Hi ThierryHi Thomas,
You would say 2 2nd order all-pass ?
I'm getting out...
No, this was correct: a 2nd order all-pass has a 360° phase shift and corresponds to an 4th order LR crossover summation.
I implemented them into rephase essentially for simulation purpose, like when designing a correction targeting an IIR DSP unit.
And honestly/sadly that is what I use rephase for most of the time
Yeah, they called this distortion, the old quasioptimumxoversquad's main concern...
Then you sadly discover that linear phase has also its cons, as latency, not good for video, not good for musicians, not good for pa live events...
Rephase is great tool, but the name kind of ridiculous...
Thx again, Pos, BYRTT, and all,
I see that nearly everybody that commented since I asked about inverted 2nd order all-pass knows more about filters than I do, so I'm thinking to just put out what i was interested in using them for, and ask for alternatives.
I've been getting very good tuning results on several different 3-way pro sound rigs, using rePhase, with a channel of miniDSP's openDRC per driver.
The speakers are EAW kf965s, JTR 3TXs without their passive x-overs, and the Peter Morris DIYs.....all made for live sound....although I sure as heck love em for life sized playback
For playback, the 64ms latency of using all the openDRC's taps (6144) is no problem.
But I've begun using the gear for live sound, and after a number of gigs, 15ms seems to be a desirable target.
I know such a target is easily reached with IIR crossovers, but I've come to appreciate the ease of tuning, clarity, and better trasient response that flatter phase brings.
Right now, with my current tuning ability, I can stay at about 15ms latency using linear phase LR4s at 100Hz. So that get's rid of the IIR crossover phase wraps.
Where I'm have trouble with phase, is with the lower end of the mid drivers combining with the subs...ie thru the 100Hz xover region.
The number of taps available at 15ms is insufficient to flatten the mid driver's raw response there. Out of band EQ helps, filters linearization helps, parametric eq helps, but still I'm left a bit short.
IME, whenever I look at a drivers's raw magnitude response thru it's passband , I see a rough upside-down bowl shape.
And for phase, a big smiley face. It's the downward sloping, low frequency end of the smiley face that i thought 2nd order inverted all-pass would be perfect for adjusting to flat.
I was hoping it could be done IIR.
But now I see that's a task for FIR and needs more taps just like the filters linearization and parametric phase do.
So my speaker tuning guru friends....what are some of the most "latency efficient" ways of tuning down low ???
Sorry for such a long *** question
I see that nearly everybody that commented since I asked about inverted 2nd order all-pass knows more about filters than I do, so I'm thinking to just put out what i was interested in using them for, and ask for alternatives.
I've been getting very good tuning results on several different 3-way pro sound rigs, using rePhase, with a channel of miniDSP's openDRC per driver.
The speakers are EAW kf965s, JTR 3TXs without their passive x-overs, and the Peter Morris DIYs.....all made for live sound....although I sure as heck love em for life sized playback
For playback, the 64ms latency of using all the openDRC's taps (6144) is no problem.
But I've begun using the gear for live sound, and after a number of gigs, 15ms seems to be a desirable target.
I know such a target is easily reached with IIR crossovers, but I've come to appreciate the ease of tuning, clarity, and better trasient response that flatter phase brings.
Right now, with my current tuning ability, I can stay at about 15ms latency using linear phase LR4s at 100Hz. So that get's rid of the IIR crossover phase wraps.
Where I'm have trouble with phase, is with the lower end of the mid drivers combining with the subs...ie thru the 100Hz xover region.
The number of taps available at 15ms is insufficient to flatten the mid driver's raw response there. Out of band EQ helps, filters linearization helps, parametric eq helps, but still I'm left a bit short.
IME, whenever I look at a drivers's raw magnitude response thru it's passband , I see a rough upside-down bowl shape.
And for phase, a big smiley face. It's the downward sloping, low frequency end of the smiley face that i thought 2nd order inverted all-pass would be perfect for adjusting to flat.
I was hoping it could be done IIR.
But now I see that's a task for FIR and needs more taps just like the filters linearization and parametric phase do.
So my speaker tuning guru friends....what are some of the most "latency efficient" ways of tuning down low ???
Sorry for such a long *** question
If possible how about develop some subs that perform better than normal up into mid range area, that should up your lowest XO point higher in frequency and thereby for lowest system XO point less taps are needed.
Yeah, they called this distortion, the old quasioptimumxoversquad's main concern...
Then you sadly discover that linear phase has also its cons, as latency, not good for video, not good for musicians, not good for pa live events...
Rephase is great tool, but the name kind of ridiculous...
Bufff better think this twice... Been there, heard these voices too...
Taps, taps, always taps...
We only try to correct some limitations in devices used to record and reproduce real world sounds here, so minimum phase lover to stop us Rephase users correct in time domain please come up with some one point wide band microphone and speaker drivers that covers say DC-3-500kHz
.Hi BYRTT, yes, that's a viable option. I've built some bass-reflex (and sealed) subs that use the BMS 18N862 18", and it's amazing how high and smooth they go. Really great driver/subs IMO.
But when outside for live, i need about 6 of those per main (PM design).
So instead, I use labhorns or another horn loaded sub that sound better 100Hz or less.
We only try to correct some limitations in devices used to record and reproduce real world sounds here, so minimum phase lover to stop us Rephase users correct in time domain please come up with some one point wide band microphone and speaker drivers that covers say DC-3-500kHz
Well said
...although I'd be quite happy w 20-20K..
We only try to correct some limitations in devices used to record and reproduce real world sounds here, so minimum phase lover to stop us Rephase users correct in time domain please come up with some one point wide band microphone and speaker drivers that covers say DC-3-500kHz
Yeah, time domain is free, like Catalonia or Danemark!
Sorry but this sounds more as ideology to me, than science.
Last edited:
What is your acoustical target slope?
Are you linearizing an existing IIR filter or building the whole crossover with FIR?
Even if you are limited to 15ms you can still use 6144 taps and force centering to "15ms". That will let you use minimum-phase EQs and/or filters toghether with your linear-phase filters.
I'd like as steep an acoustical slope as possible, really 96dB oct if I could get it without latency. Subs and mains sometimes get co-located, sometimes not, ....which of course causes large summation variances in different audience spots.
I don't hear "preringing" from steep x-overs nearly as well as I hear the summation irregularities at the different listening spots. I've found minimizing acoustical overlap really helps reduce magnitude holes resulting from different sub to main distances.
When doing live sound, I'm using a Linea Research SC-48 processor...which has what are essentially linear phase LR4s that you can change on the fly.
So, with this processor I do the best IIR work I can do, driver by driver, then apply the linear-phase LR4s and adjust timing.
I will try forcing centering at 15ms, thx.
I'm trying to use rePhase and the openDRCs to better what I can do with the Linea.....again for live sound...otherwise this is all alot easier....
Yeah, time domain is free, like Catalonia or Danemark!
Sorry but this sounds more as ideology to me, than science.
Right if its science for our hearing but that front also moves from time to time, regarding real world wave shapes i can do my own science and see output verse input is hopeless distorted no matter what we hear
.Wave shape output 50Hz SQ wave from a DC - light-speed domain:
Wave shape output 50Hz SQ wave from summed DC - 500Hz pass-band with a 2nd order LR XO point then 500Hz - light-speed pass.band:
Wave shape output 50Hz SQ wave from a 20-20kHz domain:
Wave shape output 50Hz SQ wave from summed 20 - 500Hz pass-band with a 2nd order LR XO point then 500Hz - 20kHz pass-band:
Attachments
Last edited: