Hydro Series H115i Extreme (high temps) and AX1200 (surges) dying simultaneously?

I'll try to capture everything that's happened in the past few days with my machine, as I've already experienced multiple symptoms and tried various fixes.

 

SUMMARY/TL;DR:

 

Computer randomly shuts down in Unity Editor, launching SteamVR or playing video games (most recently The Long Dark), and temps are high despite recently reapplying thermal paste (Antec Formula 7 Nano Diamond).

 

DETAILS:

 

About a week or so ago, maybe less, I noticed high temps on my CPU. TBH, I don't remember what prompted me to notice apart from the fact that I was trying to troubleshoot why my system is always so loud despite configuring them with the Fan Xpert 4 software that comes with Asus AI Suite and running the silent profile. The temps were in the 40s-50s, which seemed oddly high to me. So I removed the cooler plate from my CPU, cleaned it with isopropyl, applied a bead of the above thermal compound and reseated it. Following this, temps were in the 30s, which wasn't great but seemed acceptable, as we've been having a heat wave here so ambient temps were high.

 

Things seemed okay for a few days. I compared several machines in my home at the task of rendering a 3D model and textures in Autodesk Remake from high resolution photographs taken with my DSLR (otherwise known as photogrammetry). This is because I'm developing a game in which I want to capture hand-built assets to use as models, and I wanted to know which machine was fastest. I build PCs and repurpose old parts into new machines for different purposes, and have also inherited a few machines, so my house is full of PCs. My machine won, which surprised me, because two of those machines are dual-socket Xeon systems, one X79 and the other X99, and the X99 system has 3x Titan Blacks in SLI. So I was pretty pleased with the results, and didn't notice any strangeness other than the fact that after these tests - I rendered a model from the same 50 or so photos 3 times on this machine with different settings - the loudness was back.

 

I was having a hard time tracking the loudness down, as by turning my fans up by loading the Turbo profile and maxing my GPU fans via MSI Afterburner, I wasn't able to disambiguate the noise. It seemed to not be affected by either of these actions. My secondary GPU until just yesterday was a Geforce GTX 780. I had read before about problems with this card and the fans running harder than they need to. So I removed it and the machine seemed quiet again. Then yesterday I pulled a second Titan Black from another machine and added it to this machine to replace the 780. I ran another rendering test with Remake to see whether having two Titan Blacks in SLI made a difference in rendering times (incidentally, it didn't, even though 3 Titan Blacks in SLI made a huge difference over non-SLI on the X99 Xeon machine). It seemed to behave fine, although the noise was kind of an issue again. Fan noise is normal during a full load, but should abate as the load tapers off and the machine cools back down to idle. What will happen with my machine is that fan noise will increase with load, then drop after the load is finished, but not back down to typical idle levels. It remains relatively loud, just not as loud as under a full load. I also looked at my temps again and they were hovering in the low 40s after this and not going down, so the effects of re-application of the thermal compound seem to have been temporary.

 

I didn't do anything at that moment because I had work to do and the machine still seemed to be operating within safe temperatures.

 

Next I uninstalled SteamVR and all of the Vive bloatware from my machine because it has been a royal pain and I didn't realize you can just install it directly via Steam without all the HTC bloat. Then I did several hours of Unity development. The development involved refactoring input code from a Vive Controller to support Xbox 360 controllers as well; should not be particularly intensive. They do recommend something like a 980 for VR, but it's been my experience that the higher end Titans from the 700 series work just fine, and I've never had issues with Titan Blacks for VR. At one point, my machine got really loud and wonky - the order things happened gets a little hazy here. I don't remember exactly what happened, but I launched RealTemp and started a Sensor Test, and the temps immediately jumped to ~100C :!:. I've never hit the Stop button on that test to prevent the machine from shutting down before, but I had to this time. TJ Max is set to 105 in my BIOS, the default. Note that before I started the test, the temps were already pretty high, in the 60s to 70s.

