INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: How to create a +6b/oct dipole rolloff compensation filter  (Read 3439 times)

ths61

  • World Citizen
  • ***
  • Posts: 161
How to create a +6b/oct dipole rolloff compensation filter
« on: March 06, 2013, 01:01:05 am »

I am playing with a dipole subwoofer and would like to employ a dipole rolloff compensation filter.

Does anyone know how to enter/create a "+6b/oct dipole rolloff compensation filter" in the MC 18 (e.g. 100Hz down to 20Hz)?

I tried adding a "low shelf filter" in the Parametric Equalizer, but that is a flat boost across a specific range.

Do I have to add a series of "equalizer filters"?  If so how (which is first, are they additive, etc.)?

100Hz - 0dB - bandwidth(Q) ?
50Hz - 6dB - bandwidth(Q) ?
25Hz - 12db - bandwidth(Q) ?

TIA
Logged
Main - JRMC31 -> custom ALSA_cdsp -> CamillaDSP(2x8 channel 64-bit FIR convolution) -> 8 channel DAC
Office - JRMC31 -> Asus Xonar Essence STX -> W4S STI-1000 -> Mini-Magnepans
Shop - JRMC31 -> W4S MicroDAC -> Adcom GFA-2535 -> B&W Rock Solid

Nyal Mellor

  • Recent member
  • *
  • Posts: 34
Re: How to create a +6b/oct dipole rolloff compensation fitler
« Reply #1 on: March 06, 2013, 09:57:20 am »

It seems there are two approaches to implementing shelving filters, one is to use Q, gain and frequency the other slope, gain and frequency. I do not believe the two are interchangeable, nor is there any mathematical conversion one can do between the two.
Logged

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5181
  • "Linux Merit Badge" Recipient
Re: How to create a +6b/oct dipole rolloff compensation fitler
« Reply #2 on: March 06, 2013, 10:17:24 am »

I am playing with a dipole subwoofer and would like to employ a dipole rolloff compensation filter.

Does anyone know how to enter/create a "+6b/oct dipole rolloff compensation filter" in the MC 18 (e.g. 100Hz down to 20Hz)?

I tried adding a "low shelf filter" in the Parametric Equalizer, but that is a flat boost across a specific range.

Do I have to add a series of "equalizer filters"?  If so how (which is first, are they additive, etc.)?

100Hz - 0dB - bandwidth(Q) ?
50Hz - 6dB - bandwidth(Q) ?
25Hz - 12db - bandwidth(Q) ?

TIA

The shelf isn't just a flat boost above or below a specified range, the Q lets you specify how quickly or slowly the boost is applied from the "shoulder" frequency.  For example, I currently have baffle step compensation dialed in to provide 4 dB of boost over two octaves. My low shelf frequency is 200Hz and my Q is .66 (two octaves).  The boost begins getting applied slightly above 200Hz, and the signal isn't the full 4dB up until a little below 50 Hz.  Below 50 Hz it levels out.  So I, in effect, have a 2dB per octave boost between 200Hz and 50 Hz.

You could easily dial in a shelf with a 12db gain and a two octave Q (.66) starting at 100 Hz.  That would get you close to what you want (12 dB up at 25 Hz, 6 dB up at 50 Hz), but you'd need to go farther if you wanted subsonic adjustment (18 dB gain with a three octave Q, etc.).

I'd recommend that you dial in the 12 dB filter I mentioned and then play the full-band pink noise calibration clip while looking at the DSP analyzer.  You can then see if the boost starts in the right place, is high enough, etc.  The analyzer graph will let you see if the boost is being applied the way you want it to be, and then you can paddle around until it looks (and measures) correctly.

If you need a non-linear boost, you could probably also use the linkwitz transform DSP module, but the shelf is a little easier (for me) for a linear boost.
Logged

ths61

  • World Citizen
  • ***
  • Posts: 161
Re: How to create a +6b/oct dipole rolloff compensation filter
« Reply #3 on: March 06, 2013, 11:16:33 am »


You could easily dial in a shelf with a 12db gain and a two octave Q (.66) starting at 100 Hz.  That would get you close to what you want (12 dB up at 25 Hz, 6 dB up at 50 Hz), but you'd need to go farther if you wanted subsonic adjustment (18 dB gain with a three octave Q, etc.).

