REW PEQ Help

[Oski1928]

I have been using REW with a Umik 1 to generate PEQ filters which I then manually put into my WIIM Ultra. When I do this I have REW set to basically only reduce frequencies with no boosting. For basically all frequency reductions that REW has generated it has given me two filters (one boosting and one cutting) that are very close to each other in frequency. The result is that it reduces the desired frequency, but why does it do it this way?

Is this the normal way to implement PEQ filters? If so can someone explain why this is better than just one filter reducing the frequency?

 

[staticV3]

but why does it do it this way?

Likely because the measured peak is not a particularly good match for the shape of a single Peaking filter.

By placing a seond Peaking filter next to it, the shape of the peasured peak can be better matched.

Screenshots would tell us more.

 

[levimax]

I have been using REW with a Umik 1 to generate PEQ filters which I then manually put into my WIIM Ultra. When I do this I have REW set to basically only reduce frequencies with no boosting. For basically all frequency reductions that REW has generated it has given me two filters (one boosting and one cutting) that are very close to each other in frequency. The result is that it reduces the desired frequency, but why does it do it this way?

Is this the normal way to implement PEQ filters? If so can someone explain why this is better than just one filter reducing the frequency?

Automatic filter generation uses algorithms to try to generate the best filters. There are different theories on the best way to do this and for commercial products this is their "secret sauce". I think REW's algorithms are explained somewhere in their documents but if it is making filters as you describe it is because the filter used to cut the peak is spilling into areas that don't need a cut so it has to bring those up.

 

[NTK]

John's supposedly follows the algorithm described in this paper to calculate the PEQ parameters in REW.
See this post and follow the link to download the paper from OneDrive.

A

Post in thread 'How to compute parameter values (frequency, gain, Q-value) to EQ a loudspeaker'

Oct 18, 2020

Someone asked John M. what algorithm he used in REW to do the filter computations. I can't find that post now, but he pointed the user to this article.

• andyc56

• 

 

[ErVikingo]

On a related topic, I measure my setup with REW and UMIK1 (with calibration file) and get a (to me) pretty decent curve. I ask REW to calculate PEQ and load it into minidsp. When I measure with the PEQ the resulting curve is nothing like the projected corrected curve. I like that it tames a peak at approx 40, 55 to 80 and 210 but it creates a large peak approx 150hZ and a dip at 45hZ.

Filters applied <500hZ; Green no EQ; Red EQd

No subwoofer on my system. The speakers go down and I have them limited to 15hZ

Thoughts?

(edit: I wish I understood REW and DSP better)

 

[staticV3]

Thoughts?

Measure using the Moving Microphone Method.

Single point sweeps can have large run-to-run variance.

 

[ErVikingo]

This is measured vs Predicted

 

[levimax]

On a related topic, I measure my setup with REW and UMIK1 (with calibration file) and get a (to me) pretty decent curve. I ask REW to calculate PEQ and load it into minidsp. When I measure with the PEQ the resulting curve is nothing like the projected corrected curve. I like that it tames a peak at approx 40, 55 to 80 and 210 but it creates a large peak approx 150hZ and a dip at 45hZ.

Filters applied <500hZ; Green no EQ; Red EQd

No subwoofer on my system. The speakers go down and I have them limited to 15hZ

Thoughts?

(edit: I wish I understood REW and DSP better)

View attachment 486647

This actually looks as expected and pretty good except for the large new peak at 140 Hz. Generally sharp dips are not audible but peaks are so knocking down the peaks and leaving some dips is OK. Can you post the setting you used in REW to generate the filters. See screenshot below of the setting I am asking about.

 

[ErVikingo]

This actually looks as expected and pretty good except for the large new peak at 140 Hz. Generally sharp dips are not audible but peaks are so knocking down the peaks and leaving some dips is OK. Can you post the setting you used in REW to generate the filters. See screenshot below of the setting I am asking about.

View attachment 486652

Thanks!

What i would like to tame is mostly the peak I have at 210 approx. I dont mind correcting below 80.

 

[levimax]

Thanks!
View attachment 486654

What i would like to tame is mostly the peak I have at 210 approx. I dont mind correcting below 80.

The measured response in Post #5 and Post #7 are different. Are these both actually measured responses or is the one in Post #5 a predicted response? Also as mentioned are these "sweeps" or "MMM" (Moving Mic Measurements) ?

 

[ErVikingo]

Post 5 is measured. Actually so is 7 but smooting is different.

If it helps, I'll do all sweeps again tonight and will post back )at office now). Any other details I should share?

 

[ErVikingo]

Also as mentioned are these "sweeps" or "MMM" (Moving Mic Measurements) ?

These are sweeps.

 

[levimax]

These are sweeps.

For confirming FR I like MMM measurements as they are quick and easy and accurate. The problem is they won't work for timing or distortion or other things but for what you are trying to accomplish they are fine or even better than sweeps. To get consistent results with sweeps you need to take a lot of them (5 to 10) at slightly different locations and heights around the LP and then average them. Here is a video that shows how quick and easy MMM measurements can be.

 

[ErVikingo]

All looks proper *did the moving mic per the instructions on the video. got good measurements L, R and Both. Ran EQ and loaded BIQUADS.

Pretty sure i have something screwed up on my REW settings

 

[levimax]

All looks proper *did the moving mic per the instructions on the video. got good measurements L, R and Both. Ran EQ and loaded BIQUADS.

View attachment 486772

Pretty sure i have something screwed up on my REW settings

Can you post all of your REW settings starting from the top?

 

[Keith_W]

1. Read this eBook (does not contain instructions on how to make PEQ's for DSP though). Just so you know how to take room measurements.
2. Post your MDAT.

 

[janbth]

All looks proper *did the moving mic per the instructions on the video. got good measurements L, R and Both. Ran EQ and loaded BIQUADS.

View attachment 486772

Pretty sure i have something screwed up on my REW settings

Two comments:

1. The vertical axis should cover something like 50 dB, which is customary. Let’s say from 50 to 100 dB.
2. Use Var smoothing to show relevant details in the plots.