Inverse: A Highly Versatile Steam-Box Design

Discussion in 'Small Form Factor Systems' started by esplin2966, Feb 4, 2015.

  1. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    I have the 500W SFX-L PSU in my room, and I'll test with it next. A little nervous now! :eek:
     
  2. SaperPL

    SaperPL [H]ard|Gawd

    Messages:
    1,808
    Joined:
    Aug 31, 2014
    Is it the silverstone or chieftec? I think chieftec has longer cables - 450mm while silverstone has 350/400 cables

    little tip - sfx-l fan grill has to be mounted upside down :)
     
  3. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    It's the silverstone PSU. Thanks for the tip! Saves me a lot of headache :D
     
  4. {NG}Fidel

    {NG}Fidel [H]ardness Supreme

    Messages:
    6,291
    Joined:
    Jan 17, 2005
    This looks amazing.
     
  5. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    So glad you like it! Don't hesitate to let me know if you have any comments about the design or anything else :)
     
  6. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    Do you think there is space to put a 3.5" disk over the motherboard when an 240mm AIO is present, and still rig a couple of SSDs somewhere?
    I'm thinking of a X99e-ITX/6-core build cooled by a 240mm AIO, and all the hard disk a person can use :)
     
  7. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Whether a 3.5" drive will fit over the motherboard depends on the height of the water block. For example, the cooler master seidon 240M (which I have ordered) has a water block that is only 27mm tall. In that case, I think a 3.5" drive might just barely fit over it. I'll test it to make sure.

    I think that there might be space for a 2.5" drive between the SFX (not SFX-L) PSU and the 240mm radiator, though there are no holes there for it right now. I will check, and if it fits I will add mounting holes in the next iteration of the design. Another possibility is, if I find in my testing that having fans over the GPU doesn't really help with the cooling, I might move the GPU up in the next design iteration to make room for hard drives/slim ODDs beneath the GPU.
     
  8. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    I wonder how possible it is to have 3 slots and the ability to mount closer or further away from the air holes, depending on whether the user wants to use the case fans.
     
  9. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Hey guys, just put together the build and...

    [​IMG]

    ...IT LIVES!!

    Preliminary temperature tests are promising :) I'll do a more detailed write-up once I do a systematic test and gather up enough data.

    Triple slot? It might be insane to hold one of the Powercolor Devil13 dual GPU cards in this case. I'll have to check the dimensions, but it might work. Good thinking!
     
  10. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    Oh no... Idea stolen from Ncase M1. But the flexibility is great if someone using 2-slot GPU without case fans can mount the GPU from the second slot onwards, closer to the holes to prevent air recirculation. And if someone wants to mount 2 slot + fans or 3-slot GPU can mount it further away from the holes, if you get what I mean.
     
  11. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Yeah, I understand exactly what you mean. Now I just gotta figure out if there is enough vertical space to do it :rolleyes:
     
  12. iFreilicht

    iFreilicht [H]ard|Gawd

    Messages:
    1,348
    Joined:
    Sep 23, 2014
    Well you could also make two double-slots that are offset by less than the regular 20.32mm so you can mount the GPU close to the panel or ~15mm away. So you'd have 2,5 slots if that makes any sense.
     
  13. EdZ

    EdZ Gawd

    Messages:
    773
    Joined:
    Jan 28, 2008
    Should be easy with a fully open rear and two pairs of hole positions, but it may be tricky to hold the bracket tabs in the upper position (a protrusion to hold the lower tab in the upper position would impede insertion in the lower position). Maybe if the tab retention bracket were a separate plate, it could be 'reversed' by removing it, rotating it 180°, then reattaching it.
     
  14. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Hey everyone, I'm getting ready to run some systematic tests. Here is what I have planned so far:

    1) CPU Temperature Tests - With Prime95
    a) With outer panels off
    i) Intel Stock Cooler
    ii) Silverstone AR05
    iii) CM 120V AIO Cooler
    iv) CM 240M AIO Cooler​
    b) With outer panels on
    i) Intel Stock Cooler
    ii) Silverstone AR05
    iii) CM 120V AIO Cooler
    iv) CM 240M AIO Cooler​

    2) GPU Temperature Tests - With Unigine Valley
    a) With outer panels off
    i) With fans over GPU
    ii) Without fans over GPU​
    b) With outer panels on
    i) With fans over GPU
    ii) Without fans over GPU​

    3) Structural Tests
    Put my 50 inch TV on top of the case (including components in it) and look for buckling, bending, and other modes of physical failure.

