Linear Phase Crossovers: What are the benefits? Who makes them? - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #1 of 48 Old 08-05-2009, 09:55 AM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
so like the thread title indicates,

1. what are the benefits to a linear phase crossover?

2. what companies make linear phase crossovers?

is there an alternative device that can "fix" (linearize) the phase of a system after it is in place?

here is a dolby paper on linear phase crossovers:

http://www.dolby.com/uploadedFiles/z...0115%20NYC.pdf

this post was inspired by reports from tom danley and others that the subjective level of "punch" of a system is, in part, related to having all the various sounds arrive at the listener at the same instant (minimal group delay/minimal phase changes) as well as subjective reports that when linear phase crossovers were employed at stag theater (skywalker ranch), they "cleaned up" the sound quite a bit.

Listen. It's All Good.
LTD02 is offline  
Sponsored Links
Advertisement
 
post #2 of 48 Old 08-05-2009, 02:42 PM
Member
 
cc00541's Avatar
 
Join Date: Jan 2006
Location: latitude 40.8510 longitude -96.7592 altitude 362 meters
Posts: 123
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Do a search on 'infinite slope'. Joseph Audio has been doing this passively for years, although I'd think it would be easier today in the digital domain. The bottom line: All the issues inherent with such extreme slopes are in such a narrow passband that they are relatively inaudible.

C

http://www.freepatentsonline.com/7085389.html

Here's another article on fast/slow bass suggests that driver integration is the main causality. I guess this was an issue 10 years ago too...

http://www.soundstage.com/maxdb/maxdb061999.htm
cc00541 is offline  
post #3 of 48 Old 08-05-2009, 04:02 PM
AVS Special Member
 
tvrgeek's Avatar
 
Join Date: Feb 2009
Location: Bowie, MD.
Posts: 2,133
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 10 Post(s)
Liked: 23
Add that the drivers are not linear, it is a bit of a challenge. Linkwitz discusses it a bit in his Orion and Phoenix for the active domain. Gets a tad clumsy in the passive. Digital has it's own problems.
tvrgeek is offline  
post #4 of 48 Old 08-05-2009, 05:53 PM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by cc00541 View Post

Do a search on 'infinite slope'. Joseph Audio has been doing this passively for years, although I’d think it would be easier today in the digital domain. The bottom line: All the issues inherent with such extreme slopes are in such a narrow passband that they are relatively inaudible.

C

http://www.freepatentsonline.com/7085389.html

Here's another article on fast/slow bass suggests that driver integration is the main causality. I guess this was an issue 10 years ago too...

http://www.soundstage.com/maxdb/maxdb061999.htm

Curt, I only skimmed the patent but my understanding of the Joseph Audio XO is it's a variation on the Cauer elliptical theme -- using a notch filter to increase the slope of a standard analog XO immediately above/below the XO frequency. Each of the filters is still minimum phase although the sum isn't. Linear-phase filters are different -- no phase shift as the magnitude changes.

Bruno Putzeys and Siegfried Linkwitz have both done pieces showing why linear-phase crossovers can be 'perfect' on axis but they ring when you move off axis. Bruno is a digital kind of guy (invented the UcD amps) but he favors low-slope analog or IIR (digital equivalent of analog) crossover filters.

Dennis H
catapult is offline  
post #5 of 48 Old 08-05-2009, 07:37 PM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
personally, i am only interested in active crossovers, but if linear phase is possible passively some folks would surely be intereted in that.

Listen. It's All Good.
LTD02 is offline  
post #6 of 48 Old 08-06-2009, 02:05 AM
AVS Special Member
 
vasyachkin's Avatar
 
Join Date: Jan 2008
Location: DIY-AV.NET
Posts: 2,134
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by catapult View Post

low-slope analog or FIR (digital equivalent of analog) crossover filters.

you got it backwards.

http://en.wikipedia.org/wiki/Audio_crossover

Quote:


