1. OCAU Merchandise is available! Check out our 20th Anniversary Mugs, Classic Logo Shirts and much more! Discussion in this thread.
    Dismiss Notice

Bad Loop Design - Recoverable?

Discussion in 'Extreme and Water Cooling' started by Sunder, Jun 23, 2024.

  1. Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,604
    Pretty sure I've broken every single "accepted wisdom" with watercooling loops on this build.

    1. "Don't worry about loop order. The difference in temp between parts of the loop are 2-3*C at most". Yeah? Well explain why in the below photo, the left hand tube (Coming from the radiator, deep purple) is 22*C, after the video card (orange - directly above the word "G.Skill), it's 27*C, and after the CPU (Bright yello, the one the right of the G.Skill memory), it's 32*C?

    upload_2024-6-23_14-3-17.png

    2. "Don't worry about tube diameter or fittings being restrictions. In the scheme of things, it's trivial" I don't doubt you're right, but you underestimate my desire to keep putting more and more things in the loop.

    *sigh* Yep, aside from the piping you see there (There are a couple spacers and right angle adaptors that are hard to see in a thermal image, I use a through connector to enter and exit the case. That's a 16mm hard pipe to the through connector, to a 13mm soft tubing barb underneath. Of course, it then exits a different part of the case, and and does a 13mm barb, to through connector, to right angle swivel, to 13mm barb, to almost 1m of hose to a Mo-Ra3 420, before coming back.

    Each individual fitting might be not very restrictive, but I've got a lot in there.

    As a result, one of the bigger bumps on the market (EKWB Kinetic DDC - 7m head, and 1000L/hour pump), has slowed to a trickle.

    I've gained about 8-10*C on what the loop was before without all the fancy piping. It's workable, it's quiet, but performance is much more like air than a custom loop.

    What are your thoughts? Is this fixable? I can put a larger in-line pump to keep the water moving faster, though that will probably mean the end of silence.

    It's also a pain to bleed. With the water going so slow originally, I thought I had a vapour lock somewhere. Interestingly, straight after the CPU I do very slowly get air bubbles. I don't know why there and not in higher parts of the loop, but it takes several weeks to build up. I suspect the water temp is getting high enough that dissolved gasses come out of solution at that point.

    I'm going to admit this is a shit design. But is it recoverable? Anything I can do to make it work more efficiently?

    Thanks.
     
  2. ashenIC

    ashenIC Member

    Joined:
    Sep 21, 2019
    Messages:
    594
    Pump->Radiator->CPU->GPU->Reservoir->Pump
    or
    Pump->CPU->GPU->Radiator->Reservoir->Pump
    or
    Pump->Radiator->CPU->GPU->Radiator->Reservoir->Pump

    The first will have warm water passing through the pump the second won't, the third is best of both. You want pressure if you have to move water against gravity or a restriction. The less restriction you have in a loop, the more flow you will want. Also keep tubing to a minimum.
     
    Last edited: Jun 23, 2024
  3. The Beast

    The Beast Member

    Joined:
    Jun 27, 2001
    Messages:
    6,149
    Location:
    Back down unda.
    Isn't that that literally how a cooling loop works?

    If the water didn't heat up after passing over hot components then it's clearly not doing it's job to thermally transfer energy to the radiator.

    I've never checked my loop with a thermal imaging system, so I don't know if this is typical, but if there is 32C water going out and 22C water going in then I would say things are working. Right?

    Show us your loop. What do you have in there? Generally speaking the blocks will be the highest restriction, fittings and radiators don't slow flow that much.

    What was the change exactly? What did you from and to in order to get a 8-10C rise? Not clear from your post.

    If you have a Mo-Ra and two blocks then your build isn't that restrictive, why not use a D5?
     
  4. OP
    OP
    Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,604
    Yeah, but usually the water flows fast enough that there is a minimal difference in before and after a block

    If you have an ideal system, the water at the hottest part is usually 2-3*C above the coldest part. A gap of 10*C indicates the flow is way too slow.


    Kinda hard to show in a single photo, but it's:

    Pump -> pass-through (every time I say Pass-through, it includes the two connectors) -> GPU -> CPU -> Pass-through -> Pass-through + right angle -> Mo-RA3 -> Pass-through + right angle x 2 -> Reservoir -> Back to Pump. There's also a few spacers/offsets to make the hard pipe a bit easier to line up

    Sorry, I should have mentioned that. While I was waiting for all the hard piping bits to arrive, I had it built basically on a piece of scrap timber (Just drilled in the stand-offs from a template), and everything joined together using the shortest possible soft tubing. The blocks, radiators, pumps and reservoir was all the same, but no pass-through or right angles. I suspect all of those are what's adding up to cause the extra resistance, but also, when you've got your build on a piece of board, and your pipes are flexible, it's also very easy to bleed and make sure you don't have vapour lock.

    The flow rate is REALLY low. I'm trying to figure out if that's "normal" for the number of connectors I have, or I have a hidden vapour lock I can't see, or if there's something else I've done wrong.
     
  5. The Beast

    The Beast Member

    Joined:
    Jun 27, 2001
    Messages:
    6,149
    Location:
    Back down unda.
    By pass through do you mean quick disconnects?

    This is generally no where near enough to cause the flow rate to drop as you're describing. I reckon there is a hard blockage somewhere.

