Pump Vibration Isolation

Discussion in 'Water Cooling' started by Zarathustra[H], Mar 2, 2019.

  1. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    So,

    The main reason I went with a custom water loop a couple of years ago (build thread here) was because I longed for a no-compromise solution. I wanted the best overclock I could get, with absolute silence. I know this was never going to be fully achievable at load, but I figured during light use I could make it silent.

    I only partially succeeded in this goal.

    The rig hits great overclocks, and is certainly much quieter under load than it ever was before (due to me going overkill on the radiator capacity) and at idle the fans are completely inaudible due to my custom fan curves in my Aquaero.

    Here is an older pic (I've since swapped out all the bends for bitspower and replaced the PSU with a Seasonic PRIME)

    64767_IMG_20160818_211359.jpg



    There is just one thing that is still bugging me. There is that constant hum from the pump.

    People have suggested I probably just need more vibration isolation, so today I tried to address that.

    From my build thread, you'll see I went with a XSPC D5 Photon 270.

    photon.png


    It has a mounting bracket on the back, that has a soft pad along its length. There are 4 threaded holes that come with short little M4 screws with fine thread. When I initially installed it, I also put a little rubber screw bushing between the head of the screw and the case metal.

    I took those out today while working. Looks like they got mashed and torn:

    DSC_0324.JPG



    So, here's what I decided to try.


    There is a US company named Isolate It! that sells Sorbothane vibration dampeners. They seem to primarily market themselves to the Audiphile industry. I had some of their half inch thick dampeners left over from when I tried to solve a speaker rattling problem a while back Their dampeners are great, and should be WAY overkill for this application, so I grabbed 8 of them from my parts bin for this project.

    isolate.png

    I went to home Depot and got some 25mm long M4x0.7 machine screws and a bag of #8 washers (so the screws don't pull through the squishy rubber)

    Then I drilled holes through the center of the dampeners.

    The goal was something like this:

    (screw head)(washer)(damper)(case sheet metal)(damper)(pump/res)

    Here they are before installation:


    DSC_0325.JPG


    (these things are slightly sticky and any dust they come across just attaches itself)


    Installation was a pain in the ass due to having to compress the dampers on either side of the case while pressing the screw and the pump/res together blind, and twisting the screw hoping that it will catch. Eventually they all did, but only after holding my China made LED flashlight in my mouth to at least give me some visibility. I hope I don't get poisoned :p


    Here is an installed picture:

    DSC_0328.JPG

    (Those screws are surprisingly shiny, and I needed the flash, so that threw off the focus and metering completely, but you get the idea. This is not art photography)

    Everything installed exactly like I hoped. I can grab the res and pull it slightly back and forth, and it gently complies indicating it has a soft dampened mount.

    Success! (right?)


    Well, not so fast. Power the thing on, the hum is still there. :(

    It may be slightly quieter than it was before, tough to tell, but it is definitely not gone.

    What else can I try that I haven't tried yet?

    I don't hear other people complaining about this. Maybe they just have louder fans and don't notice it? Maybe I got a bad pump that runs a bit loud? Maybe I'm just stupidly picky?


    Any thoughts?
     
    Last edited: Mar 2, 2019
  2. Brian_B

    Brian_B 2[H]4U

    Messages:
    2,564
    Joined:
    Mar 23, 2012
  3. Teerex20

    Teerex20 Gawd

    Messages:
    802
    Joined:
    Sep 25, 2002
    hititnquitit and FearTheCow like this.
  4. Deimos

    Deimos Gawd

    Messages:
    949
    Joined:
    Aug 10, 2004
    Hmm, two enormous rads, CPU and GPU block, why not add a second pump?
     
  5. FearTheCow

    FearTheCow [H]ardness Supreme

    Messages:
    4,656
    Joined:
    May 2, 2006
  6. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    Well, they look more squished dead on than they do if you look at them from the side.

    I did think about the level of tightening before install, but I was concerned about not providing enough support, and the whole damned thing coming off. The Photon 270 is a large res, and the coolant inside (690ml according to the spec) has a not insignificant weight.

    These sorbothane feet/discs are 50 durometer on the OO Shore scale, so they are pretty soft. They need to be compressed a bit in order to provide any support.

    I did try loosening the screws by about a turn, but that's about as far as I dared go. It may have had a small impact on the vibration noise. It is tough to tell. I do have a decibel meter. but I forgot to do a proper baseline before, so I have nothing to compare to.

