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
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 29th August 2019, 03:20 AM   #2751
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
This is a very nifty feature!

So I basically was able to use the compensate feature to sorta put what looked like a backwards filter to tell me what the actual rollofff I was getting. Than apply the filter I wanted.

So in the end a 48db linearization seemed to be very close to what I was doing with compensation and adding a 24db minimum phase filter


After I did the compensate and applied the filter (by the way with compensate it looked flat like picture perfect I was very impressed) I than cleared the measurement so I could see only the generated fir without the measurement, than I turned on a 48db linearization and a 24db minimum phase filter and it looked very close to the same with much less samples required to convolve.

So I went with 48db linearization on the high pass and a 24db linearization on the low pass and I am out of taps using a rectangular window. I didn’t have enough to do all of the other without the magnitude moving and changing shape.

So the low pass is only halfway linearization I couldn’t get it the whole way so had to do the other half in the 2x4hd. It seems to work, I did a 24db linearization on the opendrc and a 24db linearization on a 2x4hd (running as the DAC and in 2x2 mode)
I had to kick the sub off the other 2ch to get the taps needed but I have already ordered another 2x4hd for the sub should be here tomorrow.


Overall it seems to work. Took a lot more fir than I thought
I wonder if I can use delay to move some of the phase so I don’t need so many taps.

Last edited by Oabeieo; 29th August 2019 at 03:22 AM.
  Reply With Quote
Old 29th August 2019, 01:52 PM   #2752
pos is offline pos  Europe
diyAudio Member
 
pos's Avatar
 
Join Date: Feb 2008
Location: Paris
Quote:
Originally Posted by Oabeieo View Post
This is a very nifty feature!
I must say I am quite proud of that compensate functionality
Once you add the target filter it can be seen as a generalized Linkwitz transform.


Quote:
I than cleared the measurement so I could see only the generated fir without the measurement, than I turned on a 48db linearization and a 24db minimum phase filter and it looked very close to the same with much less samples required to convolve.
That should not be the case if the result is exactly the same: any mean of achieving a given transfer function will "consume" the same amount of taps. So the difference in taps needed is probably due to some minor modification in the transfer function.

Quote:
So I went with 48db linearization on the high pass and a 24db linearization on the low pass and I am out of taps using a rectangular window. I didnít have enough to do all of the other without the magnitude moving and changing shape.
Have you tried playing with the impulse centering? As you compensate some existing minimum-phase behavior the ideal center would probably be in the second half of the FIR (ie over 50%: you can use pourcentages in the centering field).

Quote:
Overall it seems to work. Took a lot more fir than I thought
I wonder if I can use delay to move some of the phase so I donít need so many taps.
You can play with delays to help linearizing the phase of a low-pass filter, but unfortunately not a high-pass.
In your case I would personally use the minimum slope needed (ie giving a good enough attenuation where you need it) to avoid using too much taps as well as avoid too much (pre)ringing.
Partially linearizing the phase of a filter is also perfectly acceptable, and will turn some of the preringing into some more postringing.
__________________
2019-01-16: rePhase 1.4.3
  Reply With Quote
Old 29th August 2019, 02:26 PM   #2753
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by pos View Post
I must say I am quite proud of that compensate functionality
Once you add the target filter it can be seen as a generalized Linkwitz transform.


That should not be the case if the result is exactly the same: any mean of achieving a given transfer function will "consume" the same amount of taps. So the difference in taps needed is probably due to some minor modification in the transfer function.

Have you tried playing with the impulse centering? As you compensate some existing minimum-phase behavior the ideal center would probably be in the second half of the FIR (ie over 50%: you can use pourcentages in the centering field).

You can play with delays to help linearizing the phase of a low-pass filter, but unfortunately not a high-pass.
In your case I would personally use the minimum slope needed (ie giving a good enough attenuation where you need it) to avoid using too much taps as well as avoid too much (pre)ringing.
Partially linearizing the phase of a filter is also perfectly acceptable, and will turn some of the preringing into some more postringing.
I don’t know how to do different centering.

I just use exact centering. And energy on bass if taps start to run out.

I watched a few videos on impulse centering and went way way over my head (they busted out a bunch of math instead of explaining it) .


I remember the old rephase had a % sign but the new version is gone so I wouldn’t know how to turn it on.






As far as the part about exactly the same^ well it was close enough but wasn’t exactly, but very close to. Still sounded better.
I took off the low pass and just did the high pass in the fir with compensate
It’s better. Maybe need another opendrc or just keep the minimum phase on highpass. Haven’t decided yet, I’ve tryed a few different versions. Definitely sounds better than trying to fir everything. The magnitude was wiggles, the halfway on the LP is decent also.

