FILTER

[라디오]

//this sets W3C DHTML specification values used in NN6 & IE5.5 if (document.getElementById) { visible = 'visible'; hidden = 'hidden'; //this sets Netscape values } else if (document.layers) { visible = 'show'; hidden = 'hide'; //this sets IE values } else if (document.all) { visible = 'visible'; hidden = 'hidden'; } var lastItem = 'div0'; function showMenu(item) { if (document.all) { hideLast = document.all[lastItem].style; showItem = document.all[item].style; } else if (document.layers) { hideLast = document.layers[lastItem]; showItem = document.layers[item]; } else if (document.getElementById) { hideLast = document.getElementById(lastItem).style; showItem = document.getElementById(item).style; } hideLast.visibility = hidden; showItem.visibility = visible; lastItem = item; }  

Translate PSW!

	About | Advertising | Ken's Links | News | E-News | Industry Biz | Chat | Classifieds | Calendar | Fun | Site Search | Forum Search

	Articles | Old Soundman | Forums | Best Threads | Industry Biz | Syn-Aud-Con | NSCA | Live Sound Int.

	Articles | Mixerman Diaries | Strictly Tech | Forums | Best of Rec Pit | DAW Nation | Tape Op | Industry Biz

	Hands On | Resource | Project Profiles | Church Sound Forums | Church Sound Magazine | TFWM Archives

	LAB | LAB Lounge | R/E/P | MARSH | Install | Lighting | Subwoofer Project | Church Sound | Networking | Chat

	Audio Basics | Measurement | Business | Cabling | PCs & Networks | History | Wireless | Recording | Stagecraft | SR Tech | SAC

Translate this site with Google! Copy the url above, click on Translate PSW! and paste into the "translate a web page" form on the next page. Then select your language option from the drop down list (English to . . . )

Welcome to the ProSoundWeb sound reinforcement forums, home of more than 50,000 audio professionals.
Important Notice: Full Names are required to post in these forums; please put your full name in the alias field in your profile.

Search SR Forum

Advanced Search

LAB Archives

Church Sound Forums

R/E/P Forums

MARSH Forum

Networking Forums

Study Hall

Ken's Pro Audio Links

Breaking Industry News

Search   FAQ    Register    Login   Home

Home » SR: LAB » LAB: The Classic Live Audio Board » Anyone knows a Pro FIR filter Speakerprocessor?

Show: Today's Messages  :: Show Polls :: Message Navigator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #72716 is a reply to message #72581 ]

Wed, 31 August 2005 18:36

Peter James Messages: 179 Registered: April 2004

Has No Life

Hi Jon,

A couple of points, when you reach the HF limit of your compression driver and its output rolls of sharply the phase is going to take a dive, it has to by definition, and there is no point trying to fix it, there isn’t much output to hear in any case! The same thing will happens to the LF, but the plot I posted the LF was on an XTA 226 using normal filters.

I should also point out there was no attempt to correct for the phase with the above plot, it just used FIR filters and EQ to correct for the driver response and time alignment for their relative position.

Have I AB it on the same system – yes, were both systems flat – yes, did it sound better – yes, to everyone - noticeably. The sound is more natural; it has greater depth, finer detail and appears to come more from the stage than the speakers. To me the sound is less harsh; perhaps this is due to the improved impulse response, less ringing in the filters, maybe the higher roll off rate to the HF (don’t really know). I could go into a lot more detail but have a read of the White Paper Lake have on their site. I posted the link above.

I guess the bottom line is – there was nothing I could do that would make the systems sound as good as it did on the Lake.

The only negative thing I noticed (if you can call it that) was comb filtering. With a LR stack as the sound became more accurate the comb filtering became more noticeable as you walk across the playing field. By definition this is exactly what you would expect, it maybe not what you want but the laws of physics define that this is exactly what will happen.

What I would really love to do is to be able to cancel the horn reflection Dave Guinness style….


Peter

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #72732 is a reply to message #72706 ]

Wed, 31 August 2005 19:30

Mike Babcock Messages: 420 Registered: April 2004 Location: near the beach

Has No Life

Peter wrote on Wed, 31 August 2005 18:04

Oooops – sorry abou that, if you just click on the link it should down load and auto size.

Peter

Sure when I click on the link it becomes a small picture.
But in the thread it is 4 times wider than my screen and
makes the thread unreadable unless I scroll horizontal
as I read. Please edit your post and remove the picture,
but leave the link. I would like to read this thread as it
appears very interesting.

I have cut my sentences short to force this message to be
read on monitors smaller than 60" wide.