    Hard to diagnose with just these descriptions, but if all you've done it remove some soft tubing and replace it with hardline then you should see almost no change to flow rate.

    A few pic of your build would help a lot in understanding the situation here.
     
  6. OP
    OP
    Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,604
    No, by passthrough I mean one of these:

    upload_2024-6-23_21-19-14.png
    upload_2024-6-23_21-19-30.png

    Drill a hole through the case, push this through it, then a connector on either side.

    I did have quick connects originally, but I noted that 16OD hard tube = 13mm ID, and those quick connects look like about 5-6mm ID. So I removed those, and replaced all the 10mm ID with 13mm ID soft tube. That actually did help a little.

    It was a LOT more than just replace soft tubing to hard tubing. Because mounted on the board, it didn't have to pass through anything, no right angle connectors, no pass throughs, and at the time the radiator was sitting on the desk, not mounted to the under side of it.

    I think you're right - what I've done is not enough to cause the change in flow. I'm not convinced it's a hard blockage though, because I've checked. But I have completely flushed the system and refilled it. I reckon somewhere Testing it for leaks now, and will give it another go tomorrow. I reckon I have a second vapour lock somewhere I can't see - like maybe inside the radiator, or one of the soft pipes I can't really check in.

    Thanks for the sanity check so far. I'll let you know if it's better tomorrow.
     
    The Beast likes this.
  7. Butcher9_9

    Butcher9_9 Member

    Joined:
    Aug 5, 2006
    Messages:
    3,158
    Location:
    Perth , St James
    Do you have a flow meter?

    Have you tried bypassing a few of those components with soft tubing to see if you can figure out where the restriction is coming from? Don't even really need the system on if you have a flow meter just remove parts and check the flow.
     
    Last edited: Jun 24, 2024
  8. jjjc_93

    jjjc_93 Member

    Joined:
    Jul 3, 2009
    Messages:
    5,240
    Location:
    Perth
    If you have a 10c delta in water temperatures that would suggest flow rate is far too low to me. I had a temperature sensor at the beginning and end of my loop and saw a difference of 3c @ 120l/h with a load of ~550w.
     
  9. OP
    OP
    Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,604
    I own one. In the soft tube loop, it was showing 18L/m which is actually above what the EKWB DDC pump is rated for, so I was pretty suspicious of its accuracy. (Barrow - so a cheap one). I didn't reinstall it in the hard tube loop.

    The most I could easily bypass is the radiator. as others have suggested, the Mo-RA3 really isn't a big restriction, and it didn't seem to help.

    Yeah, that wasn't really under doubt. The real question was why. I still don't think I have the full answer. but I have an improvement. After the flush and bleed, it improved somewhat:
    upload_2024-6-24_21-47-55.png

    Still a visual difference, but it's about 5-6*C now (Camera auto-scales the colours, so you can't really compare the two photos, but even within the same photo, you can see the pipe coming off the video card is no longer bright orange, and the one coming off the CPU is no longer bright yellow.) I think there may still be bubbles trapped somewhere lowering the flow, but this is much less concerning now. - Actually, you can see a very small one in the visible light photo below. Top of the pipe coming from the CPU. It looks like one large bubble, but it's actually a series of small ones. I'll give it a week or two to collect more bubbles there, then bleed again.

    Visible light photo of the set up. Cabling (meant to be the main reason I built it like this) is a bit untidy simply because I've been tilting and turning and adjusting everything to get it bled. Once I'm happy I'll tidy up the cables.
    upload_2024-6-24_21-49-54.png
     
  10. The Beast

    The Beast Member

    Joined:
    Jun 27, 2001
    Messages:
    6,149
    Location:
    Back down unda.
    I still think the single DDC is struggling here. Can you add another pump?
     
    Sunder likes this.
  11. OP
    OP
    Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,604
    I tend to agree with you. There is definitely room for a second pump in that black acrylic box (It just holds the PSU + hides all the cables). But I also have a water injection pump from an abandoned Lancer Evo tuning project (wow... Nearly a decade ago). It happens to be BSP 1/4, which happens to be G1/4 thread compatible... Only down side is that at 12v, it's rated to 45psi. I suspect that may blow off some hoses if there is a big enough restriction to cause a substantial pressure difference.
     
  12. havabeer

    havabeer Member

    Joined:
    Dec 12, 2010
    Messages:
    8,257
    make sure you have all the fittings going into the blocks in the correct orientation.

    I was pulling my hair out with a loop that wouldn't pump properly... turns out I had the jet plate parallel with the fins and not across them.



    but with your first point thats how a closed loop system works. the heat load is all added together and then removed by the rad. If you hooked your loop straight up to your garden hose, the loop order would matter because you would get heat pick up over each bit of hardware and then the water would just get dumped, but it would constantly be replaced with new 20 deg water.

    our closed loop circulates the same water over and over. so it doesn't matter where the heat is added or the cooling is removing it its the same water.
     
    juzz86 likes this.
  13. juzz86

    juzz86 Member

    Joined:
    Nov 15, 2013
    Messages:
    2,787
    Location:
    Forth, NW TAS
    Yep, caught with this a couple of weeks back too :weirdo:
    Then I rebuilt and left it out by accident.
    Then I put it back in properly and temps were worse!
     

Share This Page

Advertisement: