Dual Tumult in room resp and BFD set-up

Discussion in 'Home Theater Projects' started by Rich X, Sep 14, 2003.

  1. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    I've been busy trying to set-up my BFD this weekend for my Tumults. I have two Tumults in 18" sealed enclosures with a QSC PLX3402 amp and LT.

    Measurements at the listening position using Stryke CD and Radio Shack SPL meter w/ corr. values are below. Seems like a very flat response other than a few nasty nulls. I've attempted some corrections with the BFD. Any additional advice would be appreciated. I have not attempted any real room corrections yet (Bass Traps, etc.)

    Freq RAW BFD
    10hz 76.0 76.0
    12.5 78.5 80.5
    15hz 82.9 82.9
    17.5hz 83.0 84.0
    20hz 84.0 85.0
    22hz 84.5 87.5
    25hz 73.0 83.0
    28hz 75.1 82.1
    32hz 74.0 83.0
    36hz 84.7 87.7
    40hz 86.5 88.5
    45hz 85.0 86.0
    50hz 84.5 84.5
    56hz 85.5 85.5
    63hz 72.5 73.5
    71hz 61.5 62.5
    80hz 75.5 76.0
    89hz 73.8 73.8
    100hz 70.0 71.0
    112hz 69.0 70.0
    126hz 73.5 74.5
     
  2. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,208
    Likes Received:
    57
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Rich,

    Based on these readings, it doesn’t look like things are going well here at all. You did a good job of raising the depressed area between 22 and 36Hz, but you also raised 36-56Hz area. This was not needed and indeed negated most of the ground gained by raising the depressed 22-36Hz area.

    Specifically, before equalizing there was a 12dB differential between the two areas. After equalizing the differential is 8dB - only 4dB improved. Thus equalized response is only marginally better than unequalized.

    Actually, unequalized response going in is pretty poor, which suggests to me that we can probably make some improvements by experimenting with placement.

    A few basics: With an exception being rectangular rooms, typically only one location will give the best unequalized response. Ideally, what you are looking for is the place where:
    • Average SPL remains the highest.
    • Response is the smoothest – the fewest peaks and dips.
    • The smallest differential between the lowest dip and highest peak.
    Of course, no position will deliver all three of these in spades, or else you wouldn’t need to equalize. But one location should deliver these goals, or at least two out of three, better than others.

    If you haven’t already tried it, a good corner location – one with the longest uninterrupted wall length on both directions - should be one of the locations you take measurements from. I haven’t come across too many rooms where another location gave the best starting point for equalization.

    If there is only one prime location in the room, then it stands to reason that if you have two subs, they should both be located there. Placing the second sub in an inferior location obviously means you have one sub performing more poorly than the other. In addition, it is extremely difficult to equalize subs that are in separate locations, delivering significantly different response curves. (Again, the exception here would be a rectangular room, were there is more than one ideal corner).

    With two subs it is often to use separate placement, asymmetrical placement, etc. With experimentation, many report that they can achieve smoother response this way, (although it hasn’t worked for me). But IMO this is only an exercise for installations where equalization is not used. I don’t expect an arrangement like this will often – if ever -outperform dual equalized subs that are both located at the room’s prime position.

    In addition, your reading should be taken with both the subs and mains on. This will allow equalization to address any phase problems introduced by the crossover.

    Since you’re using the BFD, there is no reason to use bass traps, too. The EQ is much more effective, IMO.

    So – why don’t you try finding a better location and post some new numbers for us. The BFD is capable of delivering much better response than we’re getting here.

    Regards,
    Wayne A. Pflughaupt
     
  3. Brian Bunge

    Brian Bunge Producer

    Joined:
    Sep 11, 2000
    Messages:
    3,716
    Likes Received:
    0
    Trophy Points:
    0
    Wayne,

    That's not going to happen. Rich has his Tumults inside a cabinet to the L/R sides of the screen. So playing with placement isn't an option.
     
