Both Q and BW on generic equalizer panel?

Greg Dunn

Member
Thread Starter
Joined
Oct 4, 2017
Messages
107
Hi John,

I am using a "generic EQ" (actually Apple's AUNBandEQ) to set up my system - unfortunately, it wants the Q factor translated into the octave bandwidth factor BW before you enter the filter data. REW's Generic EQ of course calculates the standard Q values. It's easy enough to translate these values, but for multiple filter sets and transferring data over from REW each time it can become tedious. I wish AUNBandEQ allowed filter data to be imported, but alas.

I have tried using the FBQ2496 and XP2040 templates because they accept the BW value, but unfortunately, these EQ models don't support shelf, LP and HP filters among others. Would it be possible to add the BW value to the Generic EQ data fields so that you could either see it displayed or possibly enter it instead of the Q value? I'd even be willing to donate a third time if I could get something along the lines of this added to the EQ panel, to save me a few calculations. ;)
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
The Generic equaliser uses a Q (and hence bandwidth) definition based on the mid-gain points on the response, equalisers using octave bandwidth (like the XP2040) tend to use a bandwidth based on the nominal -3 dB points. Do the AUNBandEQ filter shapes match the XP2040 shapes for the same bandwidth settings?

To add an equaliser setting it would also be necessary to know the ranges and resolutions of the frequency, gain and bandwidth settings, the list of filter types supported, and how shelf filters were specified - their gain range and resolution and whether they have a slope or Q or bandwidth parameter, for example, and if so over what range. It would also be necessary to know the frequency at which the equaliser operated for accurate emulation at higher frequencies, but a system-wide equaliser is likely to operate at the current system sample rate rather than a fixed frequency so the emulation would only be accurate at one chosen sample rate.
 

Greg Dunn

Member
Thread Starter
Joined
Oct 4, 2017
Messages
107
Thanks. I do not have an XP2040; from examining the EQ graph, the individual PEQs look similar but the gains end up being notably lower with the XP2040 to get the same shape in the resultant curve as with the generic - so there must be some differences in the effective Q values. I attached the resultant curves.

What I can say with confidence, though, is that as a test, I used the calculator at


and entered the converted Q->BW values from the "generic" EQ into AUNBandEQ and the result was spot on. So at least for that path, the simple calculation worked just fine; my measured response after setting up AUNBandEQ was quite close to the target curve and sounds fine. I had a fairly wide variation of gains (-12 to +10) and Q values (~1 to 10) in my EQ set, too, and they summed together nicely in the equalizer plugin - so the calculation must have been decent.

Edit: sorry, I neglected to put the AUNBandEQ plot in originally. It's hard to normalize the graph so it is probably not as helpful as it could be.

38771


38770


38772
 
Last edited:

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
To add an equaliser setting it would also be necessary to know the ranges and resolutions of the frequency, gain and bandwidth settings, the list of filter types supported, and how shelf filters were specified - their gain range and resolution and whether they have a slope or Q or bandwidth parameter, for example, and if so over what range. It would also be necessary to know the frequency at which the equaliser operated for accurate emulation at higher frequencies, but a system-wide equaliser is likely to operate at the current system sample rate rather than a fixed frequency so the emulation would only be accurate at one chosen sample rate.
 

Greg Dunn

Member
Thread Starter
Joined
Oct 4, 2017
Messages
107
As far as I can tell, AUNBandEQ handles gains of -96 to +24 dB (1 decimal precision), BW < 5 (2 decimal precision), and frequency up to half the sample rate (sample rate appears to support 44.1, 48, 88.2 and 96) at 3 decimals of precision. It will handle up to 16 bands.

The filter types are:
Parametric (freq, BW, gain)
Butterworth LP (freq)
Butterworth HP (freq)
Resonant LP (freq, BW)
Resonant HP (freq, BW)
Bandpass (freq, BW)
Bandstop (freq, BW)
Low shelf (freq, gain)
High shelf (freq, gain)
Resonant Low shelf (freq, BW, gain)
Resonant High shelf (freq, BW, gain)

Let me know if you need any further details about its operation!
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
What is the minimum BW figure?

Are you sure the resonant LP/HP/shelf filters use a bandwidth figure in the same way as the parametric filters? How do the shapes compare to the Generic LPQ/HPQ/LSQ/HSQ filter shapes? For example, an LPQ filter with a Q of 1.925 will have a 6 dB peak before it rolls off, what is the N-band EQ BW figure for the same peak for a resonant LP? Similarly an LSQ filter with a Q of 2.055 will have 3 dB of overshoot and undershoot, what is the N-band EQ BW figure for the same overshoot/undershoot for a resonant shelf?
 

Greg Dunn

Member
Thread Starter
Joined
Oct 4, 2017
Messages
107
OK, hopefully I've answered the questions you asked in a comprehensible fashion.

The minimum BW (at least that the AU will accept and calculate a filter) is 0, so 0 < BW < 5.

A resonant LP with a 6 dB peak has a BW value of 0.707.

A resonant low shelf with a gain of 3dB and 3 dB overshoot has a BW value of 0.45. With 0 gain there is no overshoot, and it varies with different gain levels.

Screen caps:

38781


38783
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
One more question, at what BW figure does a 6 dB gain resonant low shelf shape match a 6 dB gain low shelf without a BW setting?
 

Greg Dunn

Member
Thread Starter
Joined
Oct 4, 2017
Messages
107
It looks like the resonant low shelf matches a standard 6 dB low shelf when you set BW to 1.9. I would have expected 2.0 based on experience, but 1.9 is definitely closer.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
OK, thanks. That matches REW's generic LS/HS option. I sent you a message with a link to a test build that has LS 12dB/HS 12 dB, I'll replace those with LS/HS in the next build.
 
Top Bottom