Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Multi-Way Conventional loudspeakers with crossovers

rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool
rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool
rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 22nd November 2019, 07:48 AM   #2861
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by mitchba View Post


Haha! Oh thank you sir!

Yes! I will definitely try this absolutely, and more importantly this page has some excellent keywords and terminology that I have never heard of before to get me on the track to my own research. I have something to do searches on now and find other stuff.

Most excellent, oh Iíll be busy for at least a month with this thank you thank you.
  Reply With Quote
Old 24th November 2019, 04:16 AM   #2862
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
@ ByRTT

So I got the cascade implemented just as you drew out.
And this time I did it right.

Sorry for all my double posts (it’s the old man in my spinning my wheels)


As far as this

Click the image to open in full size.

What if those speakers don’t roll off picture perfect,

I’ve been using peq on the speaker output to shape things. Wouldn’t that affect the phase of other speakers? Should I be using global eq on everything?

Like my horns crossed at 1000LR4, do I match the drivers gains to form the crossovers so I can use global eq on everything? If I want every crossover and eq to globally affect everything which I realize is ideal how do I solve the +15db peak at 4K in my horns and the +6db peak at 1.6. On horns without killing the efficiency at the top end.

If global eq is ideal I’ll figure out a way to make it work.



So the results. Got it all dialed in and working and sounding great.

A few points

1. I had no idea that it would be noticed this much especially the horns and subs

Adding the 2nd low pass on the sub made the midbass sound like the crossover was dropped an oactave and everything sounds much more “open” and airy.
And the singer voice sounds more real and life like rather than sounding like coming out of a pair of speakers. I always thought about an oactave past the crossover the speaker is no audible so why would it matter. But it seems as if it’s not attenuating anything, it seems like adding the cascade pushes the timing on the sub and horns in a way that lines them up with all the midrange frequencies.
The snare has attack now where I’ve never really experienced that before. Very interesting trick, I’ll always do this now. So thank you!

2. To get the midbass to get this awesome effect that makes them seem to play lower and the sub sounds more in front , with the sub I basically had to turn off the time delay between the midbass and sub. It doesn’t make perfect sense yet but I think it’s the long wavelengths or the reverb time of some kind causing this. I had 2.5ms delay of the midbass to time alignment them to the sub and that seemed to make things worse with your cascade trick. I basically just have the sub and midbass all come out at same time now even tho there’s path length differences. Definitely a lot more ambiance down low especially and up front bass that I’ve never had before.


Now on to my question.


So if linearization makes the crossover spin disappear why do I have to have this cascaded crossovers on all speaker and the firs? Can anyone in a basic dummy way explain how that works. I’m trying to wrap my head around this, I’ve run all the sims in rephase to try to figure out why and can’t make sense of it.

Thanks in advance

Last edited by Oabeieo; 24th November 2019 at 04:35 AM.
  Reply With Quote
Old 24th November 2019, 04:47 AM   #2863
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally Posted by Oabeieo View Post
@ ByRTT

So I got the cascade implemented just as you drew out.
And this time I did it right.

Sorry for all my double posts (itís the old man in my spinning my wheels)


As far as this

Click the image to open in full size.

What if those speakers donít roll off picture perfect,

Iíve been using peq on the speaker output to shape things. Wouldnít that affect the phase of other speakers? Should I be using global eq on everything?

Like my horns crossed at 1000LR4, do I match the drivers gains to form the crossovers so I can use global eq on everything? If I want every crossover and eq to globally affect everything which I realize is ideal how do I solve the +15db peak at 4K in my horns and the +6db peak at 1.6. On horns without killing the efficiency at the top end.

If global eq is ideal Iíll figure out a way to make it work.


So if linearization makes the crossover spin disappear why do I have to have this cascaded crossovers on all speaker and the firs? Can anyone in a basic dummy way explain how that works. Iím trying to wrap my head around this, Iíve run all the sims in rephase to try to figure out why and canít make sense of it.

Thanks in advance
I'm not sure that you and Byrtt are on the same page. That graph is an idealised representation of the crossover points and slopes you could have in a 4 way system with 2nd order rolloff's added at either end, because the crossovers are linear phase they add back to form a minimum phase sum that that matches the amplitude.