  4. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Wayne,
    Thanks for your detailed feedback. As Brian points out, when it comes to placement I may be screwed, as the subs are in a custom cabinet. My only placement option may be to add a THIRD sub...

    But before I go there I'd like to make the best of what I have. As you point out, I was able to improve the null that I have somewhat, however in doing that I also increased the level at the 36-56hz area. I did not make any adjustments to that range--it just happened when I boosted the 22-36hz range. I even added a cut filter to the 36-50hz range but it had no effect.

    To be clear these are the BFD adjustments that I have made thus far. This is the first time I've ever experimented with an eq like this so I may be way off from what I really should do.

    Freq= 25 BW=6 Gain=8
    Freq= 30.9 BW=8 Gain=8
    Freq=40 BW=8 Gain=-6
    Freq=63 BW=10 Gain=12

    Also, I took these measurements without my mains. The reason I did it that way is that I am most interested in their perfomance for the LFE channel. My mains may really help as they are full range. For the music listening that I've done with my mains only I have been very pleased with the low end. I'll have to take some measurements with them engadged with the subs.

    Given the limitations that I have, what should I try next?

    Thanks

    Rich
     
  5. RichardHOS

    RichardHOS Second Unit

    Joined:
    Mar 11, 2003
    Messages:
    454
    Likes Received:
    0
    Trophy Points:
    0
    I'm not sure why you have a filter set at 30.9Hz and another at 40Hz (is that 30+9/60 ?), one with a gain +8, the other -6.

    Looks like the filter at 40Hz needs more negative gain, unless that is actually another close filter at 30.9 ??
     
  6. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,208
    Likes Received:
    57
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Rich,

    Try taking main/sub readings with the mains set to small, if you haven’t already. Often if there are too many low frequency sources it can mess up response. Not to mention, if the room tends to cause a certain frequency range to peak, it will probably be exhibited in both the mains and subs – and you can’t equalize the mains.

    Naturally, after everything is EQ’d properly, if you feel that music sounds better with just the mains running at full, you can always switch over.

    Regards,
    Wayne A. Pflughaupt
     
  7. RichardHOS

    RichardHOS Second Unit

    Joined:
    Mar 11, 2003
    Messages:
    454
    Likes Received:
    0
    Trophy Points:
    0
    Hmm, not sure I really understand that reasoning. I would certainly agree that to begin with it would be best to disconnect the mains and set the sub crossover as high as possible, thus getting a good picture of the room's resonances.

    But when setting filters, I would think you first would set filters just for the sub, and then adjust the filters after adding mains back to the mix. No, you can't EQ mains, but if the mains are producing a peak you can sure lower the sub output in that frequency range to compensate.
     
  8. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
     
  9. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,208
    Likes Received:
    57
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Rich,

     
  10. Dennis XYZ

    Dennis XYZ Stunt Coordinator

    Joined:
    Jun 1, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Boosting nulls usually doesn't work very well. You're just pumping power into an acoustic black hole. Have you tried the opposite approach -- knocking down the high spots to flatten the response? That's usually much more effective.

    Also, can you move the chair or at least try measuring from some different spots? That null at 71 Hz looks pretty grim. If you filled that in somehow with EQ for one seating position, the room power response would really get messed up.
     
  11. TimForman

    TimForman Supporting Actor

    Joined:
    Dec 4, 2002
    Messages:
    847
    Likes Received:
    0
    Trophy Points:
    0
     
  12. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Tim: Thanks for the dose of reality...

    I took some new measurements last night. I set the mains to small and the cross-over at 50hz. I used a different setting on the BFD so that there were no filters engaged.

    Things stay pretty much the same through the cross-over point. The null at 71hz is still there, but is not quite as bad. From that point up, everything is pretty smooth, though at a lower level than the lower frequencies--this could be due to a level disparity between my subs and mains. I calibrated the system using Avia, but I often run the subs a little higher b/c I have liked the way that sounds in other systems that I have set-up. I've never taken detailed measurements like this before so I really didn't notice the disparity.

