New Custom Water Cooling Loop up to 90°C! Help!

baN893

New member
Hi everyone

I recently decided to go down the custom water cooling route to challenge myself a bit. Long story short. I upgraded from a be quiet! Dark Rock Pro 3 to the following Alphacool NexXxoS Cool Answer 240 D5/UT - kit. I learned about all the basics that I had to know, watched lots of instructional videos and read tons of sites on the subject. I work in IT and have been building my own PC systems for the past 5 years, so I am no novice to PC hardware.

Here's a list of the Components in the water cooling kit:
1x Alphacool NexXxoS XP3 Light - Intel/AMD
1x Alphacool Repack - Single Laing D5 - Dual 5,25 Bay Station
1x Alphacool VPP655 - Single Edition
1x Alphacool NexXxoS UT60 Full Copper
3x Alphacool tubing AlphaTube HF 13/10 (3/8“ID) - clear
6x Alphacool HF 13/10 Anschraubtülle G1/4 - Deep Black
1x Alphacool NB-eLoop 1200rpm - Bionic Lüfter (120x120x25mm)
1x Alphacool CKC Cape Kelvin Catcher Clear 1000ml
1x ATX-bridging plug (24 Pin) - black

I have set everything up, bled the system for 48 hours making sure there are no leaks etc. However, now that I have my computer running, I ran AIDA64 and quickly discovered that my temps are through the roof. like, between 87°C and 98°C depending on the core.

I am running an Asus ROG Z170 Maximus VIII Hero Motherboard with an Intel Core i7 6700K Processor in a Corsair Obsidian 750D. I tested out different clocks from 4.0GHz to 4.6GHz which has been running stable for the past 4-5 months. The i7 does NOT throttle or crash, the temps are simply really high.

I just can't figure out why the temps would be so high. I reseated the CPU cooler, applied a different and new type of thermal paste (because my other tube was 5 years old, who knows), set the fans to max settings, air is coming through the radiator.

Is my pump broken? Since this was the first time doing a water cooling loop, I turned the PSU on and off to allow the water to go through all of the components in the loop. I got a little ambitious and it took me about 4 seconds to realize what a dry pump sounds like *eeeeek*. Should I invest in a flow meter? but when I turn up the pump speed to max, air disturbance gets pumped through the loop which I can visually see moving around at a seemingly fast speed. Also the pump has no rattling noise or anything that I wouldn't think is normal operating sounds. Is the pump ruined from 4 seconds of dry running?

Is my motherboard broken? Is it possible that my motherboard is broken? ESD? I wouldn't think so since my system is stable, simply running hot.

Is my CPU broken? Could I have fried my CPU? When I still had my be quiet Dark Rock Pro 3, I used Asus' Dual Intelligent Processor 5 to OC my CPU (I know I know...) but it worked and settled at 4.7-4.8, but I set it to 4.6 just to be 100% stable at all times. Now, with my custom loop, when I run Dual Intelligent Processor 5 to OC my CPU, it cancels on the first run through and doesn't actually test to see if the CPU is able to go to any higher clocks than the mildest overclock which is 7%. The nature of the program is that it simply reboots, and goes to the next step in the OC process which is determining the fan settings, so I don't get an error message or failed etc.

Are my Alphacool NB-eLoop 1200rpm fans ? I feel air coming through the rads, but I don't know if it's enough. Even at full speed, the temps go up to 90°C.

So guys, hope you have your Sherlock hats and pipes ready, as I appreciate every single one of you who reads through this wall of text and tries to help me solve this.

Cheers,

baN

P.S. Added an image of my system that I took of it while bleeding. Is my setup completely wrong?

TZy9Esp.jpg
 
Last edited by a moderator:
Those temps sound alarmingly bad, as if something is broken or incorrectly assembled and it's just not functioning correctly rather than a little simple efficiency problem.

Does the pump still work, how is it wired? can you feel it running? running it dry momentarily shouldn't cause it to fail, they are water lubricated and cooled so it's really not advisable however if it was just seconds then it survive and if you've observed coolant flow since then it sounds like it did.

Was coolant/air flowing through the loop or just around the reservoir. Are the in/out ports on the reservoir correct, the CPU block is less critical. How well bled is it, did you get all air out or just let it sit there running. I'd advise turning it on it's back and running it for a bit and just generally moving it around to make sure the CPU block has no air inside.

Are you sure the CPU seat is ok, is the CPU block touching the capacitors around the socket? Is the spread good when you take it off? What is the CPU voltage like? What are the stock load/idle temps?


Also what software are you using to monitor temparatures, are you sure it can be trusted?