Active crossovers can be implemented digitally using a DSP chip or other microprocessor. They either use digital approximations to traditional analog circuits, known as IIR filters (Bessel, Butterworth, Linkwitz-Riley etc.), or they use Finite impulse response (FIR) filters. IIR filters have many similarities with analog filters and are relatively undemanding of CPU resources; FIR filters on the other hand usually have a higher order and therefore require more resources for similar characteristics. They can be designed and built so that they have a linear phase response, which is thought desirable by many involved in sound reproduction. There are drawbacks though - in order to achieve linear phase response, a longer delay time is incurred than would be necessary with an IIR filter. IIR filters, which are by nature recursive have the drawback that if not carefully designed they may enter limit cycles resulting in non-linear distortion.

so its IIR that we have in our Behringers. FIR is what they have in DEQX and Dolby Lake processor.

also i recall from reading the writeup on EAW NT speakers ( digital prosound speakers ) that they had to develop a new kind of digital filter that had the benefits of FIR fitlers but didn't have the drawback of TIME DELAY

now the time delay talked about here i believe is IRRELEVANT FOR HOME AUDIO. but in LIVE PERFORMANCE you would hope that the sound comes at the same time as performer moves his lips or strikes the cymbals, so time delay is BAD there.

it seems like you can't get around time delay. you can only have it come evenly at all frequencies ( linear phase ) or have it all mixed into some sort of audio soup ( IIR and Analog ).

with Analog and IIR there is a tradeoff between amount of attenuation ( 6db/oct, 12db/oct, 24db/oct etc ) and phase error ( 90 degrees, 180 degrees, 360 degrees etc ). some designers will say attenuation is more important and go with 48db/oct ( Alesis studio monitors ) and others will say phase is more important and use only 6db/oct ( Dynaudio home speakers ).

but with FIR you just use 300db/oct and no phase error ! problem solved. you just need to shell out a couple grand on the crossover ! ! !

as most digital crossovers are developed for prosound it would explain why very few of them use FIR filters. for a live performance time apparently is more important than phase. even though you probably think its the same thing but it's not !

after all they're not called "zero phase" but LINEAR phase.
vasyachkin is offline  
post #7 of 48 Old 08-06-2009, 02:15 AM
AVS Special Member
 
vasyachkin's Avatar
 
Join Date: Jan 2008
Location: DIY-AV.NET
Posts: 2,134
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by LTD02 View Post

is there an alternative device that can "fix" (linearize) the phase of a system after it is in place?

i am certain such a device could be built. but i doubt it would be cheaper to use a regular crossover and than "fix" it then to just use a FIR crossover in the first place.

a device to linearize phase might be worth it to correct for the INHERENT phase errors that arise due to FINITE bandwidth of any physical speaker.

so if such a device existed i would use it around 20hz and around 20khz to flatten speaker's phase THERE but for a crossover i would just do it right form the beginning.
vasyachkin is offline  
post #8 of 48 Old 08-06-2009, 09:00 AM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by vasyachkin View Post

you got it backwards..... its IIR that we have in our Behringers.

Oops, brain fart, I meant IIR. I edited my post.

Dennis H
catapult is offline  
post #9 of 48 Old 08-06-2009, 09:03 AM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by LTD02 View Post

is there an alternative device that can "fix" (linearize) the phase of a system after it is in place?

http://www.thuneau.com/arbitrator.htm

Dennis H
catapult is offline  
post #10 of 48 Old 08-06-2009, 10:22 AM
AVS Special Member
 
Drew Eckhardt's Avatar
 
Join Date: May 2001
Location: Sunnyvale, CA USA
Posts: 2,674
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 14
Quote:
Originally Posted by LTD02 View Post

so like the thread title indicates,

1. what are the benefits to a linear phase crossover?

Minimum phase is a more accurate term because system amplitude response deviations from flat imply a phase shift and all audio systems have finite bandwidth.

With added delay elsewhere they'll get you imaging between different driver configurations like a WMTW center channel and TM mains and they let marketing departments brag that a square wave going in looks like a square wave coming out on a scope.

This disregards that people can't hear the phase distortion of second order all-pass filters up through LR4. Event the paper you cite says

Quote:


3. APPLICATION OF LPBW TO LOUDSPEAKER ARRAYS

One area of considerable research interest is phase distortion in loudspeaker crossover networks. In previous work, including the references in this paper, the discussion has been focused upon the audibility of phase distortion within a single loudspeaker system. Through subjective and empirical tests, it has been determined that the phase distortion introduced by a conventional crossover network is insignificant.