 

Some time later, or possibly before that, I don't recall exactly, my machine suddenly shut down, seemingly while launching SteamVR, which happens automatically when launching a Unity app in the Editor. It usually stays on, but I had turned it off as I sometimes do. When the machine rebooted is when the PSU portion of this mystery began. The BIOS showed the message: "Asus anti-surge was triggered to protect system from unstable power supply." I should mention I've had intermittent TPU warnings from this machine on boot since I upgraded the motherboard a couple years ago, which I've never been able to track down... but I've never received this surge message before. After rebooting, I tried again to launch SteamVR and again, right as SteamVR was launching, it again shut down suddenly. Since reinstalling SteamVR, it has been doing something annoying while launching, namely all of my screens flicker several times will it disables the Aero theme, which it hadn't done before. I should note that I did the restart above with my Ethernet unplugged just in case malware was causing the behavior via remote execution.

 

At this point, I was already confused, because between the high temps, the PSU surge and the SteamVR behavior, I've already got what are seemingly multiple problems without a common root cause.

 

So I took out the second Titan, unplugged the Asus monitor and the Vive, again left my Ethernet disconnected, and inspected all the cables coming from the H115i and reseated them. I also removed the various panels on the case. The machine was off for about 15 minutes while I did all of this. After powering back on, everything seemed normal-ish again, i.e. idle temps in the mid 30s. This still seems a little high for this cooler. Ambient temps at the time were about 21C. So I shutdown and replaced everything - second Titan Black, monitors, Ethernet, et cetera and rebooted and again, everything was operating nominally. I finished my development - only required another half an hour or so if I remember correctly - tested it, and submitted it to source control. After this, temps were back in the low 40s and the machine was getting loud again.

 

It was then idle for several hours while I cooked dinner, or off, I don't remember which. My spouse went to bed and I played The Long Dark for a couple of hours. The machine was a little loud, but this is normal when playing certain games, so everything seemed fine. Then, suddenly, it shut down again. Again after rebooting I got the Asus anti-surge message. After logging in again and firing up RealTemp, temps were in the 70s. Since it shuts down immediately and I hadn't been running RealTemp or Afterburner while playing, I have no idea if it was GPU or CPU temps that caused the shutdown, or was indeed the PSU. While I was checking temps... it shut down again. At that point I hung it up for the evening and went to bed.

 

So, my theories, in ascending order of likelihood (more likely at the top):

 

1. The H115i isn't functioning normally somehow (although I can feel fluid moving through the tubing and the LEDs function, so it seems to have power and the pump seems to be working). This is causing the CPU to shut the machine down; the anti-surge messages are simply being triggered by the sudden shutdown, which the BIOS is misinterpreting as a surge event.

 

2. 1 is true, but the AX1200 is simultaneously dying.

 

3. 1 and/or 2 are true, but this second Titan Black is possibly also overheating or its firmware is infected with malware, and there is in fact a 'race condition' of sorts between the CPU and my second GPU - sometimes one overheats, sometimes the other. This is less likely though, because my second GPU is dedicated to PhysX while gaming and carries less than half the load of the other card.

 

4. The UPS the computer is plugged into is failing or being overloaded.

 

5. My cards are having a hard time pushing all those pixels to 2 4k monitors and a Vive.

 

6. The Vive is causing problems somehow after the reinstall.

 

7. Some malware has made it onto my machine and is causing all of these issues.

 

8. My 6900K is dying.

 

WHAT I'M TRYING NEXT:

 

- Once again cleaning, this time with a thermal compound cleaning kit, the water block plate on the H115i, re-applying paste and re-seating the block.

- Stripping the machine down to a single stick of DDR4, if possible, one card, no RAM coolers, and just my boot drive, and unplugging all other on-board cards and devices (more info below).

- Pulling an AX1200i from another machine, which unfortunately takes that machine offline for the moment - the other machine is a dual-PSU build.

 

...not sure what else to try at the moment. I will test the current AX1200 with the test cycle button, but if it's green I don't know how that helps me. I'm also not sure how to go about verifying that the pump on the H115i is working properly. It's a sealed AIO, so from what I've read no fluid refill is ever required; it doesn't appear to have any leaks.