    Part of the problem is the screw length, and the fact that they hvae to be tightened more or less blind, because I can't see what I am doing. Too long screws and you screw straight through the mounting bracket and up against the glass, and potentially crack it. Too short and you have to overtighten the rubber insulators.

    This would be easy if I were measuring rigid blocks, but because it is difficult to predict how much this soft sorbothane needs to compress, it is also difficult to predict the appropriate screw length...

    That decoupling kit does look pretty nice though. Am I to understand there is no direct metal connection between the screws on both sides? Just rubber in the middle?


    I don't know if I want to mess with any more decoupling though. I have to move next year. After that I'll probably start looking into a setup where I can run tubes into a separate room and have the radiators and pumps in there for complete silence.
     
  7. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    Why would I want to do that?

    A single D5 has an enormous amount of power and can handle much larger loops than what I have without any problems. I'd argue most people with dual pump setups absolutely don't need them.

    The only argument I could think of is the fan argument used for pus-pull setups. The one where two fans at lower speeds are always quieter than one fan at higher speeds, given the same airflow and all else being equal.

    This might work for pumps too, if the pump only got any quieter when turned down. It doesn't. This is one of the older D5 Vario types with a 5 speed selection dial. The sound level is exactly the same at all settings, 1 through 5. The only thing that changes is the pitch. Fastest speed has a higher pitch, slowest speed has a lower pitch. I didn't find the difference between the speed settings to make any difference in how much the sound annoyed me, so I just pinned it on 5 and left it there.

    In the end, the good old engineering principle applies. KISS. Keep It Simple Stupid. The more complex a design gets, generally the worse it is. Adding a pump adds complexity, so unless it is absolutely necessary, it is much better not to.
     
  8. Teerex20

    Teerex20 Gawd

    Messages:
    802
    Joined:
    Sep 25, 2002
    When I initially had same setup,before I changed it,I drilled a few holes in mounting bracket and bought some threaded rods. I attached the rods to bracket and bent them 90° to run down to base of case to support the weight.
    I then changed it to 4 rods coming straight out of bracket going to backside of case. I wanted a floating look and it also allowed me to run plumbing behind res. I used like inch long connectors? and ran rod thru them at mounting points with washers - bracket and case ends to support the weight at mounting points and had zero problems. Theres several ways you can use to mount it. There's a video somewhere if I can find it.
     
  9. Teerex20

    Teerex20 Gawd

    Messages:
    802
    Joined:
    Sep 25, 2002
    Oops just remembered I took off back plate and ran rods directing in res mounting points. Then used long rod connectors over ends of rod at Mount points to support. Can't think of name but they can connect 2 rods together.
     
  10. Ocellaris

    Ocellaris Ginger @le, an alcoholic's best friend.

    Messages:
    18,776
    Joined:
    Jan 1, 2008
    If you have to compress the dampers more than a tiny bit, they won’t work nearly as well.
     
  11. Cranky1970

    Cranky1970 [H]Lite

    Messages:
    73
    Joined:
    Sep 2, 2017
    try a different pump? My DDC with a Alphacool top is silent.
     
    Last edited: Mar 3, 2019
  12. Deimos

    Deimos Gawd

    Messages:
    949
    Joined:
    Aug 10, 2004
    Apples and oranges. Jeez where do I even start...

    You can't compare water pumps to fans, axial fans aren't even remotely similar to water pumps.

    2 pumps make less noise than one, the primary reason is due to the restriction of flow in the loop. EVERY loop restricts flow, the greater the restriction, the noisier the pump. Speaking from experience, 2 pumps are quieter than a single pump, the easier it is for the pump to push water through the loop, the less vibration (and therefore noise) you get.

    As a point of reference, I have run two separate loops for CPU and GPU, then moved to a single loop with an extra GPU, two pumps in series and a whole lot more radiator, the two pumps in series are far quieter than the dual loop.

    Also, an extra pump adds redundancy, redundancy = good. the only trade off is the small amount of extra heat in the loop from the pump. WTF are you talking about with this bullshit "more complex = worse"? we are talking about pumping water FFS, not an engine.
     
    Cranky1970 likes this.
  13. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    What do you mean when you say "silent"? Can you literally sit in a silent room with your eyes closed listening, and not tell a difference between on and off?

