Room Treatment

Discussion in 'DIY Discussion' started by ShinOBIWAN, Dec 5, 2008.

  1. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    Shin,

    Can you post the before and after waterfall plots below say 300Hz? Those other plots make my eyes go all funny :)
     
    andyoz, Dec 10, 2008
    #41
    1. Advertisements

  2. ShinOBIWAN

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    Happy 1,000th post Andy! ;)
     
    Tenson, Dec 10, 2008
    #42
    1. Advertisements

  3. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    OK will do Andy. I'm going to try applying DRC later today and then take more measurements. I'll redo the plots correctly since the ones above are wrong too - 44.1Khz(before measurements) isn't half of 96Khz(after) which I incorrectly assumed when chopping up the impulse yesterday. This will actually make things look a tad better than they already do for the after measurements.

    I'll do a couple of regular waterfalls for below 300hz. I assume you want a long window here to see what's happening over a longer period of time?
     
    ShinOBIWAN, Dec 11, 2008
    #43
  4. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    Simon,

    I checked the fabric and the panels yesterday and found nothing flapping unduly. I also ran a 70hz sine right were the resonance seems worse and the panels were just barely moving compared to their resonance frequency. I don't think the fabric nor panels are causing the problem, there's no energy coming off them from what I can see, although I'm no expert. How can I (non destructively) test more thoroughly than just checking for vibration or is that enough.

    Its looking more like the cabinet at the front IMO. I might try dialling in the target curve to be applied with the DRC and start rolling the bass off early and shallow since there's a fair amount of gain going on down low. A lowpass with a Q of 0.5 at a suitable point might provide a bandaid.
     
    ShinOBIWAN, Dec 11, 2008
    #44
  5. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    What's the room dimensions? Can't recall if you mentioned them already.

    I am guessing that the room is 2.4m high and about 2.5m wide. There's your potential 70Hz problem. Can you do a few waterfall plots at half way down the room near your listening position, 300mm from each side wall and one directly in the middle.

    Can you do the waterfalls at 150Hz and below...at the moment I can't make out any of the axis on your current plots.

    I think those membrane absorbers need to be tuned lower??
     
    andyoz, Dec 11, 2008
    #45
  6. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    I just read that comment and I think what you're hearing is a room that's highly damped above say 100Hz but is a bit of a mess below 100Hz.

    The room modes were always there at 70Hz, they just sound worse now that the stuff higher up the frequency range is damped. Your ear is picking out the room modes more effectively as they are not masked by the energy in adjacent frequency ranges.

    It's a common problem is small voice recording booths for example. We see alot of studios where they take a bare 3 x 3m room, apply loads of 100mm foam to the walls and don't install any LF treatments. When recording male voice, these small rooms tend to "boom" around 70-80Hz and they have to EQ the hell out of the recording to get anything useful. The whole response of the room is cocked-up. The room modes were always there, they have just made them more apparent.
     
    andyoz, Dec 11, 2008
    #46
  7. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    Not got as much time as I'd like to dedicate to this right now but I did look a little further into it.

    For now I've come up with a bandaid just to make the system listenable, actually its much better but I'm still aware of a problem in the bass. What I've done is put a high Q and deep notch filter in at 70hz along with digital room correction. Oddly despite the deep notch showing as a severe drop in the frequency response its not that noticeable because its filled in by the resonance. Hardly fidelity though since your listening to controlled boom at 70hz.

    Before treatment
    [​IMG]

    [​IMG]


    After and with notch at 70hz + DRC
    [​IMG]

    [​IMG]

    I'm now 99% sure its the cabinet at the front of the room and loudspeaker positioning in relation to it. I dropped the cabinet face down onto the floor then ran a tone generator with sine through the right speaker as before and the weird phasey boom around 70hz disappeared.

    That's a bitch because the cabinet absolutely has to stay put. I'm now looking at ways to lessen the problem and here's what I've come up with:

    - Perhaps as Simon pointed out, the problem will be smoothed once the left channel is playing thus placing another low frequency source within the room. Usually such a move would smooth bass problems. But I'm not at all confident this will be true since the left channel is in a mirror image situation with the right so it makes sense that the same problems as already noted will be present.

    - Its been suggested to tune the existing panel traps lower by subbing the 4mm ply with 6mm or 8mm. I'd rather not go this route unless absolutely necessary since it will mean destroying a lot of hard work. Consider this a final option.

    - I was wondering about membrane traps located in the alcoves where the loudspeaker sit? Each alcove could accommodate 1x 60cm x 170cm panel and 2x 40cm x 170cm panels. All with a max total depth of 4". Can these be tuned low enough to affect 50-100hz with the concentration around 70hz?

    - What about multiple small subs? I could very easily and quickly build 3 small sealed cubes based around the unused Peerless XLS10 drivers I have already. I've read the Harman paper about multiple subs for smoothing room response and one at the mid point for side walls and rear wall looks promising. These would work from around 80-90hz.

