FIR/Brick Wall Crossovers - diyAudio
Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

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 11th January 2012, 09:20 PM   #1
diyAudio Member
 
rvrazvan's Avatar
 
Join Date: Jul 2010
Default FIR/Brick Wall Crossovers

I would like to start a discussion regarding all the aspects of FIR type crossovers.

Some basic info:
Finite impulse response - Wikipedia, the free encyclopedia
http://dolby.invisionzone.com/index....ch&attach_id=8
Part II: Digital Filters


Subjects of debate:

-Advantages/disadvantages
-Implementation
-Sound QUALITY

Opinions?


We all know this is the future, and the lack of condensed information is not helping anyone.
  Reply With Quote
Old 29th December 2013, 01:01 PM   #2
diyAudio Member
 
Join Date: Jan 2013
Bump! Not one response?
Well... fwiw, I'm just getting to the point where I believe that all my media content, and 99.999% of that consumed globally is going to be digital. If we presume this, then the obvious direction is digital amplification and near brick wall digital crossovers, which should decrease costs, lower power consumption and dramatically improve sound quality.
So my first question is can a brick wall xo from C1-C2Hz be developed? The answer would appear to be yes, with advantages including 0 distortion at the digital level and "perfect" phase alignment... however if such a dsp would require fir iir is debateable and will need a bit more research... I guess this is what you're referring to though?
I have to say though, I do wonder on the lifespan of the loudspeaker driver compared to bmr or magneplaner type technologies...
  Reply With Quote
Old 29th December 2013, 01:52 PM   #3
OllBoll is offline OllBoll  Sweden
diyAudio Member
 
Join Date: Nov 2010
Bruno Putzeys at Hypex advocates that you use FIR for phase correction only:

Use IIR to get the drivers flat, then apply the crossover slopes ( also IIR ) you want and then use FIR inverse all passes to fix the phase. It's cheaper on taps and works well.
  Reply With Quote
Old 29th December 2013, 02:01 PM   #4
diyAudio Member
 
Join Date: Apr 2005
Location: -
Quote:
Originally Posted by nannoo View Post
Bump! Not one response?
Well... fwiw, I'm just getting to the point where I believe that all my media content, and 99.999% of that consumed globally is going to be digital. If we presume this, then the obvious direction is digital amplification and near brick wall digital crossovers, which should decrease costs, lower power consumption and dramatically improve sound quality.
So my first question is can a brick wall xo from C1-C2Hz be developed? The answer would appear to be yes, with advantages including 0 distortion at the digital level and "perfect" phase alignment... however if such a dsp would require fir iir is debateable and will need a bit more research... I guess this is what you're referring to though?
I have to say though, I do wonder on the lifespan of the loudspeaker driver compared to bmr or magneplaner type technologies...
A perfect brickwall filter is trivial to implement: gather full digital signal (e.g. a song), perform FFT, set all values for frequencies above/below your cut-off to zero, perform inverse FFTs to get a pair of perfect brickwall filtered signals. This will not result in a dramatic improvement in sound quality for a speaker crossover. The problems with such "perfection" are well documented.
  Reply With Quote
Old 29th December 2013, 02:17 PM   #5
fpitas is offline fpitas  United States
diyAudio Member
 
Join Date: Apr 2011
Location: Leesburg, VA
I'm not sure the brick wall approach will ever work well, except for drivers that are more or less coaxial, or in an MTM arrangement. The lack of overlap makes transitions between drivers abrupt, and not blend well at the crossover. Perhaps if you're far enough away from the speaker, or in a particularly reverberant room, it may work out.
__________________
Francis
  Reply With Quote
Old 31st December 2013, 09:26 AM   #6
diyAudio Member
 
Join Date: Apr 2007
Brickwall filters are a good way to get the vertical directivity clean. In usual speakers with a vertical distribution of drivers and flat filters (up to 4th order) the directivity of the drivers may match but there're still interferences in the vertical plane. Use a brickwall filter instead and they're nearly gone. It doesn't improve the sound that much, but don't we want it perfect?

But these filters allow other driver placements. Think about line arrays, where the midrange and tweeters are placed side by side and have intereferences in the vertical plane, limiting the sweet spot. Use a brickwall filter and the problem is gone.

Also, FIR filters allow easy automated EQ'ing of drivers. With IIRs you either do it by hand or use a software which needs a lot of iterations to get the optimal frequency response - and it is still not as perfect as with a FIR. A 512 taps FIR provides a good compensation down to 200 Hz. Below the use of IIRs is more clever, because there's usually the modal region in the room and the FIRs filter will become to long introducing delays and high computational costs.
  Reply With Quote
Old 31st December 2013, 11:14 AM   #7
diyAudio Member
 
Join Date: Apr 2005
Location: -
Quote:
Originally Posted by Baseballbat View Post
But these filters allow other driver placements. Think about line arrays, where the midrange and tweeters are placed side by side and have intereferences in the vertical plane, limiting the sweet spot. Use a brickwall filter and the problem is gone.
The brickwall filter is not usually considered appropriate for crossovers. The sharp step in frequency results in a broad "ringing" in time. It has a large component at the crossover frequency and is symmetrical. This means that during a passage of silence before a transient both the tweeter and midrange cones will start moving in a manner that is supposed to sum to zero. This is not possible for drivers of different sizes located a distance apart. This deficiency is audible particularly as the brain extracts a lot of information from the initial part of an event/note.