 

Any help or suggestions are greatly appreciated; I'm somewhat stumped. I'm writing this post on the machine in question. I unplugged all but the two Dell monitors from it. It started up kind of hot, but has settled down to low 40s to low 50s - pretty warm for writing a forum post. Max on any core in RealTemp right now is 63C. I may be online intermittently as I work on this machine, but I will probably check in on one of my other machines from time to time. Current load on the UPS is displayed at ~270W. I haven't checked it while under load; it maxes out at 865W. There are 3 TrippLite Isobar power strips plugged into the UPS as well powering various devices like my monitors and a 24-port TP-Link switch.

 

ADDITIONAL INFORMATION:

 

A couple weeks ago I came into possession of a used Dell UP3216Q monitor. I run it in 4K over DisplayPort. I made it my main monitor. My old monitor, an Asus PB287Q, also a 4K monitor, I mounted to the left and flipped it into portrait for coding. I also have an old 19" Dell monitor connected over DVI, and a Vive headset is connected via HDMI. All but the Asus are connected via my first card; the Asus is plugged into the DisplayPort on the second card.

 

Other devices:

 

- Case front USB ports

- Xbox 360 Wireless Receiver

- USB 3 hub

- A pair of Superspeed USB 3 ports mounted in a front bay

- A hard drive power switch mounted in a front bay that lets me kill power to specific drives

- 7 fans, mostly of the 140 mm variety with a couple of 120's

- A fan controller PCB screwed to the interior of the case

- Asus wifi antenna that came with the mobo

- Rockus Sound Science speakers and subwoofer

- Powered through an APC Back-UPs Pro 1500

 

Security:

 

I'm running Heimdal Pro and Norton Security. I ran TDSSKiller and it didn't pick anything up. My wifi router was hacked, twice, a couple of months apart, prompting me to flash it with DD-WRT a couple months back; doesn't appear to have been hacked since. Since then, though, I've been capturing all of my network traffic with Wireshark with a dedicated machine and a two-way LAN tap. The machine pulling the packets is inaccessible over the network due to how the LAN tap is soldered. My mother has been having some issues with ID theft, hacking and break-ins to her house that I have been helping her with remotely (she lives in another state). I have verified some, but not all of her claims; there is a remote chance that I have been hacked by the same individuals, but I find this very unlikely.

 

History of the PC:

 

- Built originally ~2012 with the same case, but completely different components - Rampage IV Extreme, 3930K and another RAM kit (16 GB Vengeance, I believe).

- The Rampage IV Extreme mysteriously died several years ago connected to a different UPS; my machine simply would not power on and did not respond to any troubleshooting. After troubleshooting, I determined the mobo had died and replaced it with a Rampage IV Extreme Black Edition, which did in fact solve the issue. I was never able to get that mobo to POST or even power on ever again.

- My Vengeance RAM failed shortly after that and had to be replaced - with a Dominator 32 GB kit.

- I replaced my original SSD with the Samsung 840 EVO at some point. I don't recall where my old drive went or if it failed.

- When I upgraded to the current configuration, the RIVE Black, RAM and the 3930K went to another machine. Those components have never had any issues, BTW.

- As I mentioned, since installing the X99-II Deluxe, on and off I will get messages on start-up about the machine being tampered with.

- I've had the same Windows 7 OEM license on this machine since about 2014. I think I originally ran a Windows 7 retail copy. Looking at Speccy, 19 October 2014, 19:38 is the last time I reinstalled (I know, it's been awhile - probably should do that as well).

As a quick follow-up, I just remembered that I have a PSU tester lying around somewhere, so I'll use that on the PSU as well.

Ok, so I broke the whole build down except for the mobo and the front bay peripherals, which I've left in the case. I pulled the PSU and hooked it up to my Thermaltake Dr. Power II. I've attached some pics of what I've done so far. The inside of the case and the PSU are a little dirty (we have two cats and a dog), but nothing that would account for the kind of temps I'm seeing.

