Watercooled Gaming Rig Is Cactus

Hi team,

1.5 years ago I built a water cooled gaming rig. At the time I wanted to do something different and watercooling was a new challenge. For about a year everything went great, it was reasonably high end for the time, Intel i7 CPU, RTX 2080 GPU, 32 Mb of RAM Nvme drive for the OS and 2 x SSD drives for local storage. I played around with some very very mild overclocking but none of the games i was playing needed any serious overclocking (at the time Witcher wild hunt etc) so I just plodded on. Ran the temperature monitors and never overheated.

Just after the start of Iso, around June 2020 I started getting loads of crashes, random times. I started fault finding. Event log was showing power crashes. I did all manner of virus scans (that came up clear) using different malware and virus tools (Kasperky, Malwarebytes, and a couple of the free ones as well). I reinstalled Windows, I changed some Windows settings, I updated all of the drivers and bought and ran several diagnostic programs/tools. Nothing touched it and everything pointed to a hardware problem. I started switching parts:

  • New larger Power Supply, this was my initial thought but it made no difference
  • New UPS, I wondered if I had a dodgy local power supply so I replaced my 10 year old UPS (which was on its second battery)
  • New motherboard, I attempted to rollback a BIOS update on the old motherboard and, you guessed it, the system decided to crash half way through the BIOS update essentially bricking the MB. I tried removing the battery but the MB seemed bricked so I replaced it.
  • New CPU
  • New waterpump

Nothing touched it. I also ran a battery of tests on the SSDs and the memory. I flushed, cleaned and replaced most of the hoses in the cooling system. Pulled the water blocks for the CPU and GPU off and cleaned them.

Every time I thought I must have fixed it it went back to the same random crashes. Once after replacing the MB it ran for a day without crashing and i thought I had fixed it until ….. the next day it started all over again.

Massively frustrated i finally decided to seek professional help, tried a couple of local computer shops but as soon as I mentioned water cooled they ran a mile. Unless it was a laptop or I was a pensioner who needed help setting up my internet browser everything seemed too hard. At the end of my tether and sick of throwing money at the thing I eventually went back to the shop where I bought most of the parts and all of the watercooling stuff, PLE computers. they told me to bring it in and drop it off and they could do some diagnostics and have their watercooling guru have a look at it. The cost would be $79 to diagnose and then $120 an hour for any work.

This afternoon i took the rig out to PLE, rang them first to make sure they were open and let them know I was coming. Arrived and spent what seemed like ages whilst a junior guy asked me a few questions then went out 'the back' to ask the expert then came back with a suggestion (might be the graphics card etc) then when i asked if I could just drop it off and pay for their diagnostic service rather than drop another $1K on a new graphics card, he had to go out 'the back' to ask. In the end I was having a remote conversation with the young guy who would relay my responses to someone out the back and come back with half answers. Ended up I was going to have to sign some kind of contract indemnifying them before they would look at it (which I was OK to do), pay them the diagnostic fee (also OK to do) and the insinuation was that i would have to give them the OK to rack up the hours as needed at $120 an hour. I asked for an estimate and was told they couldn't give me one until they pulled it apart (which i understand). I said that I might end up spending more on their investigation and parts than what it would cost for a new PC and the young guy told me to wait again until he relayed this to the guy out the back. Frustrated i told him not to bother, picked it up and left.

Not sure what to do now:

  • Look for another guru that might be able to repair it?

  • Buy a barebones system that I know works and start switching in any parts I can salvage (PSU, RAM, drives etc) until something goes cactus

  • Write it off and buy a new system acknowledging that the busted one is 2 years old and a new one would be the newer architecture etc.

I don't want to keep throwing $$ at a money pit! It could literally be a dodgy cable connector somewhere causing it all or it could be the graphics card. At the moment it won't even hit the BIOS screen on startup.

Any recommendations?

Comments

  • +2

    Username checks out

    • Strong first comment

  • Seems you've done a lot of troubleshooting already. Does it crash without graphics card (or can you not check as CPU has no integrated GPU)? Also is 32GB enough for the OS?

    All 3 options look fine to me but for me it would depend on: (1) my current level of frustration, (2) how soon I want it fixed (3) how much willing to spend. Option 2 is good if you want to save money and have spare time. With 2, you still have option 3 up your sleeve and can sell off parts of the old system.

    At the moment it wont even hit the BIOS screen on startup.

    Sorry, just noticed your second last sentence. Are you saying it is now crashing before the bios screen displays or you cannot enter bios on start?

    • Nope cant hit BIOS on start, have been able to in the past but at the moment I'm not even getting that.

  • Not CPU, PSU or motherboard. Temps seem ok (so shouldn't be the cooler). So basically leaves memory, storage and GPU. Bad power cord can also cause random crashes.

    • tried a couple of power cords, it could be one of those little RGB cables or internal switch cables for the case etc.

  • TLTR

    • See title.