    Any thoughts on these idea's?

    I like the idea of trying the membranes around the speakers first 'cause its cheap and no problems building them. But will they work and how would I design them?
     
    ShinOBIWAN, Dec 14, 2008
    #47
  8. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    Shin,

    Can you redo those waterfalls but show 500ms and about 50dB drop in level (you currently show about 30dB drop)

    Also, can you play around with the settings to show "time slices" and turn off all the color fill-in. Here's an example of the sort of outputs we look at when hunting for room modes...

    [​IMG]
     
    Last edited by a moderator: Dec 14, 2008
    andyoz, Dec 14, 2008
    #48
  9. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    Andy the measurement with DRC, the notch filter aren't much good since they hide much of the problem.

    Instead here is a measurement used for time alignment of the drivers and has deliberate delays applied to each driver in order to separate the peaks of their impulses and allow calculation of the required delays. This is why you see some destructive interference in the upper range. Below 100hz is fine though.

    Another thing to note is that the large resonance is appearing at 42hz and the problems between 50-85hz don't appears to be a resonance at all. They actually look like pot holes in the graph and occur within a regular interval. This may be the phasey nature I was talking about. By far the 70hz problem is the most annoying, the one at 42hz is virtually disappears after DRC.

    Here's the plot. I cannot get time slices with this software since it doesn't appear to work that way, FFT uses time slices whereas this uses wavelets in stockwell transform.

    BTW I lengthened to 500ms and -50dB range. All else is same as above.

    [​IMG]
     
    ShinOBIWAN, Dec 14, 2008
    #49
  10. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    That's easier to "read".

    Still not entirely sure what's going on around 70Hz though TBH. If the room is about 2.5m wide and 2.5m high then you could be seeing the two room modes interacting with one another. Although your test with the cabinet door off seems to suggest otherwise. Where was that measurement taken? Is it in the middle of the room about 1.2m above the floor?

    I assume the 42Hz mode is linked to the rooms long dimension. IS the room 4m long.
     
    andyoz, Dec 14, 2008
    #50
  11. ShinOBIWAN

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    Membrane traps are very possible Ant. I'll pm you some suggestions on construction when I'm not so tired.

    Regarding the cabinet, could you try filling it with a big duvet or two or lots of clothes? Damp it, basically.
     
    Tenson, Dec 15, 2008
    #51
  12. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    The cabinet may not be having much influence. I didn't measure and simply listened after I'd laid it down on the floor. I also think I had the band aid crossover in place which greatly reduces the effect, I should have tested with measurements and an unfiltered crossover because this is where it really shows.

    I'll lay the cabinet down again and take measurements this time. Assuming it isn't the cabinet and is the room, what can be done?

    The measurement was taken at about 1m height(head height sitting on couch) and 1m from rear wall/3m from front wall. Distance from right was 1.15cm and left was 1.5m.

    The room is 2.28m wide until about 1/3rd from the front wall at which point it widens to 2.65cm. The height is 2.3m and the length is 4.1m.
     
    ShinOBIWAN, Dec 15, 2008
    #52
  13. ShinOBIWAN

    ShinOBIWAN

    Joined:
    Jun 7, 2006
    Messages:
    710
    Likes Received:
    1
    Thanks Simon.

    Good idea, there's nothing in the cabinet at the moment so I'll try filling it with the slabs and offcuts of fibreglass I have left over.
     
    ShinOBIWAN, Dec 15, 2008
    #53
  14. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    I'm sure the 70Hz thing is a room mode.

    Can you take a measurement half way down the room, at 1.15m above floor level and 1.3m from the side walls. The modes should be at a SPL minima there.

    Then repeat the measurement at 0.5m above floor and 0.5m from one of the side walls. The mode "should" be stronger there. :)
     
    andyoz, Dec 15, 2008
    #54
  15. ShinOBIWAN

    andyoz

    Joined:
    Sep 18, 2004
    Messages:
    1,117
    Likes Received:
    0
    15mm MDF over 4" cavity with 80mm mineral in cavity will give you approx. 70Hz. The mineral wool must be spaced 10-20mm from the rear of the MDF panel.

    You need to do measurements without the panels at several fixed locations and introduce the panels, repeat the measurements to make sure they are working.

    Once you know what the 15mm MDF panels are doing, I would inroduce a few slighTLY "detuend" memebranes to give say +/- 10Hz awaf from the main panels. Use slightly thinner or thicker MDF (12mm or 18mm) or play around with the mineral wool packing or cavity depth.
     
    andyoz, Dec 15, 2008
    #55
  16. ShinOBIWAN

    scott_01

    Joined:
    Sep 1, 2005
    Messages:
    433
    Likes Received:
    0
    ShinOBIWAN

    Any developments on this front? Have you managed to solve your problems?
     
    scott_01, May 6, 2009
    #56
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.