 

The PSU checks out on the Dr. Power II. The 5V output is always between 4.91 and 5.01, 12V is 11.9 or 12V, 3.3V is 3.3V. I also tested every single 8-pin on the AX1200 with either a PCIe or CPU cable. Every 8-pin with a CPU cable reads 11.9V and every 8-pin with a PCIe connection reads 12V. Every 6-pin connected to the Dr. Power via molex reads 5.0V and 12.0V. The fan runs smooth and quiet.

 

The thermal compound looks fine to me, not too thin or thick. However, it does seems a little wetter than it should be.

 

I included a photo of the mobo from various perspectives to show that it appears all of the capacitors are intact; I didn't see any evidence of blow-outs or burns.

 

There is also a picture attached of the plate on the H115i. Thermal compound seems fine between what is on the plate and what is on the CPU, at least to me. Coverage looks good as well.

 

I also included a photo of one of the AX1200i I'm going to swap out for the AX1200 in this machine. Even though the PSU seems ok, I'm still going to go ahead and do this. If the anti-surge message occurs again, then I'll know for sure it's bogus, as I've never had issues with the AX1200i's in this other machine. I'm currently typing this on that machine, so I guess I'll just throw it in this machine since it seems fine. On the flip side, if I get a surge then I'll know it's the PSU.

 

I also powered up the PSU via the Dr. Power and provided power to the H115i to see if the pump is spinning up. I can feel fluid moving through the tubes; there don't appear to be any leaks; the LED comes on; and I can hear the pump motor. It sounds kind of rough, but I don't know what it's supposed to sound like, as the fans are louder and I've never held me ear up to it in an otherwise quiet room before. But it does kind of sound as if it's a little wobbly to me; so it's the best possible culprit ATM, especially with the crazy temps. I'll try it again with new compound and again after that with the other cooler.

 

The last of the attachments is a photo of the cooler on a yoga block to roughly simulate the kind of position it would be in inside the case. As I said, sounds kind of rough to me but I have no idea what it's supposed to sound like.

 

...oops, looks like I can only post 10 pics at a time. Will follow up with the rest.

Here are the remaining pics of what troubleshooting I've done so far.

No comments or input? Lol, k. Well, this thread has helped me organize my thoughts and serve as a troubleshooting log, so I'll keep posting. I'm making progress, so hopefully this thread helps someone else when I home in on the solution.

 

So last night I took my second GPU apart to look for bad paste or blown capacitors. You can see from the newest attached photos that neither is the case; so that effectively rules out the secondary GPU. The paste had good tackiness and distribution and the capacitors and MOSFETs look good.

 

This morning I used ArctiClean to clean up my CPU and the H115i again. I used the spudger from an Antec Formula 7 kit, but switched to Noctua NT-H1 for the compound. You can see I spread it thin and evenly on the CPU. I usually go for the dot in the middle approach, as I've never had issues with that method before, but wanted to make sure distribution was good.

 

I re-mounted the H115i in my stripped down system and started it up. As you can see from the attached photos, I only have one RAM stick, one GPU, an SSD, one CPU fan and the cooler. All other on-board peripherals were disconnected. I booted up and you can see a screenshot of the results - temps in the 60s. The cooler itself sounded like a jetliner. The pump seemed to kick in fine, as it's quiet enough in the case now to hear that, but obviously something isn't right with the cooler, whether the pump is running ragged or something else given those temps.

 

It looks like maybe the cooler has been bad for a while, because with everything stripped down I was able to finally pinpoint the fan whine issue I've had for a while now as the cooler's fans.

 

I've swapped the PSUs back as I believe the PSU can be ruled out. Next I'm going to pull a cooler from another machine and see what that does.

 

At this point I think it's down to one of two things:

 

1. Bad cooler

2. Malware

Ok, well it's the H115i. I swapped out the cooler from the living room entertainment computer as shown in the attached photos. I didn't even clean the plate on the other cooler or my CPU, as the paste on both was fresh. At first the fans sounded like a freight train, but after tuning with Fan Xpert 4 the machine runs smooth and quietly with temps at relative idle in the 20s as you can see from the attached screenshot.