Reading the paper farther says that this is good in pro-sound setups with different speaker configurations (main and auxiliary). I was thinking of imaging in a home setting; but summed amplitude response being flat would be good too.

The most common analog realization of "linear phase" is a first-order analog cross-over, which still allows excursion to double with each dropping octave, leads to output level limits and/or IM distortion, often precludes using pistonic drivers so the system is always distorting, etc. Those things are all bad.

It also sounds different due to the broader but shallower power response dip about Fc compared to high-order filters which is audible and preferred by some people. The driver choices, counts, cross-over-points and resulting response will obviously be different too.

Quote:


is there an alternative device that can "fix" (linearize) the phase of a system after it is in place?

Yes. At least one of the big room correction boxes (TaCT?) will do it.

Quote:


here is a dolby paper on linear phase crossovers:

http://www.dolby.com/uploadedFiles/z...0115%20NYC.pdf

It's a paper on a specific steep-slope realization which in turn has a limited overlap region and therefore well-behaved polar response (which is audible) and good behavior when different speaker enclosures are summed together. That's good, especially in a pro-sound environment where early reflections are less an issue.

What's missing is how bad the cross-over rings off-axis in the time domain and whether that'd be audible in a home environment.
Drew Eckhardt is offline  
post #11 of 48 Old 08-06-2009, 10:54 AM
AVS Special Member
 
Mark Seaton's Avatar
 
Join Date: Jul 2000
Location: Chicago, IL, USA
Posts: 5,976
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 12 Post(s)
Liked: 188
Quote:
Originally Posted by Drew Eckhardt View Post

What's missing is how bad the cross-over rings off-axis in the time domain and whether that'd be audible in a home environment.

I've seen this mentioned twice here, and something important to remember is that this off axis effect is entirely dependent on the spacing and directivity of the devices being integrated. At lower frequencies and tighter spacings, it won't be a concern.

There are also matters of what we are after where an ideally flat phase response vs. a significant minimization in phase rotation through crossover are very different tasks. The implementation is one part of the discussion, the value of doing so and the trade offs involved will be the other half or more of the discussion.

Mark Seaton
Seaton Sound, Inc.
"Make no little plans; they have no magic to stir men's blood..." Daniel H. Burnham
Mark Seaton is offline  
post #12 of 48 Old 08-06-2009, 12:09 PM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by LTD02 View Post

is there an alternative device that can "fix" (linearize) the phase of a system after it is in place?

I forgot the Pioneer receivers. Their 'full-band phase control' claims to do that when it's enabled. I have no idea how well it works and I haven't seen any independent measurements to show how well it unwraps the phase rotation and cleans up the impulse response. Quoting the manual:

Quote:


The Full Band Phase Control feature calibrates the
frequency-phase characteristics of the speakers
connected.

Standard speakers designed exclusively for audio use
generally reproduce sound with the divided frequency
bands output from a speaker system consisting of
multiple speakers (in case of typical 3-way speakers, for
instance, the tweeter, the squawker (midrange), and the
woofer output sound in the high-, middle-, and lowfrequency
ranges, respectively). Though these speakers
are designed to flatten the frequency-amplitude
characteristics across wide ranges, there are cases
where the group delay characteristics are not effectively
flattened. This phase distortion of the speakers
subsequently causes group delay (the delay of lowfrequency
sound against high-frequency sound) during
audio signal playback.

This receiver analyzes the frequency-phase
characteristics of the speakers by calibrating test signals
output from the speakers with the supplied microphone,
therefore flattening the analyzed frequency-phase
characteristics during audio signal playback1 - the same
correction is made for a pair of left and right speakers.
This correction minimizes group delay between the
ranges of a speaker and improves the frequency-phase
characteristics across all ranges.
Furthermore, the enhanced frequency-phase
characteristics between channels ensure better
surround sound integration for multichannel setting.

http://www.pioneer.eu/files/eur/MCAC...ontrol/top.swf

Dennis H
catapult is offline  
post #13 of 48 Old 08-06-2009, 10:57 PM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
Quote:
Originally Posted by catapult View Post

http://www.thuneau.com/arbitrator.htm

what a pull catapult!

