Nearly fried amd 5975wx - ASRock wrx80 creator

[anton-sa]

Not sure in what order to start so I’ll just randomly blurt out what happened…
I have a Lian li o11 xl chassis with 1 x 360x60 radiator at the bottom, 1 x 360x30 radiator at the top and another 1 x 360x30 radiator at the side . This is plumbed to a strong d5 waterpump that can supposedly do 1100l/hr.
The CPU block of the threadripper pro 5975wx is a Optimus block, GPU has a waterblock aswell , GPU is Nvidia Rtx a4000.

Loop order doesn’t really matter but for kicks mine is

Resevoir primes pump, pump goes to first radiator then from 1st radiator to CPU block from CPU block to second radiator then to GPU waterblock from GPU block to last radiator and from last radiator back into waterpump.

Usually my temps are great … 5975 idles at 30odd c and peaks at around 60.
Biggest gains was with the single slot blower type Nvidia Rtx a4000 which in stock form saw upto around 80c… Now it never goes higher then 40odd c…

My pipes are.all petg hard line bends .

I’ve done plenty leak testing and machine has been rock solid .

Today … I scratched where it wasn’t itching , I opened up the “oc tuning” app, changed from high performance to power saving …

Not too long later I noticed a small drip at the one pipe by the CPU…
Next I noticed all CPU cores at 400mhz… Next I noticed the CPU temp was at 104c

At the pipe that was leaking I noticed I was able to turn the fitting much tighter …like it was loose… bottom line I felt the pipe had become malleable…

I have a strong suspicion what happened , my waterpump has a PWM header ,which I’ve connected to CPU fan header … idea behind being that if the waterpump fails then the system would shutdown as the system would see no CPU fan .

Putting the powesave mode made the waterpump turn real slow to the point that there was not enough circulation causing a hotspot at the cpu.

After I tightened the now deformed pipes I was able to start system and go into bios … There the temps were all perfect .

It’s the “oc tuning” app…

I’m a “little” worried about the Optimus water block with its acrylic base … Seems ok … not able to tighten the fittings on there any more so don’t think there is any damage …

[GigaBusterEXE]

Pwm goes off temp not power, you problems lie elsewhere

[anton-sa]

No sir , “fan speed” can be controlled by PWM signal … maybe the app has a bug?

Trust me this system has had prime95 run over 24hrs it’s had 3dmark and others run all day long … Always bulletproof . Today was the very first.time I switched to powersave mode through the oc tuning app .

Why I think Its the app that has a bug perhaps is because If anything happens and temps run rampant … the “fans” are supposed to ramp up… keep in mind 9 of my 10 x 120mm fans are on a corsair controller …only exhaust fan is on board . There is a 1 wire PWM on the CPU fan header and this goes to the water pump . It does work as when I remove.this plug waterpump runs full speed . And normally it does ramp up and down … My guess is that with the powersave function in the oc-tuning app somehow the “fan speed” doesn’t ramp.up.when temps climb

[GigaBusterEXE]

fans are controlled by pwm, which adjusts its values on the temperature
example

[einfachfaust]

Mate, what Gigabuster is trying to tell you, is that PWM is a Pulse Wave Modulated Signal, not Power like in Voltage. So it just sends off a Signal on Pin 4 to your Fan, which will then interpret the signal, lower or rise its speed, then give feedback of its RPM.

Anyways, im not too much into watercooling theese days, but your Loop sounds like one big collateral waiting to happen. 1100l/h sounds like a lot, but if the pressure isnt high enough, you start seeing problems.

I highly doubt that your fitting came loose from power save mode. Its a coincidence, nothing more. Your Energy Saving mode however sent such a low PWM that your waterpump didnt do its job right.

So start looking at this PWM, maybe you can set a Minimum in the BIOS. Also, if the CPU Fan dies, or better said the pump, your system will not be affected. Check for a CPU FAN happens only at post. At most, you could use a software like speedfan to monitor the PWM of the pump and in case of failure, ramp everything up and give you a warning.

About your corsair fan controller, how is it connected to the pump? There are numerous EHEIM and AQUERO Pumps that offer control via USB2.0 Header on the Mobo.

Im not saying your loop is a problem, but im saying take a closer look!

BR

[anton-sa]

Hi, yes I’m fully aware PWM is signal and that fan (in this case waterpump) interprets this and spins “x” speed accordingly .

The fittings were all tight prior,trust me this was checked and checked again . What did happen was due to the sheer heat of the liquid at/near the CPU fittings that piece of pipe deformed and THIS is what made the fitting loose. it just had a small “drip” type.leak it didn’t run all over the place .
The pipe became almost “soft” similiar to when I hear it up.with the heatgun to form/bend it .

Later today once I’ve replaced those two pipes I’ll definitely.be having a close look at the “fan settings” in the bios.

Waterpump is powered with a molex pwr plug, if you pull the PWM signal plug it runs full speed…so it’s definitely moderated in that the PWM signal is controlling the speed of the waterpump as it should .

I’ll definitely be having a closer look at why that oc-tuning power save mode made the signal so low that the pump was barely running …

ASRock did envisage folks potentially hooking up waterpump to the header as it’s even labelled like that on the PCB…

Anyway my loop is rocksolid never had any issues over days/weeks of testing and stress testing , my post was more intended to warn folks of how quickly things can go wrong …

I’m still puzzled however …why the “fans didn’t ramp up” in the event the CPU temp spiked… Usually… Especially in “enterprise” type gear if a fan starts failing or you remove a fan ,the rest start compensating … In this case … even in some sort of powersve mode ,CPU temp going up should have sent appropriate PWM signal to fan/waterpump to “spin” faster . It definitely didn’t
Removing the pwoersave setting solved the problem instantly and had the CPU running in the 30s Celsius .