    Freq SPL
    10 76.0
    12.5 82.5
    15 85.9
    17.5 86.0
    20 87.5
    22 85.5
    25 77.0
    28 78.1
    32 77.0
    36 86.7
    40 88.5
    45 86.0
    50 85.5
    56 87.5
    63 77.5
    71 68.5
    80 77.5
    89 75.8
    100 76.0
    112 75.0
    126 78.5
    142 79.0
    160 74.5

    As you can see, in the main area of concern, below the cross-over at 50hz, the null from 25-32hz is still there. Interestingly it is not quite as severe.

    Rich
     
  13. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Anyone have filter suggestions?
     
  14. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,208
    Likes Received:
    57
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Hi Rich,

    Often in these sub equalizing threads the question of boost vs. cut and headroom comes up. Personally, I feel it’s a virtual non-issue because any equalizing will place additional demands on both the amp and drivers. Therefore my rule-of-thumb is that you shouldn’t be equalizing if your sub is marginal to begin with. In your case, with dual 18” drivers and an amp putting out 700 watts or more per channel, I can’t imagine headroom being an issue.

    That said, let’s take a look at your curve. If you haven’t already, I suggest plotting your response on graph paper or an Excel-based program; otherwise it will be very difficult to follow what I’m saying.

    You’ve already indicated that the crossover frequency is 50Hz, so let’s focus on the region between 16Hz and 56Hz. As you can see, across that range response is fairly flat, except between 22-36Hz. Put another way, the depressed area is 2/3-octave wide, while the areas above and below the depression totals 1-1/6-octave - i.e., almost twice the bandwidth as the depressed area.

    Thus it simply makes no sense to cut the wider, high area down to the level of the much-narrower depressed area. The end result will be a net gain reduction of 10-12dB, and your sub level will to so low that you will have to increase amplifier gain by that amount. So you really haven’t “saved” any headroom at all.

    For more detail on the “nuts and bolts” of what I’m talking about here, you might want to look at my posts on this thread, especially the second one. The discussion there involved a situation exactly like the one you have.

    The question of boosting nulls also comes up frequently. While it’s true you can’t boost a null, not all underrepresented areas are true nulls. Depressed areas can also be the natural response characteristics of the sub/enclosure combination or the result of phase issues induced by the crossover. These can usually be addressed with equalization. Nulls are typically fairly narrow – 1/3-octave or less. That’s not what we have in your case, so I’ll be surprised if the depressed area won’t respond to equalization.

    What we need here is to boost the depressed area (obviously) and at the same time avoid what happened before – “dragging up” the adjacent areas that don’t need attention. The trick is to keep the filter’s bandwidth narrow enough that it only affects the target area – nothing more. Easier said than done – I’ve found that the BFD tends to affect a wider bandwidth than its settings would suggest, especially when the amount of gain applied to the filter is substantial – as it will be in your case.

    As noted, the depression we want to eliminate is 2/3-octave wide. On paper this would be addressed with a 1/3-octave filter. But since we’re going to be boosting the filter around 9dB or so, we can expect the actual affected bandwidth to be wider than 2/3-octave.

    So let’s start with a narrower, 1/6-octave filter and boost it 9dB. After applying the filter, take new readings at 22, 28 and 36Hz. Naturally 28Hz, which is the filter center, should read 9dB higher than it did. But the fringe frequencies 22Hz and 36Hz should read the same as they did before applying the filter.

    Now let’s look at a couple of other unusual situations your depression showed before equalizing.

    First, you probably noticed that the 28Hz center is not the lowest point of the depression; the adjacent frequencies, 25Hz and 32Hz, are both 1dB lower than 28Hz. What is going to happen after we apply the filter is that the differential between those three readings will increase – from a 1dB difference to probably 3-4dB – maybe more. This is a natural and unavoidable side effect of our filter, but we do want to minimize that situation.