    4) Noise Tests - Measure noise in dBA with noise meter
    a) Idle
    i) With fans over GPU
    ii) Without fans over GPU​
    b) Load
    i) With fans over GPU
    1. CPU with Prime95
    2) GPU with Unigine Valley​
    ii) Without fans over GPU
    1. CPU with Prime95
    2. GPU with Unigine Valley​

    5) Special Noise Test - Measure noise in dBA with noise meter
    a) With case inverted
    b) With case upright​

    A few questions:

    1) With the temperature tests, should I keep the fans spinning at a constant speed, or let the PWM take care of it?

    2) Is a specific kind of test that you would like to see? Let me know and I will add it in.


    Hmm, I'm not too sure I understand what you're saying. Are you saying take 2 double slot holes, and then put them 15mm apart? Wouldn't the holes overlap?

    I think I understand what you're saying, and I think it's a smart idea. I do hope that a triple slot will fit though, since it would be simpler to design and manufacture.
     
  15. iFreilicht

    iFreilicht [H]ard|Gawd

    Messages:
    1,348
    Joined:
    Sep 23, 2014
    The tests look fine to me, don't see anything obvious missing.

    What I mean is take the sketch you've used for the mounting holes and cutouts for the dual PCIe slots and apply it a second time, just a few mm's higher. That way, you will have one big cutout and four mounting holes for the PCIe brackets in sets of two, but the user can either mount the GPU on the upper set or lower set, depending on whether they want to use fans or not.

    Some people may also want to use 2.5" SSDs below the GPU, those shouldn't have too much of a problem with the heat. Not sure whether mounting holes for drives there exist already, I think they do.
     
  16. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Oh, I get it now haha. That shouldn't be too bad to implement, though the problem EdZ mentioned about tab retention brackets still exist.

    Yeah, I think having SSDs below the GPU (with something in between to block the heat) will be very useful, especially for the configuration with 240mm AIO coolers. Maybe I can also fit slim ODDs, which is a feature quite a few people want.

    Do you think I should keep the fan speeds constant for temperature tests or just stick with PWM?
     
  17. iFreilicht

    iFreilicht [H]ard|Gawd

    Messages:
    1,348
    Joined:
    Sep 23, 2014
    I'm not sure about that, maybe you should do both? No clue to be entirely honest, maybe try to look at the tests reputable sites do and try to imitate that? :confused:
     
  18. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Doing both is seriously tedious :(

    I'll check out what other tech websites do.
     
  19. reactor451

    reactor451 n00b

    Messages:
    12
    Joined:
    Jul 16, 2015
    I'd personally like to see PWM tests since that would represent better real world conditions. Either that or just run low fan speeds when running the idle temp test and then run them on full when stress testing. Just my $0.02 :)
     
  20. EdZ

    EdZ Gawd

    Messages:
    773
    Joined:
    Jan 28, 2008
    PWM, with the fan speed graphed along with temperature.
     
  21. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Looks like everyone wants to see tests with PWM, but also want fan speeds (in % maybe?) plotted with the temperature, so that's what I'll do.

    Here is an updated test list. I've also added one more test for various PCIe riser cables.

    1) CPU Temperature Tests - With Prime95
    a) With outer panels off
    i) Intel Stock Cooler
    ii) Silverstone AR05
    iii) CM 120V AIO Cooler
    iv) CM 240M AIO Cooler​
    b) With outer panels on
    i) Intel Stock Cooler
    ii) Silverstone AR05
    iii) CM 120V AIO Cooler
    iv) CM 240M AIO Cooler​

    2) GPU Temperature Tests - With Unigine Valley
    a) With outer panels off
    i) With fans over GPU
    ii) Without fans over GPU​
    b) With outer panels on
    i) With fans over GPU
    ii) Without fans over GPU​

    3) Structural Tests
    Put my 50 inch TV on top of the case (including components in it) and look for buckling, bending, and other modes of physical failure.

