DSPRelated.com
Forums

tuning a parametric equalizer

Started by Adam M. February 4, 2005
Jon Harris wrote:

> Also, I just remembered there is an app note that discusses graphic EQs in > detail: > http://www.rane.com/note101.html > I think that may address many of your questions. >
I think these RaneNotes ( http://www.rane.com/rnumber.html ) rate a mention in FAQ's some where. Not explicitly DSP but point out things us newbies should be at least aware of.
in article 1107562092.243811.38420@z14g2000cwz.googlegroups.com, AdamMil at
adamm@san.rr.com wrote on 02/04/2005 19:08:

> I changed the EQ from bandpass-in-parallel to peaking-in-series and it > appears to be working. Now I'm back to the original problem of tuning > it. Are there any guidelines for choosing the Q value for a given > frequency spread? (Again, it's a 10-band equalizer doubling from > 31.25hz to 16khz.)
this appears to be one band per octave. is that so? if it is so, i would recommend starting with the bandwidth, BW, set to 1 octave. the bandwidth parameter is a direct function of Q (and a little dependent on the center frequency, f0, due to frequency warping of the bilinear xform). it won't be exactly flat, but a decent goal would be to set BW or Q so that if all levels of the graphic EQ were raised by the same dB amount, that the resulting frequency response would be flat. in this case it won't be flat, but it *will* be approximately flat to two points for each band (if the BW in octaves is equal to the octave spread of neighboring f0). because of overlap of bands, the gain at the band centers will be a little larger than the gain setting of the peak, so maybe setting the BW to be slightly less than 1 octave would be better. that could use a little experimentation. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
"robert bristow-johnson" <rbj@audioimagination.com> wrote in message
news:BE299294.448F%rbj@audioimagination.com...
> in article 1107562092.243811.38420@z14g2000cwz.googlegroups.com, AdamMil at > adamm@san.rr.com wrote on 02/04/2005 19:08: > > > I changed the EQ from bandpass-in-parallel to peaking-in-series and it > > appears to be working. Now I'm back to the original problem of tuning > > it. Are there any guidelines for choosing the Q value for a given > > frequency spread? (Again, it's a 10-band equalizer doubling from > > 31.25hz to 16khz.) > > this appears to be one band per octave. is that so? > > if it is so, i would recommend starting with the bandwidth, BW, set to 1 > octave. the bandwidth parameter is a direct function of Q (and a little > dependent on the center frequency, f0, due to frequency warping of the > bilinear xform). > > it won't be exactly flat, but a decent goal would be to set BW or Q so that > if all levels of the graphic EQ were raised by the same dB amount, that the > resulting frequency response would be flat. in this case it won't be flat, > but it *will* be approximately flat to two points for each band (if the BW > in octaves is equal to the octave spread of neighboring f0). because of > overlap of bands, the gain at the band centers will be a little larger than > the gain setting of the peak, so maybe setting the BW to be slightly less > than 1 octave would be better. that could use a little experimentation.
RB-J has hit on some of the areas of difficulty with graphic EQs composite frequency response matching what the sliders show. It's a trade-off. Consider the case of 2 neighboring filters being boosted the same amount: if the filters are narrow enough so the resulting boost is about the same as the each band, there will be a significant dip between the bands. If they are wide enough to completely eliminate the dip, the composite boost will be significantly larger than the individual band settings. But now that DSP is replacing static filters for graphic EQs, there have been some advancements in this area--basically tweaking the filter boost and width/Q dynamically so that the composite shape better matches the slider settings. Even if these "smart adjustments" aren't made, most digital EQs show a composite frequency graph so the user can at least tweak their settings to achieve the composite curve they are looking for.