Listen. It's All Good.
LTD02 is offline  
post #14 of 48 Old 08-06-2009, 11:07 PM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
Quote:
Originally Posted by Drew Eckhardt View Post

Minimum phase is a more accurate term because system amplitude response deviations from flat imply a phase shift and all audio systems have finite bandwidth.

With added delay elsewhere they'll get you imaging between different driver configurations like a WMTW center channel and TM mains and they let marketing departments brag that a square wave going in looks like a square wave coming out on a scope.

This disregards that people can't hear the phase distortion of second order all-pass filters up through LR4. Event the paper you cite says



Reading the paper farther says that this is good in pro-sound setups with different speaker configurations (main and auxiliary). I was thinking of imaging in a home setting; but summed amplitude response being flat would be good too.

The most common analog realization of "linear phase" is a first-order analog cross-over, which still allows excursion to double with each dropping octave, leads to output level limits and/or IM distortion, often precludes using pistonic drivers so the system is always distorting, etc. Those things are all bad.

It also sounds different due to the broader but shallower power response dip about Fc compared to high-order filters which is audible and preferred by some people. The driver choices, counts, cross-over-points and resulting response will obviously be different too.



Yes. At least one of the big room correction boxes (TaCT?) will do it.



It's a paper on a specific steep-slope realization which in turn has a limited overlap region and therefore well-behaved polar response (which is audible) and good behavior when different speaker enclosures are summed together. That's good, especially in a pro-sound environment where early reflections are less an issue.

What's missing is how bad the cross-over rings off-axis in the time domain and whether that'd be audible in a home environment.

lot's to chew on there drew, much thanks! nice pull on the tact box. if i find any good whitepapers by them, i'll link them up in this thread.

Listen. It's All Good.
LTD02 is offline  
post #15 of 48 Old 08-06-2009, 11:11 PM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
Quote:
Originally Posted by Mark Seaton View Post

There are also matters of what we are after where an ideally flat phase response vs. a significant minimization in phase rotation through crossover are very different tasks. The implementation is one part of the discussion, the value of doing so and the trade offs involved will be the other half or more of the discussion.

thanks for weighing in ms, can you elaborate a little bit or link up a couple good places to read about this?

Listen. It's All Good.
LTD02 is offline  
post #16 of 48 Old 08-06-2009, 11:21 PM - Thread Starter
AVS Addicted Member
 
LTD02's Avatar
 
Join Date: Oct 2003
Posts: 15,929
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 114 Post(s)
Liked: 840
Quote:

again. nice pull. looks like the folks at pioneer are asking the same questions we are around here or vice versa. ;-)

Listen. It's All Good.
LTD02 is offline  
post #17 of 48 Old 08-07-2009, 02:00 AM
AVS Special Member
 
vasyachkin's Avatar
 
Join Date: Jan 2008
Location: DIY-AV.NET
Posts: 2,134
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by LTD02 View Post

again. nice pull. looks like the folks at pioneer are asking the same questions we are around here or vice versa. ;-)

pioneer is a monster on the technology side but they don't know how to package it

for example my pioneer car head unit had a microphone and a multi band equalizer and was supposed to calibrate speaker's response.

too bad i was using a system with big amplifiers and if i let it run its test tones it would probably blow all of my speakers up. so i never used it. i just set equalizer by ear.

that head unit also had an organic LED display - which obviously i didn't need.

pioneer is all about putting state of the art technology into useless products ( how japanese of them )

this would be in contrast to say APPLE which puts middle-of-the-road technology into products people actually want.
vasyachkin is offline  
post #18 of 48 Old 08-07-2009, 02:17 AM
AVS Special Member
 
vasyachkin's Avatar
 
Join Date: Jan 2008
Location: DIY-AV.NET
Posts: 2,134
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by catapult View Post

http://www.thuneau.com/arbitrator.htm

this technology actually is important.

while crossovers introduce MOST of the phase errors the transducers themselves also introduce plenty.

in my response to LTD i mentioned that this technology could be used to correct phase response at the fringes ( 20hz and 20khz ) but that's not all it should be used for.

each driver has its own finite bandwidth ( aside from the overall finite bandwidth of the sound system ) which results in phase errors. these should be fixed ( driver by driver ) using technology such as this thuneau.

after phase response of each driver has been flattened then the signal can be sent to to a linear phase crossover and that's the only way you can ensure overall linear phase performance of the sound system.