    4) Noise Tests - Measure noise in dBA with noise meter
    a) Idle
    i) With fans over GPU
    ii) Without fans over GPU​
    b) Load
    i) With fans over GPU
    1. CPU with Prime95
    2) GPU with Unigine Valley​
    ii) Without fans over GPU
    1. CPU with Prime95
    2. GPU with Unigine Valley​

    5) Special Noise Test - Measure noise in dBA with noise meter
    a) With case inverted
    b) With case upright​

    6) PCIe Riser Test - Compare Unigine Valley Benchmark Results
    a) Shielded 3M PCIe riser cable
    b) Unshielded PCIe riser cable
    c) Unshielded PCIe riser cable with EMI shielding tape over it​

    Let me know if you guys think of anything else to test! Thanks!

    Thanks for the feedback guys! I'll go for PWM with fan speed plotted over temperature.
     
  22. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    If there is a vertical tunnel between the separator between the GPU and the motherboard area, an AIO can possibly be used to cool the GPU. This may improve support for things like Fury X and Kraken G10. Not sure about the EVGA GTX 980Ti Hybrid since the tubes come out of the top of the GPU. The tunnel can be sealed up with a plate and screws for those who use air cooling for GPU.
     
  23. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    That's a good idea, and not too difficult to implement either. Just to clarify, when you say "tunnel", you're talking about a hole on the compartment divider wall, right?
     
  24. Nobu

    Nobu 2[H]4U

    Messages:
    3,173
    Joined:
    Jun 7, 2007
    Sounds like what he meant. To work with AIO units, it'd have to be open all the way to the edge, though. I'd be fine going at it with a dremel, but I realize that's not an option for everyone. lol
     
    Last edited: Sep 20, 2015
  25. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    Yes, I guess it has to be open along the height of the case so I called it a tunnel. :)
     
  26. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Yeah, I definitely want to minimize the amount of permanent damage people need to do to make the case work lol.

    I gotcha. Shouldn't be too hard to add to the next design iteration.
     
  27. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Hey cowsgomoo, just pm'ed you. Let me know what you think!
     
  28. cowsgomoo

    cowsgomoo Gawd

    Messages:
    547
    Joined:
    Oct 8, 2012
    Any updates on temperature tests? :)
     
  29. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Hey man! Just finished the temperature tests. Will do a write-up in the next few days.
     
  30. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Hey everyone, here are the results of the temperature tests for the !nverse case.

    Test Configuration:

    CPU: i5-4690K
    CPU Cooler: Cooler Master Seidon 240M
    Motherboard: Gigabyte Z97N-WIFI
    RAM: G.Skill Ares 2400MHz CL11 2 x 4Gb
    Storage 1: Sandisk 120Gb SSD
    Storage 2: Seagate 500Gb 7200RPM HDD
    GPU: Gigabyte R9 390 Gaming
    PSU: Silverstone SFX SX600-G
    Case Fans: 2 x Silverstone FW121

    The CPU is overclocked to 4.4GHz so that the system is sufficiently stressed in order to obtain decisive results.

    Test Setup:

    Due to suggestions from cowsgomoo, I decided to drill holes all over the case in order to allow the exhaust of the liquid cooler radiator to go directly out of the case, as well as make more exhaust holes for the GPU. This setup, with holes all over, is called "Porous". A sample image of the porous setup is shown below:

    [​IMG]

    To capture the original exhaust hole configuration of the case, I tape the extra holes closed. In this setup, the exhaust of the liquid cooler radiator would have to flow past the motherboard and the PSU to eventually exhaust out the side. There are also less exhaust holes for the GPU. This setup is called "Air Tunnel". A sample image of the air tunnel setup is shown below:

    [​IMG]

    To obtain a baseline, optimal cooling scenario, I take apart the exterior panels and open up the inside of the case to the outside. In this setup, the airflow from the GPU and the liquid cooler radiator is completely uninhibited. This setup is called "Open Air". A sample image of the open air setup is shown below:

    [​IMG]

    Furthermore, I test whether raised the case higher up from the table helps with the cooling. A sample image of the raised setup is shown below:

    [​IMG]

    I also compare the temperatures of the GPU when there are case fans over the GPU, air funnels over the GPU, and just bare GPU.

