and radiator fans are running through this.  I tried connecting the pump directly to the CPU_Fan header on the motherboard (Asus Maximus VII Gene), but it wouldn't run during the leak tests until I plugged it into fan hub.

 

I think that I'm going to replace the CPU block with something a bit better, probably an EK block.  The mounting seemed...a bit janky (there were also zero instructions included, so I took my best guess)

 

Any other thoughts?  What else might I be missing?

[ELSknutson]

Oh the joys of custom water cooling my system is due for a cleaning and im not looking forward to it  but it sound like the bock isnt seating very well. Does the fluid look like its moving when the system is on? also did you remember to take the plastic off the block before you put it on  

[Arras]

Hopefully yours goes smoother than mine!

 

 

It's hard to tell if the fluid is moving due to the color, but it does appear to be.  It was certainly moving along well enough as I was filling the system.

 

I'm putting my money on the block, with a side bet on the pump maybe not running at full bore

/edit - yes, I did remember to peel the plastic covering off the base of the cpu block - it was the least satisfying peel ever!

[zbryant91]

Have you tried unplugging the pwm wire from your pump to see if it will run at full speed (pending which pump you have) I know on my ek pump if I just plug in the molex when doing a leak check it runs at full speed and there is a noticeable hum. So That would be a quick and easy test to see if it is the mobo header not sending the correct power. 

[Arras]

5 minutes ago, zbryant91 said:

Have you tried unplugging the pwm wire from your pump to see if it will run at full speed (pending which pump you have) I know on my ek pump if I just plug in the molex when doing a leak check it runs at full speed and there is a noticeable hum. So That would be a quick and easy test to see if it is the mobo header not sending the correct power. 

The pump only has a PWM connection, no molex or anything else.  

 

I was looking at my motherboard's manual and it looks like PWM has to be enabled, so I'll do that tonight and try plugging the pump into the CPU fan header instead of using the hub

[zbryant91]

Yea I’ve had some weird results trying to run them through hubs. What brand pump is it? 

[Arras]

3 minutes ago, zbryant91 said:

Yea I’ve had some weird results trying to run them through hubs. What brand pump is it? 

It's an EK pump, the EK-XTOP SPC-60

[Mick Naughty]

Did you try turning up the pump speed while attached to the mobo? 

[Neo-revo]

9 hours ago, Arras said:

What else might I be missing?

some block's don't clearly label in/out's   and if there is 'jet plate' tech in your block you cannot run it backwards. unless you want sub-par performance

 

'**edit

you can take it apart and investigate to see if you have fluid moving the wrong way through the block too

[ewitte]

On 12/6/2018 at 8:18 AM, Arras said:

It's hard to tell if the fluid is moving due to the color, but it does appear to be.  It was certainly moving along well enough as I was filling the system.

 

I can't tell either but if I put my finger on the outlet I can feel the flow.  At least when its at the higher speeds if I drop to 3xxx rpm I can't really feel it either.

[Neo-revo]

28 minutes ago, ewitte said:

I can't tell either but if I put my finger on the outlet I can feel the flow.  At least when its at the higher speeds if I drop to 3xxx rpm I can't really feel it either.

I'm not really sure why pumps need PWM.   Fluid temp shouldn't vary enough for PWM to matter. Just set a speed and leave it

[WihGlah]

Umm - what order do you have your loop in?

 

It looks like pump > Front rad > reservoir > cpu block >  top rad > cpu block.

 

If it is - that is your problem. Move the reservoir behind the pump and you will be golden. Assuming nothing else is wrong. 

[Mick Naughty]

System looks pretty much bleed unless there’s some air stuck in the rads. At that point loop order won’t matter. Seems like another issue. 

[Arras]

On 12/7/2018 at 11:46 AM, WihGlah said:

Umm - what order do you have your loop in?

 

It looks like pump > Front rad > reservoir > cpu block >  top rad > cpu block.

 

If it is - that is your problem. Move the reservoir behind the pump and you will be golden. Assuming nothing else is wrong. 

The outlet from the pump is going to the CPU block, from there it goes to the 240 Rad, to the res, to the 120 rad, back to the pump.

 

A friend pointed out that with the pump below the CPU, it might be struggling to push the liquid up to the CPU block.  I think I'm going to move the pump higher up, let gravity do some of the work instead.  Probably attach the pump to the res in the back.  Might also skip the 120 rad, it's likely overkill 

[Neo-revo]

4 hours ago, Arras said:

Might also skip the 120 rad, it's likely overkill 

this pump placement wont matter unless the head pressure is low. and over kill for heat displacement is what you want for best performance

[Arras]

3 hours ago, Neo-revo said:

this pump placement wont matter unless the head pressure is low. and over kill for heat displacement is what you want for best performance

How do I tell if head pressure is low?

 

These are the specs from the EKWB site

Pump electrical and mechanical properties:
- Motor: Electronically commuted motor 
- Rated voltage: 12V DC
- Power consumption: 6W 
- Maximum pressure head: 3.2m 
- Maximum flow rate: 250 L/h 
- Maximum liquid temperature: 50°C 
- Materials: PPS-GF30, EPDM O-rings, Aluminium oxide, graphite 
- Power connector: 4-Pin PWM FAN connector

Technical Specifications:
- Dimensions (W x D x H): 62 x 62 x 51 mm
- Mounting hole pattern: 50x50mm

Operational regime:
- PWM duty cycle: ~ 20-100%
- Default behavior: Runs at 100% duty cycle when no PWM feedback signal is present

Enclosed:
- EK-XTOP SPC60 - Acetal
- EK-SPC60 PWM 12V DC pump motor
- Mounting mechanism
- Installation manual

[Neo-revo]

10 hours ago, Arras said:

3.2m 

Good head pressure..  But thats rated at full 12V.   dont know about lower'ed rpm (voltage) stats.