JR
 
what volts are you using on your CPU? are you sure your pump is plugged in and running at the proper speed or that there is no kink in the loop?
 
Those temps sound alarmingly bad, as if something is broken or incorrectly assembled and it's just not functioning correctly rather than a little simple efficiency problem.

Does the pump still work, how is it wired? can you feel it running? running it dry momentarily shouldn't cause it to fail, they are water lubricated and cooled so it's really not advisable however if it was just seconds then it survive and if you've observed coolant flow since then it sounds like it did.

Was coolant/air flowing through the loop or just around the reservoir. Are the in/out ports on the reservoir correct, the CPU block is less critical. How well bled is it, did you get all air out or just let it sit there running. I'd advise turning it on it's back and running it for a bit and just generally moving it around to make sure the CPU block has no air inside.

Are you sure the CPU seat is ok, is the CPU block touching the capacitors around the socket? Is the spread good when you take it off? What is the CPU voltage like? What are the stock load/idle temps?


Also what software are you using to monitor temparatures, are you sure it can be trusted?

JR

As far as I can tell, the pump works in that it pushes water through the loop. It is connected to a Molex power plug to the PSU and to the pump header on the motherboard.

I am certain I bled the system well. I moved the computer around, changed the flow rate on the pump etc. All of those things regularly resulted in pockets of air being released and flowing through the loop into the reservoir. I did this several times over the course of 48 hours. The inlet and outlet are also correct. Flow seems to be throughout the loop.

The CPU socket was tightenend in an X pattern hand tight both times. The spread of the thermal compound is also good when removing the block.

I would have to double check to make sure the socket isn't touching any capacitors, but I would be surprised if it did.

Both HWMonitor and AIDA64 showed the temp to be ~86°C at stock(!) clock. So 4.0GHz at default voltages. I reset to defaults in BIOS. I am at work atm, so if you wanted exact voltages you'd have to wait, but from memory, I think 4.6GHz was achieved with 1.38v.

Good lord dude look at that dust why didn't you clean your case out before you started adding new parts to your rig.

Thanks for your concern :o I will take the air compressor to the system as soon as possible.

what volts are you using on your CPU? are you sure your pump is plugged in and running at the proper speed or that there is no kink in the loop?

Both HWMonitor and AIDA64 showed the temp to be ~86°C at stock(!) clock. So 4.0GHz at default voltages. I reset to defaults in BIOS. I am at work atm, so if you wanted exact voltages you'd have to wait. There are no kinks in the loop and the pump seems to be able to change speeds fine. Temps are high regardless of the pump speed. I can set it from 1 - 5 and the difference between a 1 and a 3 are ~5°C.

Did you make sure to peel the protective film off the contact patch of the cpu block. Stupid to say but it happens.

I did make sure that the film was removed! The red writing on the film was hard to miss.

Thanks for all your replies so far everyone. I wonder if I shouldn't just mount my be quiet! Dark Rock Pro 3 back on and see if my temps normalize. That would at least allow me to narrow down if it is my loop, or my motherboard or CPU.
 
I'm gonna agree with JR here it sounds like you have an air lock in your system somewhere try moving the rig all different ways and see if there are any air bubbles in the loop and the pipe from your rad to the res could have been better with 90 degree angles fella are you running a D5 PWM or Vario?
 
Last edited:
Sooo. I took of the CPU block and decided to have a look at the CPU...

What now?:eek:

rbqISDO.jpg


92wCuEv.jpg
 
Last edited by a moderator:
Ok since no one else wants to be the bearer of bad tidings I guess I will.

That processor is Done. Who,What,Why or How doesn’t really matter now. With the PCB being as warped as it is Nothing will save it. With the PCB being as warped as it is the heat spreader will Never have a good “thin” thermal contact again. And before you ask; No a warped PCB cannot be straightened out. And No Intel will not warranty a processor that shows extreme thermal damage.

Go buy a processor and try again.

This of course begs the question “How in the world did you get enough heat into the PCB to warp it”. Point blank; I don’t know.

I will say that over the past 30 odd years I’ve been into computers the ONLY time I have seen PCB’s warp is under extreme heat loads over Extended time frames (IE weeks, months, years). NOT short time frames (IE minuets). PCB’s can take a massive thermal spike over the short term and they won’t warp like that. I’ve actually caught a PCB on fire (from an overclock) and the PCB didn’t warp like that. However; I Have seen PCB’s warp like that when fools try to put a big overclock on a stock Intel air cooler. It doesn’t pop immediately, but at some point down the line the heat spreader looses contact with the TIM, and then it pops.