    All in all, this comes to 18 possible setups:

    CPU:
    1. Open Air (OA)
    2. Air Tunnel - Raised (ATR)
    3. Air Tunnel - Unraised (ATU)
    4. Porous - Raised (PR)
    5. Porous - Unraised (PU)

    GPU:
    1. Open Air (OA)
    2. Air Tunnel - Raised - Bare (ATR - bare)
    3. Air Tunnel - Unraised - Bare (ATU - bare)
    4. Porous - Raised - Bare (PR - bare)
    5. Porous - Unraised - Bare (PU - bare)
    6. Air Tunnel - Raised - with Fan (ATR - fan)
    7. Air Tunnel - Unraised - with Fan (ATU - fan)
    8. Porous - Raised - with Fan (PR - fan)
    9. Porous - Unraised - with Fan (PU - fan)
    10. Air Tunnel - Raised - with Funnel (ATR - funnel)
    11. Air Tunnel - Unraised - with Funnel (ATU - funnel)
    12. Porous - Raised - with Funnel (PR - funnel)
    13. Porous - Unraised - with Funnel (PU - funnel)

    Test Methodology:

    To stress test the CPU, I run Prime95 Small FFTs for 10 minutes. To stress test the GPU, I run Unigine Valley Benchmark three times in a row. Aside from the CPU and GPU temperatures, I also measure the motherboard temperatures, hard drive temperatures, and fan speeds.

    The ambient temperature is maintained at 22.8 C (73 F) for all tests.

    Test Data:

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    Test Result Summary:

    [​IMG]
    [​IMG]

    From the summary above, we can arrive at a few conclusions:

    1. For all tests, the difference between the porous setup and the air tunnel setup is about 1-2 degrees.
    2. For the CPU tests, raising the case makes the max single core and max averaged core temperatures about 3-4 degrees cooler. For the GPU tests, raising the case makes the max temperature about 8-10 degrees cooler, except for the bare GPU setups.
    3. For the CPU tests, the best, non-open-air setup produces max single core and max averaged core temperature that is about 8 degrees hotter than the open air setup. For the GPU tests, the best, non-open-air setup setup produces max temperature that is about 5 degrees hotter than the open air setup.
    4. For all tests, both hard drive temperatures do not exceed 45 C.

    Next Steps:

    Right now, the case feet is about 8mm tall. I want to figure out exactly how tall the case feet needs to be to achieve the same level of cooling as the raised setups.
     
  31. fullvietFX

    fullvietFX [H]ard|Gawd

    Messages:
    1,786
    Joined:
    Sep 1, 2004
    I want to order this ASAP!!!!! :(
     
  32. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    ASAP is what I'm aiming for too! But I definitely want to give you guys the best case I can make, so it'll have to be a bit longer as we hash out some of the design details. Please bear with me a bit longer!
     
  33. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Update 10/13/2015:

    Put a 50" TV overtop the !nverse Prototype. Here are some photos:

    [​IMG]
    [​IMG]

    As can be seen, the case holds the TV over it with no problem.

    Also finished the fitment tests. Here are some photos:

    Air Cooling Configuration:

    [​IMG]
    [​IMG]

    If using SFX PSU and SFX-L PSU, this configuration fits 3 x 3.5" hard drive and 2 x 2.5" hard drive.

    120mm AIO Cooling Configuration:

    [​IMG]
    [​IMG]

    If using SFX PSU, this configuration fits 2 x 3.5" hard drive and 2 x 2.5" hard drive. If using SFX-L PSU, this configuration fits 1 x 3.5" hard drive and 2 x 2.5" hard drive.

    240mm AIO Cooling Configuration:

    [​IMG]
    [​IMG]

    If using either SFX or SFX-L PSU, this configuration fits 2 x 2.5" hard drive.