I guess the bios isn’t set right out of box for this type of situation .

If I go look in the bios later today I’ll probably find all the fan Min values are probably like 10% or something . I’ll DEFINITELY be adjusting the appropriate header to like Min 50% from here on in …

I’m pretty sure similar as above would of happened if I was running the CPU on the.noctua air cooler (140mm fan x 2) just I would not have noticed anything other than real high temps and I guess system would have started become laggy (throttleling)

The take away here is that with STOCK standard out of box bios settings, the power saving mode on the “OC tuning” is “questionable”…

[anton-sa]

I’ll definitely be taking a close look at the Min % setting on my CPU fan header as clearly the default setting / signal isn’t making the waterpump turn fast enough … Never noticed it before as was so busy benchmarking and as such always had everything on max performance…

[anton-sa]

Ok so progress finally got a moment…
Managed to remove the two pipes without draining all 2literes of fluid …lost just a few ml of coolant
Running prime 95 …temps are decent
No leaks
Pics :
My machine
Warped pipes I replaced

[mutation666]

@anton-sa

This is plumbed to a strong d5 waterpump

@anton-sa

1 x 360x60 radiator at the bottom, 1 x 360x30 radiator at the top and another 1 x 360x30

@anton-sa

My pipes are.all petg hard line bends .

Thats a lot of restriction you prob need dual pump / max power on pump to get any kind of flow rate

[anton-sa]

Might look like it but it’s not so bad ,there plenty plenty flow … temps are in the 30s Celsius now that’s basically same as ambient where I am right now …
Busy setting up.to access bmc now to see what all I can set fan wise … pump reports 4800rpm… The latest bmc update does deal with a fan issue …
I want to lock down Min % fan(wp)
And I want to set CPU temp alarm / action

[anton-sa]

Though pricey , I’m thinking this pump is in my future … What’s the chance of both pumps failing … EK-XTOP Revo Dual D5 PWM Serial - (incl. 2x pump) – EK Webshop

[anton-sa]

So some further developments :
Did some testing/observing .

1. The issue is not that my waterloop is restrictive or theres some fault in the way its layed out, The cooling numbers proves this, cpu hovers around 30c (ambient outside temp at about 34c), and during stress test around 55c which I think is reasonable.

2)The issue is also not the “power save mode” on the A-tuning app although chain of events pointed that way.

Here is what ive now proven and I can reproduce this at will.

Anytime I put the machine in powersaving mode and then shutdown OR if the machine has been sitting for a while, And this is something I didnt notice 2 days ago at the time…
But what is happening is Ill switch the computer on and the Waterpump wont switch on (its plugged into cpu fan header)
Next if I boot into windows, by which time temps are around 80c and climbing… still does not switch on, EVEN if i unplug the pwm plug of the waterpump.

If I at this point switch the machine off and then on again, the waterpump turns on and within like 2-3 seconds the temps are low 30s celsius.

Keep in mind ive setup no fan profiles at all. everything is stock factory default.

It as if when the temp is low enough, the board doesnt pulse pwm to switch waterpump on.

The motherboard pwm seems to only turn the fan on when temps are above X degree…

when waterpump is running machine is rock solid and temps are perfect.

there seems to be a bug / misconfiguration on my part.

Also to prove this problem, I can leave the pwm plug unplugged. the waterpump as predicted switches on every and any time i turn the pc on.

I did notice when i accessed the bmc that all the fan settings / profiles were/are blank. as in there seems to be “no” default settings.

So,. in a nutshell, the board is not switching on the waterpump(fan) when it thinks the temps are all good. and it doesnt ramp up either.,

Guess I need to configure a profile for this, I kind of assumed (wrongly) that there would be some basic defaults… apparently not.

Im open to suggestions…

[alpha17]

I also have the same MB and CPU config.
What are the bios & bmc version you’re using?

[anton-sa]

BMC 2.05.00
Bios 6.06

[anton-sa]

ok, so tested the fan(bmc) settings by using standalone loose pwm fan. and in the manual mode i tested, switching it 100% 50% 70% etc works …

Dont think its anything with the board, im actually suspecting something weird with the waterpump. it works well in that when running… cpu temps are perfect. buuuut (even with pwm plug unplugged) if the system has been off a while, switching it on it doesnt seem to go on and then switching off and back on again brings it up and then its 100%

dont mine the wp running 100% but…commonsense tells me chance of it realy failing is better if it runs 100% day in and out… which is why i was hoping more for a 75% or so

its a byksi d5 pump which i believe is a xylem clone… think i need to shell out for a good quality ekwb pump or something…

next challenge, I want to set this system up so that it can shutdown can even be a nasty sudden shutdown…when temps reach X degrees.

for the life of me i cant seem to find any option to do this in the board bios.

[anton-sa]

totally bizarre, same test run on the waterpump itself. while having cpuid hwmonitor open…

put the pwm on 20% in the bmc and you can clearly see the coolant flow reduce drastically… put it on 100… and its super fast and strong…

so after a whole long long convoluted post by me, I can confirm the motherboard and its pwm etc works exactly as it should!

its my waterpump that sometimes needs to be switched on of and back on again…

my next coarse of action while i shop for the best waterpump money can buy to protect my 3300$ cpu … is to see if i can get some sort of temp cutout going.