    It's very difficult to compare noise levels, but here is an attempt via YouTube video.



    Am I just being too picky?
     
  14. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    In order for that to be the case I would expect the pump to be quieter at a lower setting. But it isn't.

    And yes, I know , pumps are COMPLETELY different than fans. My analogy was just to compare how two quieter sources add up compared to one louder source. For this it doesn't matter if we are talking fans, pumps, car engines or subwoofers for that matter.

    Two subwoofers of the same size tend to be ~3db louder than using just one of them.
     
  15. Master_shake_

    Master_shake_ [H]ardForum Junkie

    Messages:
    8,408
    Joined:
    Apr 9, 2012
    2 strips cut out of a thick mouse pad would do it too.
     
  16. Deimos

    Deimos Gawd

    Messages:
    949
    Joined:
    Aug 10, 2004
    Based on what? your experience with your loop with a single pump? Which apparently represents every use case scenario? gimme a break.

    WTF, why are we comparing with subwoofers now? Subwoofers are literally designed to make noise, to generate SPL in a room and db is a non-linear scale, it isn't 3 dB, it's 6 dB, which is four times louder, are you trying to say that 2 fans are four times louder than a single fan? You sir, have a seriously warped view of reality.

    Subwoofers work together to create spl, they are producing the exact same signal in an enclosed area. Fans are designed to move air and make random noise, combining random noise with random noise doesn't always = louder.

    Example, I have a heat pump directly above my PC. With the heat pump on a high fan setting, I cannot perceive the sound coming from my pc, on or off. That is because my PC does not add its sound level to the heat pump.
     
    Last edited: Mar 3, 2019
  17. FrgMstr

    FrgMstr Just Plain Mean Staff Member

    Messages:
    48,104
    Joined:
    May 18, 1997
    I have 3 D5 from XSPC in my office and I cannot hear any of them at full RPM. I always have to feel the bottom of them to tell if they are running..
     
  18. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    These are all estimated as point sources of similar sound, so they should behave the same.
     
  19. Deimos

    Deimos Gawd

    Messages:
    949
    Joined:
    Aug 10, 2004
    I don't understand your reasoning, fans should behave the same as subwoofers? Why? I'm utterly confused.
     
  20. Brian_B

    Brian_B 2[H]4U

    Messages:
    2,564
    Joined:
    Mar 23, 2012
    Pumps usually only make a lot of noise if they are rigidly connected to something that can resonate.

    Like hard mounting the pump to your case, or the pump being installed in a reservoir and the reservoir hard mounted to the case. Or using rigid piping all over to every component.

    That’s why rubber isolators can help, or using just a bit of flex tubing in a rigid installation
     
  21. Dayaks

    Dayaks [H]ardness Supreme

    Messages:
    6,720
    Joined:
    Feb 22, 2012
    I have my pumps/res/radiator in my basement and send the water upstairs. Not sure if that’s an option. In the past I’ve also put the PC in the basement and ran DP extenders (when I had a sub 0C loop).

    The rubber isolators these guys suggested is also valid. I have my pumps sitting on some packaging material because I am cheap...
     
  22. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    Sound pressure is sound pressure regardless of what produces it, as long as it is constructive and not destructive interference.

    So yes, the same rules apply to fans as do to subwoofers or anything else for that matter.
     
    Brian_B likes this.
  23. Deimos

    Deimos Gawd

    Messages:
    949
    Joined:
    Aug 10, 2004
    No they don't. Random white noise from a fan or pump does not multiply with the number of fans in the same way a speaker multiplies the volume of music. Go ahead and test it, run a single fan at full speed, then compare it to two fans at full speed, the volume will be slightly louder. That is not the same as one vs two subwoofers, the latter being four times louder, that is because subwoofer sound is not random.

    But if you want to be willfully ignorant, and post whatever garbage enters your head, whatever man.
     
  24. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    Neither fsn not pump noise is random either. For a given fan or pump, under similar conditions, the sound output will be similar and can be added just as with speaker component.

    I'm not quite sure what point you are trying to make it why you are being so argumentative.
     