i mean you must have linear phase crossover AND linear phase drivers.

whether driver phase can be linearized in a passive system as pioneer claims to do ? i don't know. to me it seems unlikely that some DSP can take misaligned signals from several transducers and recombine them into a single signal. and even if this can be done theoretically i dont know if in practice it would do more harm than good.
vasyachkin is offline  
post #19 of 48 Old 08-07-2009, 04:45 AM
AVS Special Member
 
Kevin Haskins's Avatar
 
Join Date: Feb 2005
Posts: 1,008
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
It is marketing mainly. Anytime you see companies making a big deal out of linear phase you can almost smell the BS.

You have to recognize that transducers are not linear phase devices to begin with. You also have to recognize that most tweeter-midwoofers are acoustically out of alignment and that they can only be time aligned, even with a DSP, for one position in space since the drivers are acoustically separate. Even a single driver loudspeaker, without a tweeter, will have time alignment mismatch (the high freq arrives slightly time aligned differently than the midrange/bass). In most cases, this is completely inaudible.

Time alignment & linear phase may mean different things although you see a lot of mixing of the terminology. You can have a smooth phase response through the crossover even though the tweeter-midwoofer are out of time alignment. Often that is the case as the typical dome tweeter on a baffle is 100-150us out of alignment with the midwoofer because the Voice Coils are at two different locations on the Y-axis of the loudspeaker (the tweeter VC is closer to the listening posn by an inch or two). The research shows this is pretty much inaudible. Correcting it buys you very little that is audible but it does give you another marketing checklist item.

Horn speakers are another matter. You can get path length differences of many inches, and in those cases DSP correction is the only method to bring them into some reasonable time alignment. You can still design passive networks that give reasonable FR behavior but there is nothing you can do with a passive network that will delay the signal. The exception is using allpass filters with 0-180 phase shifts but flat amplitude response to compensate for some of the time-delay. These require picking the correct amount of phase shift for a given crossover point because they don't delay all of the signal equally. Also, they are only really suitable for modest amounts of delay compensation. They work for baffle mounted dome tweeters, but not long path length differences you see in horns.

The bottom-line. Go with a good speaker designer and trust their work. You cannot just throw transducers together on a baffle, hook up an external crossover, linear phase or otherwise and get good results.

Kevin Haskins
Exodus Audio
Kevin Haskins is offline  
post #20 of 48 Old 08-07-2009, 05:34 AM
Member
 
findbuddha's Avatar
 
Join Date: Feb 2009
Location: Brisbane, AUS
Posts: 174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Here's a post explaining Hypex's take on the situation, to be featured in their upcoming DSP modules: http://www.diyaudio.com/forums/showt...73#post1833373


There are also several threads on DIYaudio about using a PC as crossover and phase/amplitude correction.
findbuddha is offline  
post #21 of 48 Old 08-07-2009, 11:07 AM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by findbuddha View Post

Here's a post explaining Hypex's take on the situation, to be featured in their upcoming DSP modules: http://www.diyaudio.com/forums/showt...73#post1833373


There are also several threads on DIYaudio about using a PC as crossover and phase/amplitude correction.

Yeah, that's by Bruno Putzeys who I mentioned above. His slide show is an interesting read. Those new Hypex amp/DSP modules look sweet for DIYers. Hopefully they'll start building them with US power supplies sometime.

Dennis H
catapult is offline  
post #22 of 48 Old 08-07-2009, 01:29 PM
Advanced Member
 
jj_0001's Avatar
 
Join Date: Sep 2006
Location: Kirkland, Wa
Posts: 653
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
The last AES Journal had a paper on audibility of high-order constant-delay crossovers.

It would be worth reading. The summary is that if you use too sharp a filter, you will get into trouble (only) moderately off axis.

If you're lucky.

Using constant-delay crossovers is not a panacea. I don't find this surprising.

James D. (jj) Johnston
jj_0001 is offline  
post #23 of 48 Old 08-07-2009, 05:53 PM
Member
 
findbuddha's Avatar
 
Join Date: Feb 2009
Location: Brisbane, AUS
Posts: 174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by catapult View Post

Yeah, that's by Bruno Putzeys who I mentioned above. His slide show is an interesting read. Those new Hypex amp/DSP modules look sweet for DIYers. Hopefully they'll start building them with US power supplies sometime.

