Linkwitz-Riley multiple crossovers proximity

DSP, Plugin and Host development discussion.
Post Reply New Topic
RELATED
PRODUCTS

Post

How can I minimize the effect of setting up crossover frequencies too close to each other?
This is a 4th order LR. With 8th it's even worse. This is with unity gain of all the split bands. What did I miss here ?
LR-proximity.png
You do not have the required permissions to view the files attached to this post.

Post

With multiple bands (beyond 2) you need allpass filtering to compensate for additional phase shifts as you add more bands. Are you doing that?

Post

Yes I do. This is a 4 band scenario, so if my understanding is correct:
- band1 needs to be filtered with allpass2(=lp2+hp2) and with allpass3(=lp3+hp3),
- band2 needs to be filtered with allpass3

Then to recombine the bands, I do band4+ band3 + allpassed(band2) + allpassed(band1)

Look at the frequency response when crossover frequencies are far away - things are as expected then.

Post

When you split from 2 to 3 bands then band 1 needs allpass filtering. Splitting again from 3 to 4 bands then band 1 and 2 will need allpass filtering. So band 1 needs filtering by both allpass filters in other words

Post

That's exactly what i do. Still can't figure out what's wrong when crossover frequencies are close to each other.

Post

Ok, I missed that. I can't really guess what the problem is. How exactly are you designing the filters?

Post

From the graph is still looks like phasing artifacts. 4th order filters have 0 deg phase shift in pass and stopbands, which could explain why you're not getting them when the filters are far apart. Are you sure your order of splitting the bands corresponds to the order of allpass filtering?

Post

matt42 wrote: Wed Oct 10, 2018 10:08 am When you split from 2 to 3 bands then band 1 needs allpass filtering. Splitting again from 3 to 4 bands then band 1 and 2 will need allpass filtering. So band 1 needs filtering by both allpass filters in other words
Two constructions introduced here by Martin Eisenberg - https://www.dsprelated.com/showthread/c ... onological

Post

juha_p wrote: Thu Oct 11, 2018 8:38 am
matt42 wrote: Wed Oct 10, 2018 10:08 am When you split from 2 to 3 bands then band 1 needs allpass filtering. Splitting again from 3 to 4 bands then band 1 and 2 will need allpass filtering. So band 1 needs filtering by both allpass filters in other words
Two constructions introduced here by Martin Eisenberg - https://www.dsprelated.com/showthread/c ... onological
Yup, what I describe is figure 4b in his comment. 4a seems a smart optimization

Post Reply

Return to “DSP and Plugin Development”