I'd recommend that you dial in the 12 dB filter I mentioned and then play the full-band pink noise calibration clip while looking at the DSP analyzer.  You can then see if the boost starts in the right place, is high enough, etc.  The analyzer graph will let you see if the boost is being applied the way you want it to be, and then you can paddle around until it looks (and measures) correctly.

If you need a non-linear boost, you could probably also use the linkwitz transform DSP module, but the shelf is a little easier (for me) for a linear boost.

Thanks for the quick response.  It looks like if I start at 100Hz (0.66 and 12 dB) with a "Low-shelf", it is raising 200 Hz and slightly at 300Hz in the Analyzer.  To get the graph to shift to the left, I need to start the "Low-shelf" filter at 25Hz so it levels off at 100Hz.  If I do this, the amplitude of the graph drops significantly.  

Creating an "Equalizer filter" with (12Hz with 0.4 (3 octaves) and 18dB) appears to look the best so far.

Am I doing something wrong?

Logged
Main - JRMC31 -> custom ALSA_cdsp -> CamillaDSP(2x8 channel 64-bit FIR convolution) -> 8 channel DAC
Office - JRMC31 -> Asus Xonar Essence STX -> W4S STI-1000 -> Mini-Magnepans
Shop - JRMC31 -> W4S MicroDAC -> Adcom GFA-2535 -> B&W Rock Solid

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5181
  • "Linux Merit Badge" Recipient
Re: How to create a +6b/oct dipole rolloff compensation fitler
« Reply #4 on: March 06, 2013, 11:31:45 am »

Thanks for the quick response.  It looks like if I start at 100Hz (0.66 and 12 dB) with a "Low-shelf", it is raising 200 Hz and slightly at 300Hz in the Analyzer.  To get the graph to shift to the left, I need to start the "Low-shelf" filter at 25Hz so it levels off at 100Hz.  If I do this, the amplitude of the graph drops significantly.

Am I doing something wrong?



To make sure that I understand what's happening, you currently have the shelf shoulder at 100Hz with a .66 Q and a +12 dB gain.  You're seeing a slight rise at 300 and additional rise at 200; that's normal.  But are you seeing the expected 12 dB rise below 100Hz?  Or are you only seeing gain above 100Hz (which would be odd)?

You're going to get some gain before your shoulder point; it's kind of like a high or low pass filter that way (i.e. in a butterworth low or high pass filter the signal is already down 3 dB at the shoulder).  And depending on what your dipoles are doing at 100Hz  that "above the shoulder" gain might actually be desirable.  For example, if 100Hz is your 3dB down point, you might want some extra gain above the shoulder.  If 100 Hz is right where the speaker starts to roll down from a flat response, you wouldn't want that "above the shoulder" gain.

That's why I suggested you paddle around a bit.  If you want to move the curve to the left, try lowering the shoulder frequency 5 or 10 Hz at a time while watching the analyzer.  Once you've got the slope how you want it, make sure it's still 12 dB up where you want it to be.  Try widening or narrowing the Q a little at a time, or boosting the gain a few dB.  Just fiddle incrementally (with the pink noise running) until it looks how you want it.

Then (if you have a mic and an RTA analyzer) I'd recommend measuring the speaker's actual output to make sure it's actually behaving as expected.
Logged

ths61

  • World Citizen
  • ***
  • Posts: 161
Re: How to create a +6b/oct dipole rolloff compensation filter
« Reply #5 on: March 06, 2013, 11:42:30 am »

To make sure that I understand what's happening, you currently have the shelf shoulder at 100Hz with a .66 Q and a +12 dB gain.  You're seeing a slight rise at 300 and additional rise at 200; that's normal.  But are you seeing the expected 12 dB rise below 100Hz?  Or are you only seeing gain above 100Hz (which would be odd)?