Quote:
Originally Posted by Baseballbat View Post
Also, FIR filters allow easy automated EQ'ing of drivers. With IIRs you either do it by hand or use a software which needs a lot of iterations to get the optimal frequency response - and it is still not as perfect as with a FIR. A 512 taps FIR provides a good compensation down to 200 Hz. Below the use of IIRs is more clever, because there's usually the modal region in the room and the FIRs filter will become to long introducing delays and high computational costs.
I would agree that FIR filters are useful for a range of audio processing. If you want to use FIR filters at low frequencies then you should transform your time series to a lower sampling rate, perform the filtering operation and transform back. How the high frequency information in the signal is handled will depend on the operation. If you use common IIR filter approaches at bass frequencies the total group delay including the driver and cabinet may become audible.
  Reply With Quote
Old 31st December 2013, 11:46 AM   #8
diyAudio Member
 
Join Date: Apr 2007
Quote:
Originally Posted by andy19191 View Post
The brickwall filter is not usually considered appropriate for crossovers. The sharp step in frequency results in a broad "ringing" in time. It has a large component at the crossover frequency and is symmetrical. This means that during a passage of silence before a transient both the tweeter and midrange cones will start moving in a manner that is supposed to sum to zero. This is not possible for drivers of different sizes located a distance apart. This deficiency is audible particularly as the brain extracts a lot of information from the initial part of an event/note.
Yes, this can be true, but not with short filters. As I wrote already 512 taps* provide good results down to 200 Hz, and the little preringing of them is really irrelevant.

Quote:
I would agree that FIR filters are useful for a range of audio processing. If you want to use FIR filters at low frequencies then you should transform your time series to a lower sampling rate, perform the filtering operation and transform back. How the high frequency information in the signal is handled will depend on the operation.
Which is computationally very costly and introduces high delay (like every long linear-phase FIR). The latter is no problem with music, but can be problematic with real-time applications. Think in hundreds of milliseconds.

Quote:
If you use common IIR filter approaches at bass frequencies the total group delay including the driver and cabinet may become audible.
Yes, of course. But think about this: cross at 200 Hz with a 512 taps FIR, and equalize below that with IIRs. That's what I was talking about. Group delay? No problem! Computational costs? Very low. That's the clever way to do it.

*That's an approx 11 ms long impulse response at 48 kHz, with higher sampling rates you'll need longer filters, of course, but the pre-ringing still keeps irrelevant.
  Reply With Quote
Old 31st December 2013, 01:11 PM   #9
diyAudio Member
 
Join Date: Apr 2005
Location: -
Quote:
Originally Posted by Baseballbat View Post
Yes, this can be true, but not with short filters. As I wrote already 512 taps* provide good results down to 200 Hz, and the little preringing of them is really irrelevant.
I am not sure how many would agree with you particularly as methods to avoid it are well developed. It is easy enough to check for those that are interested.

Quote:
Originally Posted by Baseballbat View Post
Which is computationally very costly and introduces high delay (like every long linear-phase FIR). The latter is no problem with music, but can be problematic with real-time applications. Think in hundreds of milliseconds.
It is not computationally costly. The point of the approach is to avoid the high cost of using a huge number of taps while still enabling the full and effective manipulation of low frequencies using FIR filters.

Quote:
Originally Posted by Baseballbat View Post
Yes, of course. But think about this: cross at 200 Hz with a 512 taps FIR, and equalize below that with IIRs. That's what I was talking about. Group delay? No problem! Computational costs? Very low. That's the clever way to do it.
A group delay at low frequencies that is large enough to be audible would be considered a problem by many. Not sure about clever but the approach you outline is not an unreasonable one. However it may introduce audible artifacts that alternative approaches may not.
  Reply With Quote
Old 31st December 2013, 02:22 PM   #10
diyAudio Member
 
Join Date: Apr 2007
Quote:
Originally Posted by andy19191 View Post
It is not computationally costly. The point of the approach is to avoid the high cost of using a huge number of taps while still enabling the full and effective manipulation of low frequencies using FIR filters.
You need a decimation and probably an interpolation filter (depends on the used output sample rate of the filtered channel), the reduction in computational cost compared to a plain approach is not that high.

Quote:
A group delay at low frequencies that is large enough to be audible would be considered a problem by many. Not sure about clever but the approach you outline is not an unreasonable one. However it may introduce audible artifacts that alternative approaches may not.
EQ'ing with IIRs does not necessarily create more group delay. Think of EQ'ing room modes, they HAVE high group delay, after EQ'ing it is not so high.
  Reply With Quote

Reply


Hide 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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
DIY In-{1/2m-Brick}-Wall (HT) Speakers !!! Medenyx Multi-Way 21 26th May 2011 07:31 AM
Another brick in the wall Supersasha Introductions 6 16th February 2008 08:53 PM
in-wall/on-wall speaker? (axiom w-22 clone?) Scott_fx Multi-Way 0 22nd February 2007 03:08 AM
Brick wall schematic wanted! e96mlo Analogue Source 11 8th March 2003 12:11 AM


New To Site? Need Help?

All times are GMT. The time now is 01:10 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2