Using delay with min phase LP seems to sound the best. I think I’m trying to do too much or not doing it right. There’s definitely a element of echoes in what sounds like the next octave up

Last edited by Oabeieo; 29th August 2019 at 02:46 PM.
  Reply With Quote
Old 29th August 2019, 03:58 PM   #2754
pos is offline pos  Europe
diyAudio Member
 
pos's Avatar
 
Join Date: Feb 2008
Location: Paris
Quote:
Originally Posted by Oabeieo View Post
I remember the old rephase had a % sign but the new version is gone so I wouldn’t know how to turn it on.
You can just enter any value you like, like "8ms", "8" (for 8 samples), "10cm", "60%". "middle" is equivalent to "50%".
It will complain with an "unknown centering" error when clicking the "generate" button if the value is not a valid one.
If you click on the drop down menu on the centering entry (the upper one) you will see a few examples.
You can also combine them like "middle+8ms"
__________________
2019-01-16: rePhase 1.4.3
  Reply With Quote
Old 30th August 2019, 04:54 AM   #2755
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by pos View Post
You can just enter any value you like, like "8ms", "8" (for 8 samples), "10cm", "60%". "middle" is equivalent to "50%".
It will complain with an "unknown centering" error when clicking the "generate" button if the value is not a valid one.
If you click on the drop down menu on the centering entry (the upper one) you will see a few examples.
You can also combine them like "middle+8ms"


Okay I heard you guys talk about that but never tried it. (Cause I donít know what itís for)

Iíll try that tomorrow. I fiddled with it a little tonight.

I took off the opendrc and put it back on the sub so I could do my linear phase crossover and do box correction (sealed qtc .8 46hz) and that btw sounds soooo good in a car my goodness that 18Ē sounds as transient and in time as a 6.5Ē sub itís just amazing.

So I got the 4th 2x4hd today and dedicated it to these 10s

2042t is all I got now. After auditions between the opendrc and the 2x4hd together and just the HD by itself , itís such a narrow bandwidth I canít really tell a whole lot difference when everything is playing, and I liked the sub better with all its correction.


So now that I know what the box and crossover are doing I just tried some linearization only with minim phase crossovers.

A had to spread by bandwidth because the inband was too attenuated so itís now 65hz-110hz (48db slopes both sides) have me a 16.6ms fir which mates up fine to my 17.66ms fit to sub with 3 more ms added to this 10Ē.

So I used this fir and added 3ms delay on top and itís sounding very close to when it had the open and Hd in series.

Pic below (yeah it looks bad but sounds ok)


What is it about narrow bandwidth that makes the speaker sound spitty?

Maybe linearization isnít the problem Iím hearing. (I figured the giant smiley face in time domain was the cause) but after getting it smoothed with the opendrc and HD it didnít make it super smooth. The timing part was much better but something else is going on.


Iíve heard of ppl doing super narrow bandwidth like in H-K filers and what have you, itís like the speaker hates it and thereís so much attenuation caused by the filters.

Would brickwall filters be better for this? Like 96db ? So I can get more gain on the passband, how little it is?

Iím not afraid of buying dsps to do the task, would a pair of opendrcs with 48k plug in do the job?




When I play this box with an oactave of bandwidth it sounds great!
B&C 10NWL-64-4 neos, I just want this massive hole at 70hz gone and this box works great for it, just trying to get it to sound smooth seems challenging a bit.
I chose this bass speaker for its low Q for small box and so I can give it 350wrms if I need to, would rant her not.

I guess what Iím getting at , is , is there a bass driver specifically designed for narrow bandwidth? Iím not a speaker design expert, i can muddle my way through winisd fairly well, but that so far on this seems maybe questionable now


Click the image to open in full size.
  Reply With Quote
Old 30th August 2019, 06:28 AM   #2756
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,

Maybe i misunderstand your situation but can remember from past when pos did link to article over Linkwitz site about how summing of multiple band pass especially if band pass gets narrow can lead to nasty summing errors if side band slopes is not calculated into final slope, to show what i mean see below two graphs from VituixCAD and take a close look in block diagram area that when side band filters are bypassed (greyed out) we get huge summing errors seen in the second graph, reason seems to be at 20dB down and below on center the narrow filler we have three transducers in a advanced sum and the small variation in final slopes then output the relative huge comb pattern.
Attached Images
File Type: png 1050.PNG (133.6 KB, 284 views)
File Type: png 1051.PNG (139.6 KB, 282 views)
  Reply With Quote
Old 30th August 2019, 06:42 PM   #2757
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Quote:
Originally Posted by BYRTT View Post
Hi Oabeieo,

Maybe i misunderstand your situation but can remember from past when pos did link to article over Linkwitz site about how summing of multiple band pass especially if band pass gets narrow can lead to nasty summing errors if side band slopes is not calculated into final slope, to show what i mean see below two graphs from VituixCAD and take a close look in block diagram area that when side band filters are bypassed (greyed out) we get huge summing errors seen in the second graph, reason seems to be at 20dB down and below on center the narrow filler we have three transducers in a advanced sum and the small variation in final slopes then output the relative huge comb pattern.


