2-way-advanced = 20 PEQs per channel

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
After it gathering dust for many months (previously as a two-way active crossover for a pair of Magnepans), I am giving the MiniDSP (2x4) a try as being a "Bose 901 (II) active equalizer" and room correction at once. I am a recent user of REW and I think have the basics of REW and MiniDSP's software. My question centers on this: to input REW data to the MiniDSP, I bought the 2-way crossover (PEQ) advanced. Here is where it gets intersting. REW will export values for the MiniDSP, but only 6 filters. However, this plug-in can do (per channel) 6 filters per input, and 6 per output channel, so for use as a 2x2 PEQ, I have 12 filters and -- this part I am still foggy about -- I think one can use the "crossover" section for an additonal 8 filters. This is not meant to be a REW question, but I think I found a way to make this plug-in a 20-band PEQ (for 2 channels in out). Here is what I did:

1. IN REW, I had my baseline room measurements (in my case, an average of about 4 or more readings.)

2. After much fiddling, I selected the "generic" EQ unit in REW which conveniently has twenty filters. I settled on 1/48 octave (I think...) smoothing, and +/- 1 db for the entire audio range. I let the auto eq function do its thing and I got a set of twenty PEQ filters. I did a "control-printscreen" + paste screen dump to printer so I had a hard copy to work with. i did not find any easy way to export from REW these values. No biggie...

3. Start the MiniDSP plug-in. There is a menu allowing one to calculate biquads so that is what I did for all twenty values. I made three separate text files to suit the plug-in, just cutting and pastting from plug-in to text file, and manually naming the biquads in order (1...6) and (1...8) for the "crossover" one. Watch those commas too. I named the files the hopefully clear names of "input.txt", "crossover.text" and "output.txt" to show which of the three sections of the plug-in they would be imported to.

4. I then imported using the "advanced" setting of the plug-in, the "input.txt" file to the input section and the "output.txt" section to the channel 1 output. Since my EQ is "mono", I copied as needed from L to R channel in/out.

5. The "crossover" section is similar. It appears to take the "crossover.txt" as the "low-pass" filter, and then you tell it the high-pass is bypassed. Do the same for the other channel.

6. Save your work and load to the hardware as usual. It seems to work.

This seems to be a cumbersome way to do it, but it looks as though you end up with a twenty band parametric equalizer. Not bad!
 
Today (03-22-2012) I re-did it a bit easier: I use REW to compute and export the settings for MiniDSP (to a file); I analyze the total response (averaged) as bass, treble, and highs. I look at bass (20-100 Hz, no smoothing) and see that I have major nulls at 88, 178 Hz. So I will not try and boost those. I EQ'ed (6 filters) the range 20-200 Hz for smoothest response, minimzing major peaks (worst) at 42, 70 and 140 Hz. This was "bass.txt"; Next I chose the range 201-1000 Hz for "mids"; I used some smoothing there and fiddled for the flattest graph. This becames "mids.txt"; finally the range of 1 - 10 KHz was mostly a gradual drop-off (about 7 db/oct > 2 KHz); I EQ'ed this with 1/3 smoothing and results to "highs.txt"; in this case, the entry into plug-in is much easier since it is either import from a file (PEQ) or cut&Paste for the "crossover"; with my set-up I can't easily A/B the different settings but the (03-21) set didn't sound right. THe (03-22) seems a bit more balanced. I lose 2 filters on the "crossover" but still have 18 per channel :)
 
Yes, thank you for the advice. I am aware that nulls are the acoustic version of a Black Hole, sucking in vibrations into another continuum of silence. "Equalization is futile. You will be assimilated."

Returning (ahem!) to the topic of using the MiniDSP as a super-EQ, I am still refining my "Bose 901 & Room" EQ. I have not yet run a REW sweep to see what my corrected response is...but I am getting close to what sounds good.

One pitfall occurs to me: my technique, as described so far, may result in errors on the boundaries (of low, mids, highs) because I assume REW does not "know" about other EQ's that are being applied outside the range I ask it to assign filters for. Since I am using the 2x4's three separate filter sections, it would be logical to make the changes in steps, i.e. do the "bass" (input 6 filters) first, apply EQ, new measurement and tweak until right, then work on the "mids" ("crossover"), EQ, tweak, etc.

I am considering writing a tutorial but obviously I don't completely know what I'm doing yet, so maybe another month or so :dunno:
 
Report for March 24

Today's filter set: I chose frequency nulls as the breakpoints for low, mids, and high sets of filters. Why: I am just reducing peaks, not trying to fill nulls. This seemed to work well for lows (20-177) and mids (177-2K Hz); however I have run into trouble with the highs. For the Bose 901 II, I have a roll-off above 2K. REW did not seem to handle this well, at least with the EQ function. Using the manual settings and huge amounts of boost, I was able to get response to 20K reasonably flat (at least the "predicted" response.) When loaded into the 2x4, however, they show that I am trying to boost > the +20 dB limit (?) of the plug-in. Never the less, the sound from this config set sounds to be working quite well.
 
Are you using analog in, and doing the volume control before the DSP? With that kind of boost and a full signal into the DSP there's a good chance you're clipping the outputs, although if it sounds good maybe not. If you attenuate before the boards you should be ok.

I use the mini's full boost capability, doing the volume control onboard, and I need to be creative to not clip the output. This is quite audible!
 
Guilty on all charges :) With the current filter, I attenuate 20 db both inputs but still get clipping due to the massive boost between 2K-20K, although I only hear distortion during the frequency sweep. In theory, I am trying to cut everything and boost nothing. This is not the first 901 I have equalized.

Memo to self: to save time and avoid erroneous measurements, make sure you have the DEQ2496's dynamic range expansion out of the circuit before running REW sweeps :headbash:

"The piano has been drinking, not me" -- Tom Waits from [of course!] "The Piano Has Been Drinking" :drink: :Piano:
 
Last edited:
To anyone following this thread, do you know of a biquad calculator that would easily give me something like a boost of 7 dB/octave > 2 KHz? That is my approximate roll-off in my installation. I am still kind of an amateur at these newfangled biquadratic thingies. REW appears to be useless for huge boosts and I otherwise must do unspeakable things with PEQ filters and huge boosts.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.