 

I'm glad it wasn't malware because I don't really want to reinstall the OS at the moment, though I need to do that soon-ish. If it turned out not to be malware after that, then I would have to address the possibility of my network being breached and do a deep dive of my packet captures and bone up on packet forensics.

 

Anyway, HTH someone trying to troubleshoot similar issues.

Well, that is definitely the long way around... On any liquid cooler with an internal coolant temp sensor, you can separate cooler problems from everything else in a few minutes. When you see your coolant temperature hovering in the 50-60C range at the desktop, you have a flow issue of some type -- usually pump failure or partial blockage. The absence of a pump speed and/or CPU boot error often gives it away. My guess is you are not running the Link software for monitoring the cooler. It isn't necessary, but has obvious uses for troubleshooting. Much easier to install it and take a look then to repaste anything. There are also other monitoring utilities like AIDA or SIV that can track the H115i Temp (coolant temp).

 

Hopefully this was your sole issue. If not, there is still a lot of ground to cover.

@c-attack, heh, yeah, I try to avoid linking up hardware with additional user-level software, particularly PSU, UPS and cooler for security reasons, but in the future I will probably install the Corsair software. I was able to get an H100i v2 delivered in under an hour with Prime Now and have already put the other H100 back in the living room PC and installed the H100i v2 on this machine.

 

I've also had some other long-term issues and wanted to investigate all of those.

 

So far, fingers crossed, temps seem to be the only issue. The PSU seems fine based on the Dr. Power tests, so hopefully there are no more sudden shutdowns with ant-surge warnings from the BIOS.

I should also note that the H100i v2 doesn't perform quite as well as the H115i, which is to be expected, but the additional room it afford me in the case means I have room for a push-pull configuration, which I didn't have space for before. When cash flow is better I may put a compact custom loop in this system, which I hadn't considered before, but considering the load I'm starting to put on this machine it might be prudent.

@c-attack, heh, yeah, I try to avoid linking up hardware with additional user-level software, particularly PSU, UPS and cooler for security reasons, but in the future I will probably install the Corsair software.

 

You can skip it and then only install if you need to troubleshoot. Some of those other packages may have other benefits for you. I am not sure I could live without Aida anymore. The downside of not using Link is you miss out on fan control via coolant temp. If you don't use link, better to run the fans from the board, which I think you are already doing. If you want it, you can get coolant temp control with a thermistor line placed at the exhaust flow point. As long as you have a motherboard that allows you to link fan speed to the probe, it's just as good.

c-attack, thanks, I might do that. I have a lot of those crufting around from past builds.

 

Also, I included a "cheesecake" shot of my machine running quiet and cool again :sunglasse.

Hello

I want a query

I want your inquiry about AX1200i

When the computer turns on the fan is operating normally after 10 seconds or 15 seconds the fan stands and does not work again for more than two hours

* I bought it from a friend of mine. He told me the temperature should be high for the fan to work

You may have already bought it new and tried it before you can help me and thank you:!:

samir06700, If I'm interpreting you correctly, you bought an AX 1200i from a friend, and when you turn on your computer, after 10 or 15 seconds the fan on the PSU stops and then the fan doesn't turn on again for a couple of hours; and your friend told you the fan on the PSU won't turn on unless the temperature is high in the PSU.

 

As far as not turning on until it gets hot, I believe this is the correct behavior, although I don't know what the cut-off is. Looking at my AX 1200 (not an AX 1200i) the fan is currently running.

 

For the 1200i, I downloaded the manual and searched it for 'fan' and found this:

 

"Upgrading to digital provides increased efficiency. The AXi Series is certified 80 PLUS® Platinum, which results in lower heat generation and a quieter fan profile. The AXi Series is so efficient that the 140mm fan doesn’t even spin at low loads for completely noiseless power delivery."

 

So that would appear to be correct. I have 1200i's and 860i's in other machines, but I've never verified to see if they do that.

