Attn EQ Experts - Am I over-EQing?

Discussion in 'Speakers & Subwoofers' started by JasonCI, Feb 14, 2005.

  1. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0
    Hello,

    I'm trying to flatten out the frequency response curve of my PB12 using the BFD DSP1124 PEQ.

    My FR curve*

    [​IMG]

    Given the volatile nature of my FR I would assume that at least a half dozen filters are required. However, I could also see using as many as 11 of the 12 filters to create something that resembles a flat curve.

    Freq(Hz) / Gain(dB) / Bandwidth (x/60th octave)
    20 / -07 / 16
    24 / -08 / 12
    30 / -05 / 35
    40 / +12 / 18
    43 / -10 / 13
    56 / -10 / 22
    63 / +05 / 06
    68 / -02 / 09
    78 / +15 / 12
    87 / -05 / 11
    98 / +06 / 13

    Given my FR curve, is 11 filters over kill?

    Thanks all!

    Note - I've determined that 100 Hz and 32 Hz sound most similar when the 100 Hz tone is played at -12 dB. I will eventually add a shelving filter to the mix to create the 12 dB gradual decline ([​IMG] Thanks Wayne! [​IMG]).


    *Configuration Details
    • My HT room is approx 2800 ft^3 with a small open area to the kitchen and a hallway to the stairs at the rear of the room.
    • AVR was configured with 80 Hz Crossover w/ all speakers set to small.
    • Only front right, front left and subwoofer channels were used.
    • Phase at 180 degrees. There's a significant decrease in SPL between around and above 80 Hz otherwise.
    • SPL readings were taken via Radio Shack Analog SPLM and compensated for error accordingly.
    • 10 second sine waves were played at 1/12 octave intervals.
     
  2. Mark Seaton

    Mark Seaton Supporting Actor

    Joined:
    Oct 10, 1999
    Messages:
    599
    Likes Received:
    0
    Trophy Points:
    110
    Real Name:
    Mark Seaton


    That's a fairly extreme contour in my oppinion, where a 4-8dB shelf is usually plenty if you set the upper end of the subwoofer somewhat hot relative to the mains and can get a smooth transition. Again, I would maybe try another location or two to see if you can get around the pair of recessions in the response.

    Just my take on the matter,
     
  3. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0
    Thanks Mark,



    I ran home at lunch to run through my frequency response with my sub at a different location. The compensated values are shown in the chart below. I was able to get rid of the 80 Hz null. The 40 Hz null is a bit of an enigma. There is a significant drop in SPL from 39 Hz to 40 Hz and then significant increase in SPL from 40 to 41 Hz.

    Should I try a narrow bandwidth boost at 40 Hz?

    [​IMG]
     
  4. Mark Seaton

    Mark Seaton Supporting Actor

    Joined:
    Oct 10, 1999
    Messages:
    599
    Likes Received:
    0
    Trophy Points:
    110
    Real Name:
    Mark Seaton
    Stick with the first location and try what I had suggested. Don't get overly concerned about a very narrow notch.
     
  5. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,184
    Likes Received:
    51
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Jason,

    As Mark noted, that is quite a few filters. Part of the difficulty in learning how to use a parametric is determining the correct bandwidth your filters need to be. Too often people try to use lots of super-narrow filters where a few broader ones will do the trick.

    Here’s a link to some information you might find helpful. Look for my third post on this page, and reference it to the charts that are there (some of the charts will be on the thread’s previous page).

    http://hometheatertalk.com/httalk/vi...2ad9e247565d3a

    Regards,
    Wayne A. Pflughaupt
     
  6. Jon W.

    Jon W. Stunt Coordinator

    Joined:
    Jul 31, 2004
    Messages:
    121
    Likes Received:
    0
    Trophy Points:
    0
    I started out with 12 filters and kept working with it and now I have 3 filters for my dual 15's and 1 filter for my dual 10's. I found it to be alot easier to set a filter while measuring. Take your 40 hz null for instance. Set a 40 hz filter with a low bandwidth and a low gain and play the tones from say 30 hz to 50 hz and see what effect it had. Keep bumping up the gain and bandwidth till you get it where you want without effecting the frequencies you don't want. Baby steps.
     