You're going to get some gain before your shoulder point; it's kind of like a high or low pass filter that way (i.e. in a butterworth low or high pass filter the signal is already down 3 dB at the shoulder).  And depending on what your dipoles are doing at 100Hz  that "above the shoulder" gain might actually be desirable.  For example, if 100Hz is your 3dB down point, you might want some extra gain above the shoulder.  If 100 Hz is right where the speaker starts to roll down from a flat response, you wouldn't want that "above the shoulder" gain.

That's why I suggested you paddle around a bit.  If you want to move the curve to the left, try lowering the shoulder frequency 5 or 10 Hz at a time while watching the analyzer.  Once you've got the slope how you want it, make sure it's still 12 dB up where you want it to be.  Try widening or narrowing the Q a little at a time, or boosting the gain a few dB.  Just fiddle incrementally (with the pink noise running) until it looks how you want it.

Then (if you have a mic and an RTA analyzer) I'd recommend measuring the speaker's actual output to make sure it's actually behaving as expected.

I see the proper slope in the Analyzer, just that it extends above 100Hz.  Using an "Equalizer filter" (12Hz, 0.4, 18dB) yields the better graph [3 octaves = 12Hz, 24Hz, 48Hz, 96Hz].

I am experimenting with the filters on my laptop now, not hooked up to the real system.  I will try it on the real system tonight.  I want to get the filters figured out before I start experimenting with cabinet design mules.

Logged
Main - JRMC31 -> custom ALSA_cdsp -> CamillaDSP(2x8 channel 64-bit FIR convolution) -> 8 channel DAC
Office - JRMC31 -> Asus Xonar Essence STX -> W4S STI-1000 -> Mini-Magnepans
Shop - JRMC31 -> W4S MicroDAC -> Adcom GFA-2535 -> B&W Rock Solid

mwillems

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 5181
  • "Linux Merit Badge" Recipient
Re: How to create a +6b/oct dipole rolloff compensation fitler
« Reply #6 on: March 06, 2013, 11:56:02 am »

I see the proper slope in the Analyzer, just that it extends above 100Hz.  Using an "Equalizer filter" (12Hz, 0.4, 18dB) yields the better graph [3 octaves = 12Hz, 24Hz, 48Hz, 96Hz].

I am experimenting with the filters on my laptop now, not hooked up to the real system.  I will try it on the real system tonight.  I want to get the filters figured out before I start experimenting with cabinet design mules.



Sounds like you've got it figured out; are you crossing the sub over at 100Hz?  Or higher?  I only ask because you may also need to account for the frequency response dip from whatever kind of low pass you're using on the sub at the top end (especially if it's at or near 100Hz), but that should all show up in the analyzer if you're using a DSP low-pass. 

Regardless, glad you sorted it out and happy listening!
Logged

ths61

  • World Citizen
  • ***
  • Posts: 161
Re: How to create a +6b/oct dipole rolloff compensation filter
« Reply #7 on: March 06, 2013, 12:13:44 pm »

Sounds like you've got it figured out; are you crossing the sub over at 100Hz?  Or higher?  I only ask because you may also need to account for the frequency response dip from whatever kind of low pass you're using on the sub at the top end (especially if it's at or near 100Hz), but that should all show up in the analyzer if you're using a DSP low-pass.  

Regardless, glad you sorted it out and happy listening!

I am considering 2 rooms.  For the office/desktop system, it will be around 100Hz.  For the 2-channel room, it will be much lower.  This is my first DIY OB/Dipole sub so it is experimentation at this point until it is determined it is suitable for both, either or neither environment.  The Dipole subs will be integrated with Dipole panels.
Logged
Main - JRMC31 -> custom ALSA_cdsp -> CamillaDSP(2x8 channel 64-bit FIR convolution) -> 8 channel DAC
Office - JRMC31 -> Asus Xonar Essence STX -> W4S STI-1000 -> Mini-Magnepans
Shop - JRMC31 -> W4S MicroDAC -> Adcom GFA-2535 -> B&W Rock Solid
Pages: [1]   Go Up