Setting up a graphic EQ with REW

[robbnj]

Hello all.

I'm tinkering around with a programmable DSP from Parts Express (the DSPB-K). I'm not at the point of being able to dump an EQ curve from REW onto it (yet), but I have been able to create a graphic EQ setup to at least learn how it works, the quirks, etc.
I can create a fairly complex graphic EQ, with as many "sliders" as a I choose, each one at whatever center frequency and Q that I choose,

For testing, I did a sweep of one of my office speakers, and tried to create an adjustment curve. I chose the generic/generic output, with 12 bands, and got this result:

Am I correct in assuming the lowest field in each set of three is to represent the Q for each frequency I am going to center my "sliders" at?
One issue that I can already see is that there are multiple corrections that are very close in frequency, but almost the opposite in settings. Thinking maybe I should disable the "allow narrow filters below 200Hz" option (?).

Unfortunately, I can't run a white noise tone and adjust on the fly until I get things "right", so I'm trying to get my best approximations from REW. I then plug them in to the schematic, which then gets dumped on the DSP module.
Yes, I could buy a MINIDSP, but I thrive on hands-on stuff like this.

The simple beginnings of my basic equalizer module:

 

[Wayne A. Pflughaupt]

Yes, the bottom field is filter bandwidth, aka Q.

A reference graph (i.e. before EQ) would be nice to see, but there is little-to-no chance your curve needs that many filters.

Throw out any filters tighter than ~1/4-octave (5.8Q), with +/-3 dB as they will not be audible with program material. (+/- 3 dB can be audible with broad filters, though.)

The idea is to get an improvement in sound quality. This is typically achieved by addressing the worst problems, not chasing every little ripple in response.

Regards,
Wayne

 

[robbnj]

Yes, the bottom field is filter bandwidth, aka Q.

A reference graph (i.e. before EQ) would be nice to see, but there is little-to-no chance your curve needs that many filters.

Throw out any filters tighter than ~1/4-octave (5.8Q), with +/-3 dB as they will not be audible with program material. (+/- 3 dB can be audible with broad filters, though.)

The idea is to get an improvement in sound quality. This is typically achieved by addressing the worst problems, not chasing every little ripple in response.

Regards,
Wayne

Thank you for the reply! I did 12 because I'm used to 10-band EQs, and 12 is more, lol.
I did another that is 6 b/c the DSP device was being finicky.

Here is the graph of the sweep I took of the speaker, from about 12" in front of it. Not properly set up for testing, I just wanted to learn, and figured I'd use this one. It's conveniently at my desk and sounds like it's got some "honk" to it that I'm interested in correcting (the spike at 1K and the dip at 1.5K, I presume)..

 

[Wayne A. Pflughaupt]

Twelve inches is likely too close to get a meaningful measurement. Most manufacturers give response specified at a distance of three feet.

Regards,
Wayne