I see what your doing, thatís really cool how to do that.

So you think adding a 20hz hp and a 1k lp also will help?


So the situation is basically this


I have a 3 way up front of car

6nd430- 80-300
2118h 300-1.2
Horn 1.2 up

Than a 18Ē sealed to 80hz

All LR4


Because of the car shape thereís a dip at 80


So adding the 10s under the back seat and using timing tricks and time them against the sub and the midbass to reinforce that null only where I sit

It works, using an allpass on the sub also works but changes things negatively


So Iím trying to get the 10Ē so sound smooth and have it only play 65-110


I tried crosier 100 but there was no output ( the dsp wouldnít pick up the signal it was too low)

I went up in increments of 1hz ( 107,108,109,) and at 110hz audio came back

So now itís just a matter of trying to get it to sound smooth

If I push the crossover up to 160 it sounds just fine but it detracts from my stage playing that high


Maybe itís not possible
  Reply With Quote
Old 30th August 2019, 08:13 PM   #2758
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
So 1st off I super appreciate the help.

There’s no doubt at all my system would not be what it is without the help from this thread.
(I’m serious also, the last 10% made a higher difference thanks to rephase and the 2”speaker trick inside the pods)


So thank you!


Okay so I’m playing with it again right now. I tried adding a 20hz and 1khz side band filters
(If I understood that right) still sounds the same.

Here’s what I noticing. And this is the ticket to making this work right.


I put back on the opendrc and did the driver linearization ( that works better I’m keeping it)


So the 10s in question sound just fine by themself.
It’s the timing between the midbass drivers that make them sound bad


So the timing scheme that I using to fill this null I think is a little by hass effect

Like I’m using the 10s to piggyback on the wavefront from midbass upfront

If I move the delay the bad sound goes away and null comes back ,


I have the 10s like 1ms about s in front of the 6nd430s in the door .
That repairs the null but it’s hassing a tiny bit.


It’s almost as if it’s amplifying all the distortion that is normally buried in the response.

As soon as I turn off the rest of speakers and just let 10s play,again they sound just fine .

This is very unusual phenomenon. Unfortunately it seems the only delay pattern that is filling the hole properly and not changing the way the fronts sound.


I really hope someone knows what I’m talking about and has experienced this before.

If there’s a fir truck I am all ears
  Reply With Quote
Old 31st August 2019, 12:36 AM   #2759
Oabeieo is offline Oabeieo  United States
diyAudio Member
 
Join Date: Jan 2017
Location: Denver
Well completely never mind gang

Pos was right! The whole time I needed to just move the impulse....I couldn’t figure out what he was saying. Than I finally did it. I had to type it in. Oh yes. I’m such a dummy

Took off the opendrc back to just the HD , plenty of taps to do it.
It sounds killer. The wierd noise stopped. It’s working. Sounds smooth with sub. Aghhh yes!

That was it! It wasn’t in time right. I was using too much or too little fir.

Now I’m going to go re do my sub and use this trick on it and see if I can just eliminate the opendrc completely.

Before and after

Ahhh looks soo much better sounds soo much better! I love it it works!
Thank you guys once again!

Click the image to open in full size.

Click the image to open in full size.
  Reply With Quote
Old 7th September 2019, 10:58 PM   #2760
chebum is offline chebum  Poland
diyAudio Member
 
Join Date: Apr 2018
Location: Warsaw
rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool
I have a feeling that FIR-filters "oversharpens" to the sounds. It sounds like particular sounds have contours around them. Similar to what we see on oversharpenned photos from mobile phones. Similar but minimum-phase IIR-filters don't produce these artefacts.

Does anybody hear something similar?

I use rePhase to generate FIR-filters. I suppose I may not be using it correctly. If it's only me hearing the issue, could you check my filter settings? I believe there may be something wrong with my configuration.

rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool-rephase1-jpg

rePhase, a loudspeaker phase linearization, EQ and FIR filtering tool-rephase2-jpg
Attached Images
File Type: jpg rephase1.jpg (510.5 KB, 322 views)
File Type: jpg rephase2.jpg (439.4 KB, 319 views)
  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 01:35 PM
FIR filter design tool for Loudspeaker magnitude equalization ttmusic Software Tools 3 24th May 2013 08:30 PM
FIR Filtering experiences Olombo PC Based 8 10th February 2013 03:45 PM
AVX based FIR VST, crossover / EQ / DRC and delay KOON3876 PC Based 97 26th November 2012 07:18 AM
Phase EQ using FIR filters Grasso Multi-Way 2 2nd July 2003 10:37 PM


New To Site? Need Help?

All times are GMT. The time now is 06:26 PM.


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