  25. Cranky1970

    Cranky1970 [H]Lite

    Messages:
    73
    Joined:
    Sep 2, 2017
    Let me put it this way..I can hear my fans over my pump if that helps. ;)
     
  26. Nicepants42

    Nicepants42 Gawd

    Messages:
    737
    Joined:
    Aug 30, 2004
    Have you tried removing all screws, holding the pump/res assembly in place in the air with your arm, completely separated from the case by air, turning on your WC system, and checking the noise?

    If the pump is still making noise under those conditions, then you know that the motor/impeller is rattling inside the pump causing the noise, and isolating the pump won't help you.

    If the pump is quieter under those conditions, then you know that isolating the pump should actually work, and you also know what it should sound like under ideal conditions. In that case, I'd suggest getting a proper bottom for your reservoir and connecting the pump to it via standard tubing. This way, you can mount the reservoir where you want it with screws (since it's heavy), but can still decouple the pump by sizing the tubing connections such that the pump can be suspended in the air, sit on a gel pillow, etc., completely isolated from other metal parts.
     
  27. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    Good suggestions. Thank you. Wish I had known to check this back when I originally built it. It's too late to RMA it now if the impeööer is indeed the issue.

    That said, a new D5 isn't THAT expensive.

    I have never had this pump/reservoir apart. Does anyone know if I can buy any D5 on the market and just replace the pump, or do they tend to have some sort of customizations to tie them together?

    I could see myself grabbing a nice PWM version if it Will fit....
     
  28. Nicepants42

    Nicepants42 Gawd

    Messages:
    737
    Joined:
    Aug 30, 2004
    I'm pretty sure a D5 is a D5 when it comes to housing/thread/o-ring sizes, and any 3rd party D5 pump top should fit any D5, regardless of vario, PWM, single speed, etc.

    But it may be the case that any D5 will be audible to you, and it's possible that trying to RMA a D5 over what you perceive to be annoying noise/vibration might not be easy.

    First thing is to test the ideal condition.
     
  29. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    You're using the Photon 270, right?

    I had that one for a while and stuck an AquaComputer D5 in it. No fitment issues. You should be able to put whatever D5 you want in there.
     
    Zarathustra[H] likes this.
  30. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000

    Thanks for the info. That is good to know.

    Is there anything special abouy that Aquacomputer D5 that I wouldn't get with any other brand of PWM D5 when used with an Aquaero?
     
  31. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    The AquaComputer D5 is available with Aquabus connections, allowing you to control it via Aquabus (and thus it shows up as a pump in the software) instead of using a PWM channel (which just treats the pump like a fan). However, data point of one: the control board on my AquaComputer D5 died within a few months, and I wound up returning it for a PWM EK D5, which also fit with no issue. I was surprised at the failure, and don't know if it's a common issue, but it did happen to me.

    I'd say it's not worth it if you can spare one of your four PWM channels to dedicate to a pump.
     
  32. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    Thank you.

    I have googled for a guide on how to replace the pump on the reservoir, but come up short.

    Do you just twist the bottom off, or how does it work? Did you do any pressure testing after, or lube the o-rings?

    I am considering picking up an EK PWM D5 like you got, as well as a flow sensor and 4x aquabus based inline thermal probes, so I can have one before and after each block, and govern pump speed automatically based on the max of the delta T's across my two blocks. I decided on the aquabus probes, because they claim they are more accurate than standard probes (some sort of calibration wizardry in the electronics) and for accurate delta T between probes, trying to detect what will likely be a very small temperature difference at flow, I figured I'd need all the accuracy I can get.
     
  33. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    There's a knobby-looking retention ring that holds the pump up against the volute on the bottom of the reservoir. You just loosen that ring, and once the threads disengage the pump will come free. There's a single o-ring that seats around the outside perimeter of the metal pump plate (around where the impeller sits) and seals against the bottom of the res.

    I've never lubricated my o-rings - dunno if it's best practice or not but I've never leaked.
     
    Zarathustra[H] likes this.
  34. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    By the way, I want to thank you for repeatedly recommending the Aquero (at least I think that was you)

    I had always liked the device, but initially I balked at its rather steep price. Now that I have one I am very happy that I do. Everyone with a water loop should have one.
     
  35. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    I feel the same way. Next to the cost of the custom loop, the extra $100 is pretty minimal for the degree of control it gives you.
     
  36. Nicepants42

    Nicepants42 Gawd

    Messages:
    737
    Joined:
    Aug 30, 2004
    So if I've got two D5 varios, would the Aquero let me automatically control the speed? For instance, set the knobs to 5 (max) and then vary voltage via Aquero software?
     
