t.racks DSP mini and REW. How to apply REW correction into separate channels?

[Tupisac]

I have a 2.1 setup. So far I've been using software equalizer on the OS level. One curve for all, generated on the measured response from all the speakers playing simultaneously (sub + L + R).

Recently I've bought t.racks DSP mini and it has separate PEQ for each channel. It got me wondering what may be the best practice here. Should I measure each channel separately (sub / L / R) and generate separate correction curves?

Or maybe should I measure whole system after all the crossovers and delays are applied and then somehow chop the correction into separate channels in REW?

 

[staticV3]

First, if you have anechoic frequency response data for your speakers and there's something to correct there, apply that to the individual channels.

Next, use the microphone to find optimal delay, phase, and level settings for Sub integration. No PEQ at this step.

Once the system is well integrated, measure the combined L+R+Sub response with the Moving Microphone Method and apply room correction to the entire system (20Hz to ~400Hz).

Finally, you can add an LS and HS filter (Q 0.71) to the entire system and adjust bass and treble to taste.

 

[Tupisac]

Once the system is well integrated, measure the combined L+R+Sub response with the Moving Microphone Method and apply room correction to the entire system (20Hz to ~400Hz).

Yes, but I don't know how to do it correctly on the DSP Mini. EQ is only available there for individual output channels. No input or global EQ. And when there is a peak around crossover region I'd have to somehow split the suggested correction between the sub and speakers, right?

Theoretically I could use software EQ for that final correction but it adds some latency and doesn't work when I use my dawless setup via direct monitoring. I'd really love to make it all work independently from my computer.

 

[staticV3]

Yes, but I don't know how to do it correctly on the DSP Mini. EQ is only available there for individual output channels. No input or global EQ. And when there is a peak around crossover region I'd have to somehow split the suggested correction between the sub and speakers, right?

Simply apply the room mode filters from REW to the input channels InA/InB (where you feed your Stereo signal into).

DSP on the output channels then is only loudspeaker correction (if available) and sub integration.

 

[Tupisac]

Simply apply the room mode filters from REW to the input channels InA/InB (where you feed your Stereo signal into).

Unfortunately there is no option to add EQ on the inputs...

 

[staticV3]

Unfortunately there is no option to add EQ on the inputs...

Then you'll have to split the PEQ filters according to your crossover frequency and ideally avoid any PEQ adjustments directly at that frequency.

I'd apply the same PEQ filters to the L and R Mains output, unless measurements show strong response asymmetry worth correcting.

 

[90sNintendoKid]

First, if you have anechoic frequency response data for your speakers and there's something to correct there, apply that to the individual channels.

Next, use the microphone to find optimal delay, phase, and level settings for Sub integration. No PEQ at this step.

Once the system is well integrated, measure the combined L+R+Sub response with the Moving Microphone Method and apply room correction to the entire system (20Hz to ~400Hz).

Finally, you can add an LS and HS filter (Q 0.71) to the entire system and adjust bass and treble to taste.

For the anechoic corrections first and then room correction at the end between 20hz and 400hz, what’s your opinion on the interaction between the anechoic peq and room correction peq if they overlap not producing the predicted results in post EQ measurements?

This method of EQ is similar to how I do it except I use in room nearfield moving mic method measurements.

 

[Tupisac]

Then you'll have to split the PEQ filters according to your crossover frequency and ideally avoid any PEQ adjustments directly at that frequency.

With xo slopes at 24 db/oct there is still a ton of overlap. Crossed at 80 sub is still audible over 100 Hz mark.
I guess good thing is that REW predictions are really accurate. The crossed and aligned sums from raw channel signals are basically the same as final measurement (red is measured, green is predicted):

But it is quite a painful process and I don't look forward to it. Change peak's gain in one channel. Change in another. Sum the predicted measurements. Rinse and repeat.

I'd apply the same PEQ filters to the L and R Mains output, unless measurements show strong response asymmetry worth correcting.

Oh boy... L is green, R is amber:

 

[staticV3]

For the anechoic corrections first and then room correction at the end between 20hz and 400hz, what’s your opinion on the interaction between the anechoic peq and room correction peq if they overlap not producing the predicted results in post EQ measurements?

There is no unpredicted interaction here.

You apply PEQ, measure the response with PEQ, then apply additional PEQ.

 

[90sNintendoKid]

There is no unpredicted interaction here.

You apply PEQ, measure the response with PEQ, then apply additional PEQ.

Thanks, I’ve been playing around with eq stacking recently