 

if its not loud try running it at 100% since it only gets - Maximum flow rate: 250 L/h 

[ewitte]

On 12/7/2018 at 10:17 AM, Neo-revo said:

I'm not really sure why pumps need PWM.   Fluid temp shouldn't vary enough for PWM to matter. Just set a speed and leave it

It's pwm but I adjusted it manually to my liking.

[Neo-revo]

4 hours ago, ewitte said:

It's pwm but I adjusted it manually to my liking.

what i mean is unless you can set your curve to a say 5C spread there isn't much point. Also the amount of energy required to raised the closed volume of fluid would be high (ie pump speed wont really change that much)  unless there some sort of software suite for the newer boards with w_pump headers.

 

you have a pwm pump that you set to run at a speed, i have a pump that i can set at 5 different speeds to run..   i dont see the difference unless you are using a 3/2 pin pump and cannot regulate its V input easily.

also ambient would play a factor in you spread (one in winter wouldn't work for spring/summer unless you use climate control in your room already)

[ewitte]

1 hour ago, Neo-revo said:

what i mean is unless you can set your curve to a say 5C spread there isn't much point. Also the amount of energy required to raised the closed volume of fluid would be high (ie pump speed wont really change that much)  unless there some sort of software suite for the newer boards with w_pump headers.

 

you have a pwm pump that you set to run at a speed, i have a pump that i can set at 5 different speeds to run..   i dont see the difference unless you are using a 3/2 pin pump and cannot regulate its V input easily.

also ambient would play a factor in you spread (one in winter wouldn't work for spring/summer unless you use climate control in your room already)

By manually I mean the exact same % setting for every temperature rating the RPM is constant it doesn't vary at all.

[Neo-revo]

1 hour ago, ewitte said:

By manually I mean the exact same % setting for every temperature rating the RPM is constant it doesn't vary at all.

that was my point, i set it on 3 and it runs at 3, you set it at 65% and it runs at 65%

[For Science!]

On 12/9/2018 at 4:27 AM, Arras said:

How do I tell if head pressure is low?

 

These are the specs from the EKWB site

Pump electrical and mechanical properties:
- Motor: Electronically commuted motor 
- Rated voltage: 12V DC
- Power consumption: 6W 
- Maximum pressure head: 3.2m 
- Maximum flow rate: 250 L/h 
- Maximum liquid temperature: 50°C 
- Materials: PPS-GF30, EPDM O-rings, Aluminium oxide, graphite 
- Power connector: 4-Pin PWM FAN connector

Technical Specifications:
- Dimensions (W x D x H): 62 x 62 x 51 mm
- Mounting hole pattern: 50x50mm

Operational regime:
- PWM duty cycle: ~ 20-100%
- Default behavior: Runs at 100% duty cycle when no PWM feedback signal is present

Enclosed:
- EK-XTOP SPC60 - Acetal
- EK-SPC60 PWM 12V DC pump motor
- Mounting mechanism
- Installation manual

Can you unmount the cpu block, and then before cleaning the thermal paste off, take a picture of both sides. Then take a photo of your thermal paste application before remount the block? Something this serious sounding has to be in the cpu block since your pump is capable enough to fill the loop.

[Arras]

On 12/10/2018 at 5:10 AM, For Science! said:

Can you unmount the cpu block, and then before cleaning the thermal paste off, take a picture of both sides. Then take a photo of your thermal paste application before remount the block? Something this serious sounding has to be in the cpu block since your pump is capable enough to fill the loop.

Can do, I'm replacing the water block with an EK Velocity block.  I think the block itself is probably ok, but the mounting was pretty sketchy. I think it's whatever retention screws/springs aren't keeping the block firmly enough against the CPU.

 

When I was mounting it, I did the grain of rice method in the center of the chip.  There was no pre-applied paste on the block

[Neo-revo]

4 hours ago, Arras said:

There was no pre-applied paste on the block

this is a good thing, depending on 'rep' 

 

company's with no rep, you paid less so they didn't have to think about it.

 

well made parts by rep'ed company's want you to see the mirrored finish they put on it,   even though  there is no real performance difference between that and a brushed finish after you smear the stuff all over it (the whole point of smearing the stuff)

[Arras]

I'm pretty confident I'm on the path to success!  Thank you all for your encouragement and helpful advice, I'm definitely out of practice

 

Fairly certain the issue was that the original waterblock was making poor contact with the CPU - when I lifted it off to replace it, the thermal paste had barely spread at all.  How it was transferring any heat at all is anyone's guess.

I replaced the cheapo block with a better quality one from EK and the difference was immediate.  First being that there were actual installation instructions (looking at the two blocks, I'm betting the issue was the placement of the springs, I think if I originally had them in a different spot, the temps wouldn't have been so bad - but my trust and faith in that block was soured, so in the trash it goes)

 

I mounted the EK block and let it sit for a few minutes, then removed it and the thermal paste spread was night and day different.  The attached pictures should give you an idea of what the original block and paste looked like.  I forgot to get an after picture, that was clearly the issue.

 

Thanks again everyone!