  • +1

    Oh no! Very few computer shops have the expertise to DIAGNOSE anything. They're normally a mater of tial and error swapping parts and when that doesn't work they reinstall Windows.
    You could just start at BleepingComputer BlueScreen forum
    Make sure you follow these steps

  • +11

    Remove all cards from the motherboard, hopefully there is onboard graphics, and just have the psu, ram, cpu and boot drive connected, the bare minimum to get the computer working in windows. Test for errors. If there are none, progressively add back each component one at a time and test again.

    If it doesn't work with the bare minimum, one of your core parts is still not working properly. Since you said your CPU, PSU and motherboard is new, could be RAM or storage. Also reset the bios back to defaults, no overclocking.

    It would also be helpful to know your full exact system specs.

    • Yep, I'll get a list of parts, you're right I should be more methodical with the diagnosis. Complicated with the custom loop as booting without the graphics card isnt as simple as removing the graphics card and trying to bbot, I have to drain, re-route the water loop around the graphics card, re-fill the loop and then try to boot but I guess this would be the only way to test the card.

      • +1

        also when you do this bare bones test, only one stick of RAM at a time

        Though the not POSTing you mentioned at the end implies a major component failure - CPU, MB, GPU, RAM, PSU, or cables between any of those listed

        Edit: oh, and get a cheap air cooler and piss the custom loop off while you try to pin this down - that complication will slow you down so much and just make everything worse

        Is the GPU on the loop too?

        Do you have another GPU to test with if you have no integrated graphics?

        Edit2:

        Actually, strip the lot out of the case and try to fire up just the board, CPU, RAM, and if no iGPU, GPU, on a table outside your case. Could be something in the case shorting stuff out too with all the RGB and bullshit they have these days, or a random MB standoff in the wrong location for your board layout shorting the back of the board.

        Eliminate as many variables as possible - not POSTing at all should be the easiest thing to troubleshoot IF you have the spares to swap things, assuming two of those base parts aren't rooted at once, and assuming one of them isn't blowing up another one.

        This is the expensive testing the shop was probably going to do - strip it and start from scratch, or test every individual component one by one in their own testbeds - but they should have sent the guy with a clue out to talk to you once it became clear the junior wasn't up to the job. It can be very time consuming and expensive to figure it out if you have two faulty parts or one part blowing up other new parts.

  • Don't most mobos now have a bootstrap to prevent bricking?

    • I tried swapping the coin battery to reset the mobo to defaults but it still wouldnt boot. Perhaps out of frustration I stopped there and bought a new motherboard. I might have been too hasty on the old mobo but the brand new one I put in should be OK.

  • +5

    My bet is RAM is bad, the earlier crashes and now not even getting to bios is a sign. If you have multiple ram sticks try booting with a single one plugged in, test boot each one.

    • Yep, I've ran memtest (left it running overnight) but I've always had my suspicions about the RAM.

      • So did Memtest find anything???

  • +1

    If you've run tests over the SSDs and RAM, and replaced basically everything else, then it really does look like it might be the GPU. You mentioned that you overclocked it slightly. Have you tried reducing it to base, or underclocking a bit? See if that stabilises your system, assuming you can get it to start up. Can you start it in Safe Mode?

    Also, if your RAM is 2x 16GB RAM sticks then try running with only one chip in, and if that doesn't cause improvement swap it for the other chip. That way you can rule out the RAM as being the problem at least. I assume you ran Memtest over them? If not it might be an idea if possible.

    • yep, ran memtest a couple of times, I've swapped the memory sticks over but I cant remember if I've tried booting with each one to see if one will fail. At the moment it's not even hitting BIOS on startup. i get lights on the peripherals and fans kick in but nothing else. A motherboard on its own should boot into BIOS.

      • I had a similar issue with my computer a while back. Pressing the power button would start the machine with fans running for a few seconds, then the machine would power down.

        Turned out to be a faulty power switch on my case, I switched the power sw and reset sw jumpers to my motherboard and the computer now turns on fine using the reset button.

  • -8

    Stop playing around here with the kindergarten children that don't know anything and making guesses about what might be wrong; you might find it's actually a driver causing the issue.
    You need to review the Blue Scren errors and the best way to do this would be to go to the Bleeping Computer forum. The crash dumps are your fastest way to fix this.

  • And if for some weird reason, you actually DON'T want to go to the Bleeping Computer forum, boot in Safe Mode.
    If you don't get any crashes in Safe Mode it's a driver issue.

  • +2

    Get a cheap ram stick to test it. It happened to me once, I bought a new mobo unnecessarily, when it turned it was bad ram.

  • What Cartman said is your best bet.

    I'd even go slightly further and build it all from scratch. Sometimes it's something not quite connected currectly causing the issue, but my bet would be some component not playing nice all of a sudden. Check it all, but by bit with as little in as possible to get the bare bones working.

  • +3

    Sorry to tell you this, but water-cooling should only be performed by the most knowledgeable of pc builders, if you cannot determine fault yourself you should steer clear.