To be clear; I’m not accusing you of anything. I don’t know how you burned up this processor. I am simply commenting on my experience over the last 30 odd years VS the picture’s you posted. That being said, you still did.

I suspect you didn’t install the Dark Rock Pro 3 correctly originally. Then you seen your temps getting out of hand, and decided to use water cooling to get a handle on the temps. But by that time the damage was already done and nothing was going to save that processor.

And by the way “Both HWMonitor and AIDA64 showed the temp to be___” will never happen. Aida64 reads the “approved by Intel” thermal couple under the die package and the “reported” core temps. HWMonitor reads the thermal couples inside each core used for thermal throttling (not reporting), and averages the total to come up with a package temp. They will Never read the same…………
 
Ok since no one else wants to be the bearer of bad tidings I guess I will.

That processor is Done. Who,What,Why or How doesn’t really matter now. With the PCB being as warped as it is Nothing will save it. With the PCB being as warped as it is the heat spreader will Never have a good “thin” thermal contact again. And before you ask; No a warped PCB cannot be straightened out. And No Intel will not warranty a processor that shows extreme thermal damage.

Go buy a processor and try again.

This of course begs the question “How in the world did you get enough heat into the PCB to warp it”. Point blank; I don’t know.

I will say that over the past 30 odd years I’ve been into computers the ONLY time I have seen PCB’s warp is under extreme heat loads over Extended time frames (IE weeks, months, years). NOT short time frames (IE minuets). PCB’s can take a massive thermal spike over the short term and they won’t warp like that. I’ve actually caught a PCB on fire (from an overclock) and the PCB didn’t warp like that. However; I Have seen PCB’s warp like that when fools try to put a big overclock on a stock Intel air cooler. It doesn’t pop immediately, but at some point down the line the heat spreader looses contact with the TIM, and then it pops.

To be clear; I’m not accusing you of anything. I don’t know how you burned up this processor. I am simply commenting on my experience over the last 30 odd years VS the picture’s you posted. That being said, you still did.

I suspect you didn’t install the Dark Rock Pro 3 correctly originally. Then you seen your temps getting out of hand, and decided to use water cooling to get a handle on the temps. But by that time the damage was already done and nothing was going to save that processor.

It's a known fact that Skylake CPUs are prone to bending, yes you are right heat is a major factor and 30 years ago during the i386/i486 days it was ultra hard to bend a die but with todays CPUs being 14nm it's not surprising that this is happening now. Clamping pressure and heat from overclocks on a thin die was never going to end well.
 
It's a known fact that Skylake CPUs are prone to bending, yes you are right heat is a major factor and 30 years ago during the i386/i486 days it was ultra hard to bend a die but with todays CPUs being 14nm it's not surprising that this is happening now. Clamping pressure and heat from overclocks on a thin die was never going to end well.

Wow, learn something new every day. I can honestly say I haven't heard a thing (or seen) an instance of this myself until now. From some of the pics I pulled up it seems like the PCB on the skylake's are substantially thinner.................EWWWWW

When I wrote that I was thinking about some of the sandy and ivy bridge's I've seen warp like that (purely heat related from OC on a stock cooler)

I don't think Skylake being 14nm really has anything to do with it. The die doesn't carry any of the clamping force. The pcb does. And its a .8 mm PCB VS a 1.1mm PCB for the DC series. But I would guess (assuming logic holds) the PCB is .3mm thinner because the FIVR is back on the MB. But who knows it could just as well be Intel wanted to save .0000000000000001 cents per processor............LOL

In any case. What ever the reason, be it clamping force + heat Or Heat over time or gremlins I would lay dollars to bones that the die has at least partially separated from the TIM and IHS. And with the warpage (unless there is some way I'm unaware of) there's no fixing that. Unfortunatly
 
Wow, learn something new every day. I can honestly say I haven't heard a thing (or seen) an instance of this myself until now. From some of the pics I pulled up it seems like the PCB on the skylake's are substantially thinner.................EWWWWW

When I wrote that I was thinking about some of the sandy and ivy bridge's I've seen warp like that (purely heat related from OC on a stock cooler)

I don't think Skylake being 14nm really has anything to do with it. The die doesn't carry any of the clamping force. The pcb does. And its a .8 mm PCB VS a 1.1mm PCB for the DC series. But I would guess (assuming logic holds) the PCB is .3mm thinner because the FIVR is back on the MB. But who knows it could just as well be Intel wanted to save .0000000000000001 cents per processor............LOL

In any case. What ever the reason, be it clamping force + heat Or Heat over time or gremlins I would lay dollars to bones that the die has at least partially separated from the TIM and IHS. And with the warpage (unless there is some way I'm unaware of) there's no fixing that. Unfortunatly

From what i have heard It's been a common problem with Skylake.
 