The standalone DSP module will run off +/- 12V volt (or 15v, Jan-Peter wasn't sure when I asked)

I'm probably going to run it off the Aux out of a Red Rocks SMPS

findbuddha is offline  
post #24 of 48 Old 08-07-2009, 06:44 PM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
I read further in the thread you posted and Jan-Peter said you can order the plate amps with a 120V/60Hz power supply. Just specify it in the comments when you order.

Dennis H
catapult is offline  
post #25 of 48 Old 08-08-2009, 04:29 AM
AVS Special Member
 
vasyachkin's Avatar
 
Join Date: Jan 2008
Location: DIY-AV.NET
Posts: 2,134
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
can anybody explain what you people mean by running into problems off axis ?

wouldn't you run into the same problems with a regular crossover too ?
vasyachkin is offline  
post #26 of 48 Old 08-10-2009, 10:58 AM
Advanced Member
 
jj_0001's Avatar
 
Join Date: Sep 2006
Location: Kirkland, Wa
Posts: 653
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by vasyachkin View Post

can anybody explain what you people mean by running into problems off axis ?

wouldn't you run into the same problems with a regular crossover too ?

It is possible to design, using a modern DSP, a crossover of almost any slope you want. The problem, off-axis (i.e. up, down, left, right or whatever so that the two drivers being crossed over are no longer at the same distance, which of course depends on the driver arrangement) is that the "sum to 1" property does not hold when you add delays to the sum of the two filtered signals.

Drivers aren't entirely zero-phase or linear, either, of course, and this can cause a problem even when you are on axis.

The moral is that it's not smart to make a crossover TOO steep.

When you use FIR's you can get really wierd pre-echos and ringing patterns resulting from ugly frequency response about the crossover point.

When you use IIR's you can get similar problems, but they don't pre-echo, in general, they only have ringing after the main lobe of the resulting filter response.

I've made some plots to explain:

A pretty outrageous lowpass (woofer) filter (note, this plot is a bit odd, I should have plotted the frequency response using more fft points but it is an equiripple filter when you plot it with enough points): http://s238.photobucket.com/albums/f...current=lp.jpg

The matching highpass filter:
http://s238.photobucket.com/albums/f...current=hp.jpg

The sum when they are time-aligned and there is no problem with drivers:

http://s238.photobucket.com/albums/f...urrent=sum.jpg

The sum when one speaker is .3" farther away than the other:
http://s238.photobucket.com/albums/f...t=delaysum.jpg

And the impulse response of that filter (.3" delay)
http://s238.photobucket.com/albums/f...t=impdelay.jpg

As you can see, this small amount of delay is bad. More would in fact be much worse.

James D. (jj) Johnston
jj_0001 is offline  
post #27 of 48 Old 08-10-2009, 12:41 PM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by jj_0001 View Post

The last AES Journal had a paper on audibility of high-order constant-delay crossovers.

It would be worth reading. The summary is that if you use too sharp a filter, you will get into trouble (only) moderately off axis.

If you're lucky.

Using constant-delay crossovers is not a panacea. I don't find this surprising.

I'm not an AES member but the abstract of that paper makes it sound like you can still go pretty darn steep with FIR filters before they start sounding bad. (bold emphasis mine)

Quote:
Perceptual Study and Auditory Analysis on Digital Crossover Filters

Authors:Korhola, Henri; Karjalainen, Matti
Affiliation:Helsinki University of Technology, Department of Signal Processing and Acoustics, Espoo, Finland
Page:413

The extensive research on the perceptual attributes of analog filters used for loudspeaker crossover networks does not necessarily apply to digital filters. In this study finite-impulse response (FIR) and Linkwitz-Riley (LR) digital crossover filters were examined for their perceptual artifacts. Subjective tests with headphones and loudspeakers showed that for LR filters the audibility of phase distortion can be predicted by group delay errors. But FIR filters of high order produce audible artifacts because of time smear created by extensive ringing. LR filters of order 8 or less and FIR filters of about 600 were without problems. These safety limits should be respected.


