INTERACT FORUM

More => Old Versions => JRiver Media Center 26 for Windows => Topic started by: BCZ on March 10, 2020, 01:11:42 am

Title: Something is wrong with the low pass/high pass filters
Post by: BCZ on March 10, 2020, 01:11:42 am

So I'm trying to make a Linkwitz Riley filter using DSP studio to build an active speaker. I thought I'd make sure the signal is correct through loopback measurements, and I think something is wrong.

In JRiver, one makes a Linkwitz Riley low pass/high pass filter by stacking two LPF/HPF together.

Linkwitz Riley LPF/HPF has the property that if you sum the low pass filtered output with the high pass filtered output, they would sum completely flat.

However, I double and triple checked, when I stack 2 of the HPF/LPF in JRiver, and try summing them together, it does not give a flat response. It has a dip at the crossover frequency.

Please take a look. Thank you

Title: Re: Something is wrong with the low pass/high pass filters
Post by: Matt on March 11, 2020, 07:59:38 am

There's a little bit of energy left right at the cross-over.  If you look at the graphs playing pink noise with a high pass and with a low pass, you'll see.

I think that's just the way it is, because the graphs of each cross-over look correct to me.

Title: Re: Something is wrong with the low pass/high pass filters
Post by: mwillems on March 11, 2020, 08:42:07 am

Quote from: BCZ on March 10, 2020, 01:11:42 am

So I'm trying to make a Linkwitz Riley filter using DSP studio to build an active speaker. I thought I'd make sure the signal is correct through loopback measurements, and I think something is wrong.

In JRiver, one makes a Linkwitz Riley low pass/high pass filter by stacking two LPF/HPF together.

Linkwitz Riley LPF/HPF has the property that if you sum the low pass filtered output with the high pass filtered output, they would sum completely flat.

However, I double and triple checked, when I stack 2 of the HPF/LPF in JRiver, and try summing them together, it does not give a flat response. It has a dip at the crossover frequency.

Please take a look. Thank you

Can you clarify what orders of HPF and LPF you are summing to get the Linkwitz riley and the rest of your parametric EQ settings?  You may know this already, but it's worth checking: if you're summing two second order filters to create a 4th order LR there should be no dip (and I see no dip measuring my own system), but if you're summing two first order filters to create an LR 2, you need to invert either the high frequency channel or the low frequency channel to account for the phase rotation or you'll get a very serious dip at the crossover.

Title: Re: Something is wrong with the low pass/high pass filters
Post by: BCZ on March 15, 2020, 12:30:06 pm

This is weird. It sums flat if I sum in JRiver, then measure the loopback. It doesn't sum flat if I measure, then sum the results after measurement. The FIR version does sum flat.

Title: Re: Something is wrong with the low pass/high pass filters
Post by: mwillems on March 15, 2020, 03:58:04 pm

Quote from: BCZ on March 15, 2020, 12:30:06 pm

This is weird. It sums flat if I sum in JRiver, then measure the loopback. It doesn't sum flat if I measure, then sum the results after measurement. The FIR version does sum flat.

How are you time aligning the measurements?  That could be a clue.

Title: Re: Something is wrong with the low pass/high pass filters
Post by: BCZ on March 26, 2020, 01:38:52 am

I'm just using REW to measure the loopback signal, then summing using REW's add function.