  7. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0


    Hi Mark,

    My interpretation of what you're suggesting here is that I Turn off my sub and measure the SPL of my mains in the 70-120 Hz range. Then, turn off my mains, play the same frequencies though my sub and increase the gain until its SPL matches (or is slightly hotter than) that of my main speakers.

    Am I reading you correctly? Is this preferred over pink noise level matching?

    BTW, I'm running M&K LCR 750 THX Select speakers with a FR of 80 Hz - 20 KHz ± 2 dB.

    Thanks again!
     
  8. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0

    Good info Wayne, thanks.

    I think I'm getting the idea. Consider my first FR plot above. Even though my peaks and nulls aren't truly symmetric I can see how I could clean up most of them with only these 4 filters:

    Center Freq (Hz) / Gain (dB) / Bandwidth (octaves)
    23.0 / -11 / 0.50
    39.0 / +04 / 0.10
    52.6 / -12 / 0.30
    77.8 / +12 / 0.18

    Right?

    peaks and nulls in FR with 4 filters. Much of the f you consider my first FR plot I believe take my first FR plot
     
  9. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,184
    Likes Received:
    51
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne
    Yeah, that’s more like it. [​IMG] Asymmetrical problems aren’t that big a deal, really – it just takes an extra filter or two to “clean up” what the first couldn’t do.

    Regards,
    Wayne A. Pflughaupt
     
  10. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0
    I think I'm getting the hang of this. Four filters did a majority of the flattening and another three filters cleaned up what was left. Check out my before and after frequency response:

    [​IMG]

    Short of applying a shelving filter for a house curve I think this is as good as it's going to get. I tried applying additional narrow-bandwidth filters to the remaining peaks but they only had adverse effects.

    Here's the filters I used to get the final result:

    [​IMG]

    I have to thank everyone here at HTF (especially Wayne P., Sonny, Mark S., and Bruce K.) for their help. And thanks to Anthony Gomez and company for creating the PEQ spreadsheet for the BFD. It really helped me through the process. One word of caution though, if you decide to use Anthony's spreadsheet, and you should if you're using the BFD, you'll get frustrated if you are overly concerned with getting your FR "ruler" flat. Less really is more in this case![​IMG]

    Now, what should I watch first?
     
  11. EdwinK

    EdwinK Agent

    Joined:
    Oct 29, 2004
    Messages:
    38
    Likes Received:
    0
    Trophy Points:
    0
    I would still not recommend using > +6dB gain in any frequency range.
    ---
    My current settings were done through playing 20 or so "Warbles", using my DEQ2496 as an SPL-meter. In stead of needing 7+ 'sophisticated' filters using SpectraRTA, I ended up using only 4 (easy) ones. The result is surprisingly good.

    ps: A Warble is a frequency sweep over a tiny interval, making it easy to measure an average SPL for that frequency range.
     
  12. John-Tompkins

    John-Tompkins Second Unit

    Joined:
    Jan 21, 2003
    Messages:
    326
    Likes Received:
    0
    Trophy Points:
    0


    If your amp and woofer dont have a problem with a higher setting then +6 //then is it really a problem at all ? ..I have a freq set at + 8 but my amp doesnt clip, so I dont really see it as a big deal..maybe Im wrong.

    Plus, His question about using more filters still really hasnt been answered..

    Sure, if you can get it done with less filters then more power to you ..BUT is there any technical reason why using more filters would effect the sound negatively versues using less filters ??

    I had my house curve pretty much the way I wanted it with 5~6 filters ..but decided that I wanted to try and get the house curve to where it was perfect (on paper anyway) ..so by adding another 5~6 filters and messing around with it I was able to get a perfect graph

    If the graph is perfect...does it really truely matter how many filters were used to achieve it ?
     
  13. JasonCI

    JasonCI Stunt Coordinator

    Joined:
    Dec 29, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    0


    In retrospect, I now understand that 11 filters was overkill to adjust my FR. I ended up with 7 filters but could have gotten away with only the 4 centered at 23, 39, 49, and 79 Hz. When playing frequency sweeps I don't really notice much of a difference between using those 4 and the 7 filters I'm using now. I just wanted to further refine the area in the crossover range to create a more seamless transition.

    BTW, I did find that any more than 7 filters created more volatility.