Last edited:
Holy cow O.o that cpu is a gonner for sure, unfortunately..
I didn't know Skylake cpus could bend like this from heat stress. I'm assuming you have over tightened the cpu block to begin with, even though you did not use any tools apart from your fingers. If these cpu are so prone to bending, any extra pressure will be enough to kink it slightly. That combined with the extreme temps (as a result from the initial bend, either reducing the contact of the die with the IHS and/or not allowing the block the have good contact with the TIM and IHS) sure did no good to that cpu. I'm sorry for your loss, but you'll have to open up that wallet again..
 
Wow, learn something new every day. I can honestly say I haven't heard a thing (or seen) an instance of this myself until now. From some of the pics I pulled up it seems like the PCB on the skylake's are substantially thinner.................EWWWWW

When I wrote that I was thinking about some of the sandy and ivy bridge's I've seen warp like that (purely heat related from OC on a stock cooler)

I don't think Skylake being 14nm really has anything to do with it. The die doesn't carry any of the clamping force. The pcb does. And its a .8 mm PCB VS a 1.1mm PCB for the DC series. But I would guess (assuming logic holds) the PCB is .3mm thinner because the FIVR is back on the MB. But who knows it could just as well be Intel wanted to save .0000000000000001 cents per processor............LOL

In any case. What ever the reason, be it clamping force + heat Or Heat over time or gremlins I would lay dollars to bones that the die has at least partially separated from the TIM and IHS. And with the warpage (unless there is some way I'm unaware of) there's no fixing that. Unfortunatly

There was a big hoohar not long ago about Skylake PCB's bending/warping, AFAIK Intel used a thinner PCB compared to Haswell which is one of the main reasons for the warping.
 
This was every where three months ago. One or two cooler manufacturers changed their mounts over this. There are several posts over that.
 
Wow, learn something new every day. I can honestly say I haven't heard a thing (or seen) an instance of this myself until now. From some of the pics I pulled up it seems like the PCB on the skylake's are substantially thinner.................EWWWWW

When I wrote that I was thinking about some of the sandy and ivy bridge's I've seen warp like that (purely heat related from OC on a stock cooler)

I don't think Skylake being 14nm really has anything to do with it. The die doesn't carry any of the clamping force. The pcb does. And its a .8 mm PCB VS a 1.1mm PCB for the DC series. But I would guess (assuming logic holds) the PCB is .3mm thinner because the FIVR is back on the MB. But who knows it could just as well be Intel wanted to save .0000000000000001 cents per processor............LOL

In any case. What ever the reason, be it clamping force + heat Or Heat over time or gremlins I would lay dollars to bones that the die has at least partially separated from the TIM and IHS. And with the warpage (unless there is some way I'm unaware of) there's no fixing that. Unfortunatly
Indeed there was a huge hoohaa about it shortly after Skylakes' release and the cases were strong, I attribute this to user error and clamping force more than Intel being at fault. "When you hit firm, stop tightening"

Some examples of bent CPUs.
qOwWpL6-sgaV.878x0.Z-Z96KYq.jpg

dy40KGoOWoMs.878x0.Z-Z96KYq.jpg

Curved_CPU_bottom-pcgh.jpg
 
Indeed there was a huge hoohaa about it shortly after Skylakes' release and the cases were strong, I attribute this to user error and clamping force more than Intel being at fault. "When you hit firm, stop tightening"

Some examples of bent CPUs.
qOwWpL6-sgaV.878x0.Z-Z96KYq.jpg

dy40KGoOWoMs.878x0.Z-Z96KYq.jpg

Curved_CPU_bottom-pcgh.jpg

while I agree that user error is the cause to generalise it by saying firm i think is wrong. everyone is different therefore everyones judgement of firm is different. To me firm is when I cant turn something any further. that is one of the reasons i like eks precise mount, you could only turn it until it clicked and that was the perfect amount of pressure applied. I like to tighten my screws very slowly from the very top of the thread until i begin to start feeling the resistance from the spring goming into contact with the nut while counting the amount of revolutions that the nut is making. From there i give it a few more 1/4 turns each so that i apply smaller amounts of pressure evenly. when it gets to a point where i feel that it is almost enough i try and and give the block a gentle wiggle side to side (GENTLE). If it moves I know that that amount of pressure wasnt enough so i strip it and start again and counting the revolutions go until there is no wiggling.
 
Back
Top