To do that in reality means that you need to EQ each driver individually to flat at least an octave or two either side of the crossover with minimum phase EQ, time align then add textbook linear phase crossovers. The rolloff at either end will most likely not be textbook 2nd order but you can EQ to that if desired.

The idea of the global EQ is to use that to set your overall target curve and by doing that with minimum phase EQ you will not upset the crossovers that were EQ'd and aligned separately.
  Reply With Quote
Old 24th November 2019, 05:50 AM   #2864
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by fluid View Post
I'm not sure that you and Byrtt are on the same page. That graph is an idealised representation of the crossover points and slopes you could have in a 4 way system with 2nd order rolloff's added at either end, because the crossovers are linear phase they add back to form a minimum phase sum that that matches the amplitude.

To do that in reality means that you need to EQ each driver individually to flat at least an octave or two either side of the crossover with minimum phase EQ, time align then add textbook linear phase crossovers. The rolloff at either end will most likely not be textbook 2nd order but you can EQ to that if desired.

The idea of the global EQ is to use that to set your overall target curve and by doing that with minimum phase EQ you will not upset the crossovers that were EQ'd and aligned separately.


Okay thank you fluid!

So okay than I was doing it right..... that exactly how I’ve been doing it. As it says way back on the early pages. I’ve always measured every speaker flat at close mic an oactave past any crossover , than added all my linearizations or linear phase crossovers, than global eq to shape the magnitude.

This has always worked really good.
This cascade that BYRTT has definitely does something different. I’m getting better ambiance without a doubt. I just adopted his blocks and dropped my LR4s in instead of what he has and my crossover points. So now I’m really confused.

In a car it’s almost impossible to get good measurements globally as the near reflections add way too much excess phase so windowing has to be pretty aggressive compared to the room size.

Doing the cascade the phase drops a lot like his picture where how I was doing it was more flat. They both sound good, the cascade has more ambiance. Not sure if that means better, but so far seems good. I like them both, but his cascade seems better, I’ll have to listen for a week to really say for sure.


So my system is basically this

3 2x4hds to a 4way all linear phase crossovers in the fir banks

1 opendrc upstream of the HDs with gloabal eq and the linearization parts of his cascade in the fir blocks.


The HD’s I used a peq bank in advanced to get the added crossovers for each speaker in iir, and again used the opendrc to do linearizations gloabally on everything.

So each speaker gets all the linearizations gloabally on the opendrc and it’s own crossovers and IIrs seperate on the HDs along with the initial fir crossover for the inband of each driver

The blocks form just like his diagram

It’s sorta goofy I might have done it wrong, the ambiance sounds a little artificial almost like reverb ish sounding. So either I botched it or it just needs a little more fine tuning

It was like 12 linearizations done gloabally on the opendrc , the phase turned from flat to a LOT of lines going through the magnitude like stripes , the phase just looked like lots of lines going from top to bottom over and over again in rephase , the measured phase was more down hill tho ......

Last edited by Oabeieo; 24th November 2019 at 06:11 AM.
  Reply With Quote
Old 24th November 2019, 09:02 AM   #2865
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Byrtt can confirm when he sees your post but I am fairly confident that the blocks that you refer to as "cascade" are just what is needed to represent those curves in Xsim.

I don't think the idea is to try and input those as corrections.

The reason the phase curves more at the bottom and top is because the magnitude is rolling off. If your magnitude does not roll off in that way then you don't want the phase to either. The low one is 2nd order Q 0.83 at 45Hz which is the figures you provided earlier, so it's a virtual representation of that.

If you like what all that unnecessary correcting does measure it and see if you can work out why
  Reply With Quote
Old 24th November 2019, 06:55 PM   #2866
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by fluid View Post
Byrtt can confirm when he sees your post but I am fairly confident that the blocks that you refer to as "cascade" are just what is needed to represent those curves in Xsim.

I don't think the idea is to try and input those as corrections.

The reason the phase curves more at the bottom and top is because the magnitude is rolling off. If your magnitude does not roll off in that way then you don't want the phase to either. The low one is 2nd order Q 0.83 at 45Hz which is the figures you provided earlier, so it's a virtual representation of that.