 

Here's the manual link:

 

http://www.corsair.com/~/media/corsair/download-files/manuals/psu/axi_series_manual.pdf

Some things to consider:

As c-attack mentioned, use the Link software to monitor and control your fan speeds. You can also write the settings to the device's onboard memory so you don't have to have Link running. Installing Link will set up the drivers for your cooler. Once installed, the H100i V2 temps (and fan speeds) can be monitored by other software. I, personally, use HWInfo. Others use SIV. I tried Aida but I don't recall it monitoring my liquid temps. Regardless, to determine if your cooler is operating properly, you need to monitor the coolant temperature.

It's the coolant that cools the CPU. The cooler it is, the more heat it can take from your CPU. The air blowing through the radiator fins is what cools the coolant. Ideally, the radiator fans should be intake but that's not always feasible with every case. If your radiator is configured as exhaust, that air is coming from inside your case. When you are stressing your video card, that can get quite high. In this case, it may be a good idea to monitor the temperature of the air just before the fans blowing into the radiator and control the case fans based on this temperature - the job of the case fans would be to ensure cool air is available for both the GPU(s) and the radiator. Finally, if you have a push/pull configuration, you want both sets of fans to be the same and to be synchronized. If your pull fans are slower than your push fans, you can wind up with turbulence and pressure that will keep warm air around your radiator, making the cooling less effective. Also, I don't think that the fan controller for the H100i v2 would actually be able to handle powering 4 fans at full load.

After load, the liquid in the cooler will be warmer ... and will take time to cool down. If that liquid gets up to 45C, your CPU won't drop (significantly, at least) below that until the coolant also cools down. That takes time. IIRC, the default fan curve profile on the H100i V2 is "Quiet" and that may not be up to the task of keeping the liquid cool with the 6900K ... the fans only get up to 60%. With balanced, they only get up to 80%. Performance goes up to 100% but it's loud (60%) at my typical idle coolant temp (30C). A custom fan curve allows you to have low speed fans around your idle temp and ramp up as the coolant temperature increases. As the coolant temp increases (and decreases) pretty slowly compared to CPU temps, if you control your fans from your coolant temp, you won't have fans ramping up and down quickly; that's due to the heat capacity of the liquid.

Finally ... I've experienced strange issue with running Afterburner with my GPU under load with ROG RealBench, even without overclocking the GPU (just setting a custom fan curve). Closing Afterburner before running the load test allowed me to pass. Don't know if this is your issue but thought I'd mention it.

DevBiker, thanks, as I posted above it turned out to be a bad pump on my H115i and I replaced it with an H100i v2, which resolved it. I might install the Link software in the future; right now I'm running the ROG AI Suite and that seems to be working pretty well for me. In general, I seem to need to run the performance mode in Fan Xpert to keep my 6900K nice and cool for most games and productivity; if I'm just web browsing or doing something light, I can drop it into silent. I only use the full speed setting usually to quickly cool down after a performance heavy task. I could actually probably use silent a lot more, as the temps usually don't break 50, but I like to keep the processor as cool as is convenient. The freight train sound that used to come from my computer turned out to be the cooler fans working overtime trying to correct for the bad pump.

 

In the future I'm now seriously considering a simple custom loop or kit for this machine. I have a single, long custom loop with dual pumps in another machine that run through the CPU and then 3 GPUs and that machine is always nice and frosty. The AIO coolers are definitely better than air coolers generally, but they aren't as cool as I'd like. The one in my living room PC is adequate, but for my workstation I think I want more cooling. Then again, I don't like the amount of maintenance I have to put into that custom loop every year or two. I had a radiator spring a leak and had to gut it and troubleshoot the whole thing, finally discovering the radiator leak with an air bubble test.

 

My last post was responding to someone who for some reason was asking me a question about their 1200i; I have an AX 1200 on this machine, which I thought originally might be power surging, but it turns out to be working just fine. The BIOS just thought the sudden shutdowns were a result of PSU surges.