Dennis H
catapult is offline  
post #28 of 48 Old 08-10-2009, 01:07 PM
Advanced Member
 
jj_0001's Avatar
 
Join Date: Sep 2006
Location: Kirkland, Wa
Posts: 653
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by catapult View Post

I'm not an AES member but the abstract of that paper makes it sound like you can still go pretty darn steep with FIR filters before they start sounding bad. (bold emphasis mine)

Well, 600 taps isn't that steep. Don't forget, an FIR filter length is equivalent to the entire length of the IIR response in terms of total energy.

My example used a 1K filter. People have proposed much sharper. I didn't even try to figure out what would happen with a much sharper filter, I think 'wrong' is the appropriate way to put it.

Ok, I tried a 4097 length filter. The results for one sample delay (.3 inches at 44.1kHz) are just too wierd to contemplate.

James D. (jj) Johnston
jj_0001 is offline  
post #29 of 48 Old 08-10-2009, 01:16 PM
AVS Special Member
 
catapult's Avatar
 
Join Date: Sep 2001
Posts: 3,824
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Thanks James, I misunderstood. I thought he meant 600th order not 600 taps. About how steep in orders or dB/octave does 600 taps turn out to be?

Dennis H
catapult is offline  
post #30 of 48 Old 08-10-2009, 02:26 PM
Advanced Member
 
jj_0001's Avatar
 
Join Date: Sep 2006
Location: Kirkland, Wa
Posts: 653
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by catapult View Post

Thanks James, I misunderstood. I thought he meant 600th order not 600 taps. About how steep in orders or dB/octave does 600 taps turn out to be?


First, it's "jj". Please.

Well, for FIR's, the order of the filter (in terms of zeros) is the number of taps minus 1, so a 600 tap filter is a 599 order filter.

To answer your second question, it's not so easy. The answer is in terms of transition bandwidth, which is itself in terms of fs/2 being '1'.

This is why low-frequency crossovers are "interesting" digitally, and higher-frequency crossovers are "interesting" for different reasons

In order to match a 100Hz 3rd order butterworth, one would need quite a long filter.

In order to match a 3rd order butterworth at 10kHz (for 44.1 sampling rate) would require a much shorter filter. Of course, also "match" is not entirely fair, you'd have to "match" "how fast did I get to -n dB" because the filter shapes are most likely to be extremely different.

To give you an idea. a 600 tap (599th order) FIR with a transition start band at 100 Hz and stop band starting at 200Hz has in-band ripple of about .3 dB, and rejection of around -28dB. Above 200Hz, the filter is equiripple, i.e. there are many peaks coming back up to -28dB. This is not a very good filter, frankly, for a crossover.

Let's try the same 600 tap filter for 1 octave at 1khz now,cutting off at 2kHz... The in-band ripple is minescule (smaller than double-precision!) and the rejection of the filter is in excess of 180dB. (More importantly my optimization program gave up as "um, I hope this is good enough, sport, I'm not written in quadruple precision...")

So, the comparison of FIR to IIR is not so simple. an IIR in analog will have a fixed slope/octave or decade. An FIR will have an out-of-band rejection, an in-band ripple, and a transition band from one to the other.

Combinations are possible, of course.

At 300Hz the ripple and rejection are reasonable, the ripple is about .004dB and the rejection about 70dB, which is reasonably good to avoid driver interactions.

I was going to try a filter working from 10khz to 20khz, but I can't even do a 600 tap filter there.

The point is that if the transition band is 100Hz, 600 taps isn't enough at 44.1khz samplgin rate. If the transition band is 300Hz (it doesn't matter 300 Hz or 3khz as the start of the stop band), it's just enough. If the transition band is 1kHz, it's too long. 10kHz? Fergetaboutit in this universe.

To explain, if I make a crossover at 1kHz, and for which the end of the HP filter transition band is 1300Hz, I get very much the same filter performance (in terms of ripple and rejection) as I do for the 300/600Hz filter. (checking my own assertion, the two filters have equal ripple and stop-band rejection to better than .1dB stop band and 1% difference in passband)

So you can't treat FIR and IIR's the same way when you think about them. With IIR's it's dB/octave, with FIR's it's "absolute width of transition band divided by fs/2"

Hope this helps.

James D. (jj) Johnston
jj_0001 is offline  
Reply DIY Speakers and Subs

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

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