If you like what all that unnecessary correcting does measure it and see if you can work out why
Thank you again Fluid

So this morning I went and listened again. (Fresh ears)

I starting taking off the added cascade one by one as it didnít sound as natural as before.

So the part that I liked was having the 300hz low pass along with the 80hz low pass on the sub only. After reducing down and mixing out what seemed wrong the added lowpass cascade to the sub definitely sounds better.

I ended up taking everything else off and went back to the standard linear crossovers.

So I think Iíll keep it like this for awhile and like you said try and figure out why.
It makes the midbass sound like there playing sub frequencies. Itís a fantastic effect.

The car is 2.66ms wide so the 300hz crossover is right in that area. Maybe just a artifact of reflection and interference that adding the 2nd lowpass seems to help.
(Just a guess) or BYRTT is right and Iím not understanding him again.

As far as his diagram, even if it was from xsim and justba way to show theoretical flat phase with a 4way, wouldnít that technique still apply in real life.
I think he is on to something and I just havenít grasped it maybe.
Or I simply donít have the processing to do it exactly like he shows.

If each one of his lines in the block reads right to left and each row is a single channel an should not be global than I just donít have the process power to do it. (I wish the HDs had a 48k plugin)

The sub has 2042 t
Everything else only 1024 t

The opendrc has 6144 t

So I think I would need a opendrc on every pair of speakers to be able to do the
cascade on each speaker pair as he has it in rows , maybe trying to do all the linearizations on the opendrc wasnít correct and caused the reverb ish sound.
  Reply With Quote
Old 24th November 2019, 08:54 PM   #2867
BYRTT is offline BYRTT  Denmark
diyAudio Member
 
BYRTT's Avatar
 
Join Date: Dec 2009
rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool
Hi Oabeieo,

Know it can be hard to grasp therefor suggest you take some virtual study to educate your self, it is not that hard copy my example pass bands in Rephase export them as frd files import to REW where you can sum them on "ALL SPL" tab using the wonder math it can offer in "Controls" dialog. You will then see example sum perfect as was it one pointsource transducer, and when you do the same exercise but omit side and stop bands you get varius errors in amplitude and phase domains no matter you use FIR or IIR XO slopes.

Understand its a textbook example to get four way sum perfect flat when looked isolated from a real indoor enviroment, at the same we place that system into a indoor enviroment we get tons of extension in low end and ugly ripple from bounderys, but it doesn't change on that the optimal relation between pass bands is as in the example, to count for room gain plus ripple plus dial in the right house curve for system installation shall use global correction to form a perfect integration sum of the optimal textbook multi way with the room enviroment gain signature.

For example say you room enviroment gain signature was a friendly curve looking exactly as was it a linktwitz transform from Q0,83 @45Hz to Q0,7 20Hz, then we was set and happy in system pass band is now a flat minimum phase domain covering 20Hz-20kHz audio band.

Another note is i noticed my box simulation program in good old WinSpeakerz model auto cabin response as a linkwitz transfer from Q0,7071 @66Hz flat down to DC point (0Hz), if that is right think you should try transfer your system stopband in global EQ from Q0,83 @45Hz to Q0,7 @66Hz and set a 2nd order highpass in area 15-20Hz.
  Reply With Quote
Old 24th November 2019, 11:31 PM   #2868
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Maybe I confused you before saying not to input them directly, I should have said if you want to look at the cascade of filters ignore the low and high ones as they represent the system stop bands in the simulation and may not be what you have.

If you cascade the three crossovers slopes you will see what effect it has.

Here is an example using the low driver, comparing the LR24 at 80HZ in blue to the cascaded version using LR24 at 80Hz and LR24 at 300Hz.

The start of the rolloff is the same but gets to be a 48dB filter further along.If you add in the 2000Hz filter the change happens somewhere around -140dB so for real use it would be fairly safe to leave it out.

You could do the same thing by cascading the LR24 filters with IIR and then use filters linearization in a global FIR to undo the phase turn of the crossover filters. That would likely save you taps in your mindsp units.