    My suggestion is to switch to aircooling, then do a step by step full diagnosis for failure.

    The problem with this issue is it could be anything, or you could have multiple failures. So you have to do a full rundown of multiple replacements.

    i would replace the ram and power supply before the cpu. You basically replaced everything but the ram and video card.

    I always keep a cheap $50-100 video card that is rock stable, like a gtx 780 as a backup or needing testing.

  • Rip out the water loops, run a fan or AIO CPU cooler. Custom water is kinda silly these days.

    Flash bio with one ram stock plugged into slot 1, through BIOS interface.

    Run PC without anything plugged into headers, disconnect second drive, use only 1 ram chip.

    Is the ram listed on the mobo whitelist?

    When does it crash exactly? Gaming or when browsing internet?

    BSOD or pixelated colour mess?

  • double check power connectors, especially the ones that go from PSU to Motherboard. Some pins from ATX 24 pin power connector could potentially have melted and burnt out.

  • Hey are you using a PCI extender cable to vertically mount GPU?

    It could be that they can work fine and then randomly fail so something to consider.

    I would also try splitting your Ram sticks so run on one stick for a while and then slowly add back to check stability. It might be worth winding down whatever xmp profile you have.

  • I'm not sure which PLE store you went to, but if it was Bentley I can highly recommend this guy in East Vic Park - https://m.facebook.com/perthcomputerhelp/

    He runs his business from a dedicated room off his garage & also does house calls. I've used him twice now and he really knows his stuff, has very reasonable pricing & on each occasion I dropped off my desktop pc it was ready for pick-up later that same day.

  • My bet is a water leak, single drop is enough to short a component :/
    Water broke the old mobo and now the new :/
    Happens all the time with custom loops, just one of the hazards :(

    • I had a hose came loose in the early days of water-cooling and getto loops.

      Heard splashing in the case and everything went dark……..

  • +1

    Don't use water cooling.

    Swap out parts process of elimination.

    Possibly need another new motherboard (if leaks broke 2).

    Don't use water cooling.

  • Well looks like memory is the only one left to replace, personally that would have been the first thing I changed since it is the easily part to swap out assuming you're not watercooling it

  • That sounds really frustrating mate. I used to run quite a few water-cooled systems in the early days of water-cooling and am no stranger to dramas haha!

    Can't boot to BIOS? Does the mainbaord have diagnostic lights or LCD to see where it is stuck?

    I had a GPU failed from a previous build because the GPU water block wasn't contacting one of the mosfets properly due to the thermal pad sliding out of position when replacing the air-cooler.
    And that seemed to have cause similar symptoms as yours. Random crashes and eventually that mosfet fried and the card died.

    Have you tried a cheap GPU card to test that?

  • If your PC doesnt post, your first port of call should be seeing what error your mobo is indicating. This is usually via number of beeps (cheapest mobos), some error LED's on the mobo (mid-range mobos) or a small LCD which shows hexadecimal error codes [e.g. C5 or something] (more expensive mobos). Details on what those things mean should be in your mobo manual, somewhere towards the back pages.

  • $120 an hour.. tell them they're dreaming!!

    • Lol yes it's high but it seems to be the going rate in these busy big places.

      We charge $22 per 15 minutes in-house and $27.50 on-site BUT 1 hour minimum including travel time to the destination.

      To be fair though I know we've baulked at diagnosing home-built water cooled PCs in the past also, but mainly because diagnosing can literally take days if it's intermittent especially and we don't feel comfortable charging that much money, yet our technicians need to get paid. We usually politely decline these sorts of jobs.

  • What i7 chip you running? K series or KF?

    Do as some have suggested and go bare bones. Though first, don't just remove a part here or there, completely pull everything apart. I had a friend's pc not boot after he built it, we completely pulled it apart and restarted the build and it worked perfectly fine on second attempt. Unsure on original problem because it looked fine to me. Even things which seem trivial like removing the motherboard from the case, and reinstalling the standoff screws - do that too. Remove the PSU etc. Also take the time to clean all components for dust. Dust can cause a lot of problems, especially with over-heating certain components on the motherboard or within the PSU itself.

    I would suggest: Case, PSU, MOBO, RAM (single stick), CPU (if not KF), Hard drive only. If you don't have another cooler, you will be fine for 1-2 minutes to turn on the pc and make sure it posts and you can enter bios to reset everything to default. Best case though would be to find a stock intel cooler, probably worth $10 on ebay or something lol just so you don't comepletely overheat your CPU unneccessarily (a cpu will get unbelievably hot without a cooler very quickly even under seemingly menial tasks).

    Do a full reinstall of windows. Software can do weird things, sometimes you can work around it, other times you realise it's causing a problem that isn't worth fixing.Then, as with the build, do only the neccessary items. Start adding in additional RAM, GPU, Cooling, then start installing your software.

    Eventually you will find the problem, just be prepared to invest a lot of time into it.

Login or Join to leave a comment