    Second, note that the depression’s parameter frequencies - 22 and 36Hz - are both about 2dB below their outer adjacent frequencies, 20Hz and 40Hz. So it wouldn’t hurt to nudge 22 and 36Hz up a tad to get them closer to 20 and 40Hz.

    We can easily address both these situations by tweaking the bandwidth of our 28Hz-centered filter. All you need to do is adjust the bandwidth until 22Hz and 36Hz read 1dB higher than they did before applying the 9dB boost. This will simultaneously nudge up those two frequencies, while at the same time making sure that 25Hz and 32Hz have been boosted as much as this filter will allow.

    However, new readings may show the big problem area centered at 28Hz has now become two smaller problems centered at 25 and 32Hz. But they will be very narrow and most likely inaudible with program. Nevertheless, if you feel they need to be addressed you can apply a couple of very narrow filters. I’d start with 1/6-octave and squeeze them down until they do not change the readings of the adjacent frequencies. Make sense?

    The only thing left to deal with is the narrow 71Hz dip. I’m less confident we can address with this one, as it may be a true null. Not to mention, it is located in an area that the crossover is filtering downward.

    But it can’t hurt to try. It will require a filter narrower than 1/6-octave - 1/8 perhaps. A 8-9dB boost looks good on paper, but it may require more because of the effect of the crossover. Again, you want changes to the adjacent 63 and 80Hz readings to be minimal.

    If you find a substantial boost at 71Hz gets little or no change in reading at that frequency, then don’t use this filter. I’ll be the first to concur that “burning” headroom by boosting is a waste if it gets you no change in response.

    Hope this helps.

    Regards,
    Wayne A. Pflughaupt
     
  15. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Wayne,
    I did plot everything out on excel (I just didn't know how to post the graph), so I can follow what you described.

    I'll set up these filters and see how it goes. Thanks.

    Rich
     
  16. Brian Bunge

    Brian Bunge Producer

    Joined:
    Sep 11, 2000
    Messages:
    3,716
    Likes Received:
    0
    Trophy Points:
    0
    Rich,

    If you email me the Excel graphs I can host them on my website for you if you'd like.
     
  17. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    Thanks Brian,

    I'll e-mail you the graph once I have punched in the new filters and taken measurements. We're having company over tonight, but hopefully I'll be able to spend some time on it after they leave!

    Rich
     
  18. Rich X

    Rich X Stunt Coordinator

    Joined:
    Jun 18, 2002
    Messages:
    115
    Likes Received:
    0
    Trophy Points:
    0
    I finally had some time to play with a new filter scheme. I think I have gotten things pretty flat through the critical 20 - 50hz range. It took a good bit of trial and error and re-measurement.

    Here are the results (mains crossed over at 50hz):

    Freq SPL
    10 79.5
    12.5 80.5
    15 84.9
    17.5 85.0
    20 88.5
    22 88.5
    25 87.0
    28 87.1
    32 87.0
    36 88.7
    40 89.5
    45 87.0
    50 86.5
    56 88.5
    63 77.5
    71 68.5
    80 78.5
    89 77.8
    100 76.0
    112 74.0
    126 79.5
    142 79.0
    160 76.5

    Other than the null at 71hz, the difference in level above 56hz is probably due to the level difference I have set between my mains and subs (I usually run the subs a little higher)

    Here are the filters I ended up with:

    28.3Hz BW=10 Gain=9dB
    25Hz BW=2 Gain=7dB
    32Hz BW=2 Gain=5dB
    71Hz BW=8 Gain=6dB

    I may actually remove the filter at 71Hz b/c it does not seem to be doing anything. I also thought about adding a few additional tweeks to try to flatten the a few of the little humps, but they are small and probably within the error tolerance of my measurement device anyway, so that would probably be overkill.

    Brian: I'm going to e-mail my spreadsheet to you. If you can post the chart that would be great.

    Thanks to Wayne and others for your help. Let me know if you have any additional suggestions.

    Rich
     

Share This Page