  37. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    I think so. The Aquaero has four PWM fan headers, and those headers are rated at 2.5a each and can also do voltage control. You'll need to convert your pump's molex power and tach signal wires into a single 3-pin plug and dedicate one header to each pump. I've done it with a single DDC310 and had no issues, but that was a much lower wattage pump than a D5. I'd be worried about current draw, but the Aquaero has built in features that let you start pumps softly to limit current, I believe.

    Additionally, the Aquaero can slave other Aquabus enabled products like the PowerAdjust modules, which are designed to run pumps, from what I've read - that may also be an option if the current draw is too high.
     
  38. LRAStart

    LRAStart n00b

    Messages:
    8
    Joined:
    Feb 16, 2019
    This looks like it may be veering off from the original intent of the post but I love this type of DIY stuff especially for WC so ima throw in my 2 cents. I'll start by saying that I think you may want to try what Nicepants has suggested and try to see how much noise your pump by itself is actually making by removing it from the case and running it handheld. I think a D5, especially a Vario, should be pretty damn quiet on a low setting. But IMO the simplest solution, in this case buying a new pump, is usually not as fun.

    I think your vibration dampeners could be improved. You still have a rigid screw attached to your pump which could potentially cause rattle against the sheet metal of your case. Something like this would totally isolate: https://www.mcmaster.com/vibration-damping-sandwich-mounts

    Again, I would think that the majority of the noise you are hearing is probably from the pump itself, so i would start there.
     
    Teerex20 and IdiotInCharge like this.
  39. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,724
    Joined:
    Oct 29, 2000
    So, I tried setting the minimum pump speed to 50% instead of 20% and that seems to have eliminated the oscillating issue.

    It is really interesting to me how the measured flow rate scales with pump speed. At 100% pump speed, I have a flow rate just shy of 1 GPM (~0.94GPM). At 50% pump speed - however - that flow rate drops down to ~0.25 GPM.

    Anyway, I have another issue with controlling pump speed based on temperature deltas.

    Currently I have 4 calitemp sensors. CPU In, CPU Out, GPU In, and GPU out.

    A Virtual temp sensor takes the max temp value of the "out" sensors, and calls it max water. This controls the fans (I would have used all four, but apparently a virtual sensor is limited to three inputs?)

    I have two virtual temperature sensors that take the delta over the CPU and GPU respectively, and then I have a final virtual temperature sensor that takes the max of the two deltas. This is what I have set to control the pump speed.

    At full pump speed, full CPU load results in ~0.6C delta T. Full GPU load results in ~0.5C delta T. So I set the pump to max out at 0.4C Delta T, and drop down to as low as 50% at lower delta T's

    But, damn, even these fancy Calitemp temp sensors drift like all hell.

    It's one thing to use just one (or a max of a few of them ) to control fan speed, but when using the delta between two of them the random noise aspect of the delta is often greater or as great as the signal.

    The single point calibration offset option in the Aquasuite is not helpful. Maybe a multipoint linear calibration would have worked better, but that is not an option.

    I think Aquacomputer runs into limits of how much they can actually fit in the flash of the Aquaero devices, and have to impose a lot of limits on what you can do.

    I don't mean to sound like I am complaining. These are by far the best fan control devices I've ever used, but man, they could be so much better...
     
    Last edited: Apr 20, 2019
  40. VanGoghComplex

    VanGoghComplex [H]ard|Gawd

    Messages:
    1,675
    Joined:
    Apr 5, 2016
    Oh, you're using delta-t across the blocks? That's interesting... And the oscillating makes sense. Every time I've seen the "delta-t" figure used in reference to controlling a watercooled loop, it's always the delta between coolant and ambient temperature.

    My setup uses two coolant temperature sensors - one at the hottest and one at the coolest parts of the loop - averaged together for an overall coolant temp. Then I have three ambient air sensors - one each set on the fan frame behind my three 140mm "intake" fans - also averaged together.

    The virtual sensor that controls my fans is the absolute value between those two averages. The closer my coolant is to ambient, the slower the fans turn.

    I don't modulate my pump at all. As has been said, the pump seems to be the same volume all the time, just a different pitch - so I set mine to the highest speed that isn't occupying the resonant frequency of my case or desk and leave it be.