    So I second John's question. From a purist's point of view I can see it is desireable to use as few filters as possible. But will adding a few more narrow-bandwidth filters to get more of 'ruler' flat response degrade the resulting sound?

    I know a lot of boosting negatively impacts headroom but what about a few extra narrow-bandwidth cuts?
     
  14. Mark Seaton

    Mark Seaton Supporting Actor

    Joined:
    Oct 10, 1999
    Messages:
    599
    Likes Received:
    0
    Trophy Points:
    110
    Real Name:
    Mark Seaton
    A good check for the narrow band EQ questions is to quickly check a few other locations of interest in the room. Make sure the peak or dip is somewhat common between seats. There are cases where the added energy in the room and at other seating locations can make for more problems than a narrow boost solves. Similarly, you may find that a dip at one seat is a peak at another, where you generally are best off splitting the difference and not going overboard. If anything, try cutting that frequency with a very narrow band filter, not boosting, as it is obviously causing some odd interactions in the room to result in a narrow notch in the first place.

    Overal it looks like you have made some very good progress. How does it sound? With relatively flat response like that you should be able to set the subwoofer a bit hotter than the mains by 2-5dB depending on what happens in the overlap region with pleasing results. This will largely depend on your main speaker capabilites in your room and your desired listening levels.
     
  15. Wayne A. Pflughaupt

    Moderator

    Joined:
    Aug 5, 1999
    Messages:
    6,184
    Likes Received:
    51
    Trophy Points:
    9,110
    Location:
    Katy, TX
    Real Name:
    Wayne

    Sorry to let this linger – we’ve been busy moving the past few weeks and I kinda let this one fall by the wayside.

    This is a tough one to answer, “why not use a lot of filters.”

    For one thing, if you look at brucek’s excellent bench test of the BFD (see Sonnie parker’s BFD Comprehensive Set-Up Guide ) you’ll see that it has pretty severe phase issues when filters are engaged. That’s a pretty typical problem with cheap equalizers, but the BFD seems to be even worse than most in that regard.

    Naturally, the more filters you engage, the more phase issues you can have – phase issues that vary by frequency. That would be tough situation for a continuously variable phase control to address, or even one of the auto set-up features many receivers have these days that calculate and compensate for speaker distance settings (which is what phase problems amount to).

    Does this translate to anything audibly degrading? Well, I honestly can’t say as I’ve ever heard anyone complain about it. At the end of the day, any phase problems you might end up with will most likely be a small price to pay for such a significant improvement in overall performance.

    So in the end, it’s one of those situations where “correct” might not realistically translate to anything audible. That said, it just seems preferable to avoid the issue as much as possible by using as few filters as possible.

    Perhaps a poor analogy: Sure, it won’t immediately trash your engine if you use the cheap Wal-Mart brand motor oil in your car - might even work just fine for many years. But wouldn’t you just feel better using a brand-name oil?

    Regards,
    Wayne A. Pflughaupt
     
  16. WarnerL

    WarnerL Agent

    Joined:
    Aug 31, 2001
    Messages:
    34
    Likes Received:
    0
    Trophy Points:
    0
  17. David Bikeman

    David Bikeman Stunt Coordinator

    Joined:
    Aug 15, 2004
    Messages:
    164
    Likes Received:
    0
    Trophy Points:
    0
  18. WarnerL

    WarnerL Agent

    Joined:
    Aug 31, 2001
    Messages:
    34
    Likes Received:
    0
    Trophy Points:
    0
    I actually did that a couple of years ago so it is kind of hard to remember but I think maybe a couple of hours. After a few plots you quickly get the hang of it. What I did was after graphing the initial response on the Excel spreadsheet, I drew a flat ideal response line on the same graph (I clicked around and found the function to draw a line on the graph). I then started at the 20 Hz side and tweaked and adjusted filters/bandwidths one filter at a time graphing the new response each time to try and bring the response up/down to my ideal line on the graph. Each filter that I used, I tweaked the bandwidth so it would not affect or affect very little any adjacent response that I did not want to change. I moved from left to right bit by bit until the final result shown. Previously, when I was shooting for a house curve (there is a link within the previous linked page) I did the same thing, ie. drew a line on the graph as my ideal house curve response and then moved from left to right adjusting/tweaking filters to get the response to hit my ideal line.
     

Share This Page