Maybe it is this change in crossover slope that you like. The change only happens at -40dB so in reality it should be a fairly subtle effect.
Attached Images
File Type: png Screen Shot 2019-11-25 at 7.15.48 am.png (49.5 KB, 47 views)

Last edited by fluid; 24th November 2019 at 11:35 PM.
  Reply With Quote
Old 24th November 2019, 11:58 PM   #2869
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by BYRTT View Post
Hi Oabeieo,

Know it can be hard to grasp therefor suggest you take some virtual study to educate your self, it is not that hard copy my example pass bands in Rephase export them as frd files import to REW where you can sum them on "ALL SPL" tab using the wonder math it can offer in "Controls" dialog. You will then see example sum perfect as was it one pointsource transducer, and when you do the same exercise but omit side and stop bands you get varius errors in amplitude and phase domains no matter you use FIR or IIR XO slopes.

Understand its a textbook example to get four way sum perfect flat when looked isolated from a real indoor enviroment, at the same we place that system into a indoor enviroment we get tons of extension in low end and ugly ripple from bounderys, but it doesn't change on that the optimal relation between pass bands is as in the example, to count for room gain plus ripple plus dial in the right house curve for system installation shall use global correction to form a perfect integration sum of the optimal textbook multi way with the room enviroment gain signature.

For example say you room enviroment gain signature was a friendly curve looking exactly as was it a linktwitz transform from Q0,83 @45Hz to Q0,7 20Hz, then we was set and happy in system pass band is now a flat minimum phase domain covering 20Hz-20kHz audio band.

Another note is i noticed my box simulation program in good old WinSpeakerz model auto cabin response as a linkwitz transfer from Q0,7071 @66Hz flat down to DC point (0Hz), if that is right think you should try transfer your system stopband in global EQ from Q0,83 @45Hz to Q0,7 @66Hz and set a 2nd order highpass in area 15-20Hz.
Thanks BYRTT,

Alright that is making a lot more sense now. Sorry for thinking it was something different. For some reason I thought that was a model to try on a multi-way.


So itís going to take a little bit to digest this, I thought I was on to my next revelation in learning, okay so taking a step back now,Iíll definitely try that and see what happens.

So, going back to my original post. The closed box linearization that I asked about. I think what Iím trying to figure out or learn in a simple way is.

Letís say out of the car the box rolls off at f345 QTC.83, in car I get fairly flat to about 16hz. I do have to use eq at 66hz (you hit that on the head) and take it down -6db peq Q 1.4 to be flat to 20

My question is, if the box , the cabinet causes a natural roll off at 45hz but the room boosts below it and you say it acts as a Linkwitz Transform, that would mean itís minimum phase and doesn't need any farther work done from Rephase

I was wondering I guess if room gain is minimum phase now that I think about it


So I did end up learning off this, not what I expected to learn but still beneficial indeed

Iíll try and find some reading :-)

Thanks again
  Reply With Quote
Old 25th November 2019, 12:31 AM   #2870
fluid is offline fluid  Australia
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally Posted by Oabeieo View Post

My question is, if the box , the cabinet causes a natural roll off at 45hz but the room boosts below it and you say it acts as a Linkwitz Transform, that would mean itís minimum phase and doesn't need any farther work done from Rephase

I was wondering I guess if room gain is minimum phase now that I think about it
That was pretty much the discussion that was had above. If you linearize the sealed box you will change the phase away from minimum and end up with a phase that does not follow the amplitude response.

Measure the result you have and use REW with a FDW to clean up the phase, generate a minimum phase version in REW and compare to what you have. If there's excess phase you can export that and use rephase to try and correct for it and see if you like that any better.
  Reply With Quote

Reply


rePhase, a loudspeaker phase linearization, EQ and FIR filtering toolHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
FIR linear phase plugin for MiniDSP? diyjb01 miniDSP 17 9th June 2016 02:35 PM
FIR filter design tool for Loudspeaker magnitude equalization ttmusic Software Tools 3 24th May 2013 09:30 PM
FIR Filtering experiences Olombo PC Based 8 10th February 2013 04:45 PM
AVX based FIR VST, crossover / EQ / DRC and delay KOON3876 PC Based 97 26th November 2012 08:18 AM
Phase EQ using FIR filters Grasso Multi-Way 2 2nd July 2003 11:37 PM


New To Site? Need Help?

All times are GMT. The time now is 07:23 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio
Wiki