    Media Storage Configuration:

    [​IMG]
    [​IMG]
    [​IMG]

    If using SFX PSU, this configuration fits 8 x 3.5" hard drive and 4 x 2.5" hard drive. if using SFX-L PSU, this configuration fits 7 x 3.5" hard drive and 4 x 2.5" hard drive.
     
    Last edited: Oct 14, 2015
  34. iFreilicht

    iFreilicht [H]ard|Gawd

    Messages:
    1,348
    Joined:
    Sep 23, 2014
    Looking good, but I think in the first configuration only 2 3.5" HDDs fit, not 4, right?

    Also, you should really round the corners on the fan mounts, those pointy corners will definitely hurt somebody while building.
     
  35. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    I made a mistake. In the first configuration, you could fit 3 x 3.5" drives by having 2 x 3.5" drive beside the motherboard, and 2 x 3.5" drives and 2 x 2.5" drives beside the PSU. This is what's shown on the images.

    Actually, on second thought, 4 x 3.5" drives and 2 x 2.5" drives might work too. Let me test it out.

    Good thinking with rounding corners. With the paint on, the corners of the fan mounts are actually not that sharp, but people can always get hurt in random ways.

    Thanks!
     
  36. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    OK, just verified that the air cooling configuration can only fit 3 x 3.5" drives, not 4 x 3.5" drives, at least not without changing the dimensions of the 3.5" drive bracket. I'm slightly miffed by this, but it's probably a small deal in the grand scheme of things.
     
  37. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    Update 10/17/2015:

    Hey everyone, a quick update from me. I have done the PCIe riser test with 5 different risers:

    1) 300mm unshielded PCIe riser cable.

    2) 300mm unshielded PCIe riser cable wrapped in aluminum foil.

    3) 300mm unshielded PCIe riser cable wrapped in EMI shielding tape.

    4) 300mm LiHead Type B PCIe riser cable.

    5) 500mm 3M PCIe riser cable.

    Basically, using each of these cables, I run Unigine Valley Benchmark and compare the scores attained. Here are my results:

    1) System does not boot.

    2) System does not boot.

    3) System boots, but graphics crashes when running Unigine Valley.

    4) System boots, and runs Unigine Valley Benchmark fine. Average FPS is about 89.

    5) System boots, and runs Unigine Valley Benchmark fine. Average FPS is about 91.

    From this test, it is clear that DIY solutions to shielding the PCIe riser cable does not work. The LiHeat cable performs slightly worse than the 3M cable, but at less than half the price. Therefore, I think the LiHeat cables are what we'll be bundling with the !nverse case. Now we need to see what kind of bulk discount we can get.

    Thanks for reading!
     
  38. SaperPL

    SaperPL [H]ard|Gawd

    Messages:
    1,808
    Joined:
    Aug 31, 2014
    I think its not a problem of shielding or not.

    The problem is with the way your radeon draws from pci-e thus length x quality of riser. With cheap unshielded you're power starving it on pci-e power lane and your 390 is quite power hungry.

    That's of course a problem unless you've got a riser with additional power connector on the card connector side.
     
  39. esplin2966

    esplin2966 Limp Gawd

    Messages:
    215
    Joined:
    Jan 18, 2015
    So you're saying that the issue is not really the shielding, but the fact that the quality is too poor for the length I'm using? I think it might be both, since the unshielded riser started working once I put shielding tape on it.

    I can try a PCIe riser cable with additional power connector and see if that helps. The R9 390 is 275W, and it has a 8 pin and a 6 pin cable to it, which gives 225W, leaving the PCIe slot to give 50W of power. While the PCIe slot on the motherboard should be able to provide 75W of power, it may be that the length x the quality of the unshielded PCIe cable lowers that number.
     
  40. Runamok81

    Runamok81 [H]Lite

    Messages:
    86
    Joined:
    Oct 6, 2005
    Esplin2966, the solution is the LiHeat. Go with it.

    BUT! In the sake of science, and if you have the time... I would love to see SaperPLs theory tested. I am curious to see if an EMI shielded POWERED riser would net better results than your current unpowered riser. The thought being that if it makes a difference, then we could probably convince LiHeat to release a powered version of their product?