Mike

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #72767 is a reply to message #72732 ]

Thu, 01 September 2005 00:07

Peter James Messages: 179 Registered: April 2004

Has No Life

Thanks - didn't know I could change it.

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #72871 is a reply to message #72395 ]

Thu, 01 September 2005 11:49

Jon Waller Messages: 34 Registered: April 2004 Location: Michigan

Should Get Out More

Many thanks to Fred and Peter for the interesting discussion. I have begun to check out the links that Fred posted. It appears there is much more written about this than I was aware of. Last night I went home and in my mail was the latest AES journal (2005 July/August). It has a paper by Flanagan, Moore, and Stone about the audibility of group delay. They find that under nearly anechoic conditions, group delay is audible for clicks (impulses), but they conclude "These results suggest that the nonlinear phase response of loudspeakers is unlikely to have any audible effect in typical rooms". So I still don't know what to think, but will keep an open mind. But one of you made a comment about the ringing of the impulse response of a non-linear phase filter. Linear phase FIR filters, especially ones with very steep rolloffs, have massive amounts of time-domain ringing, not only after the main impulse, but also before it. Granted, if the transducers were perfect, these oscillations of the low-pass and high-pass would largely cancel out, but I am still not convinced that their residuals would not have negative audible consequences of their own.

My sympathies and prayers for those down south affected by the hurricane.

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #72981 is a reply to message #72871 ]

Thu, 01 September 2005 18:24

Peter James Messages: 179 Registered: April 2004

Has No Life

Yes FIR filter will ring "Gibbs effect" as you described but in practice there are ways of dealing with. Im not into DSP programming but if you use a Hamming window the ringing is much better, however the filter will not be quite as steep. Have a look at this impulse response, which compares a Butterworth with a FIR.



While I think there is only a very small improvement with a flat phase response it’s the sum of all of the other things that add up to a noticeable difference

· Improved phase response
· Improved impulse response
· Improved lobing errors etc.
· Good A/D’s with 96K sampling rate
· Less Xmax problems on the HF driver etc

Peter

• Attachment: Digital_filter_impulse_compare[1].gif
  (Size: 42.83KB, Downloaded 265 time(s))

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73117 is a reply to message #72981 ]

Fri, 02 September 2005 08:56

Jon Waller Messages: 34 Registered: April 2004 Location: Michigan

Should Get Out More

Sorry to keep bugging you, Peter. That plot is interesting. Where did you get it? As someone who has designed many an FIR filter (using commercial software), I find it hard to believe that the FIR impulse responses shown here, with only one oscillation or sidelobe, have very good frequency domain performance. A brick-wall filter would have many, many oscillations, and the weighting would not have much of an effect on the ones closest to the center.

A second question I would like to ask, is the plot you furnished of the KF750 with FIR xover of just a single cabinet? If not, how many cabinets where on for the measurement? And how many were on during your A/B listening comparisons? One possible explaination for some of the improvement you are hearing is as follows. The center horn on a KF750 has a mouth that I estimate is about 12 inches wide. The coverage angles of this horn are spec'ed at 35 degrees, but as the frequency gets lower, the coverage angle broadens out due to the physics of diffraction. If the crossover frequency is 1.5 KHz (an educated guess), with a Linkwitz-Riley crossover, there is still significant contribution from this horn down to 1 KHz or less, but with a brick wall filter, the output at this frequency would be much lower amplitude. But this is where the horn coverage really broadens out, so that any listening position will really hear contributions from mutiple horns. If the boxes are arrayed in an arc, as designed, the horns will have different path lengths to your ears, and therefore different arrival times, possibly smearing the impulse response more than the effect of the Linkwitz-Riley non-linear phase response does! So the brick wall filter could really help clean up the midrange response of a multi-box array both in the time domain and the frequency domain (less interaction of boxes, less amplitude ripple). But note that this improvement is brought about by the better, more ideal amplitude response of these filters, not the fact that they are linear phase!

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73213 is a reply to message #73117 ]

Sat, 03 September 2005 01:39

Peter James Messages: 179 Registered: April 2004

Has No Life

I don’t remember exactly where picture came from. It maybe on this site somewhere

http://www.mathworks.com/access/helpdesk/help/toolbox/signal /filterd8.html

I should also point out that the brick wall filters on the 750 plots are something like 48 & 78dB per oct. Low - Mid / Mid- Hi respectively not that steep by FIR standards.

From memory the impulse response with the lake was improved compared to what I found with other DSPs.

One of the problems with the HF on the 750 is that the filters ring at just under 7K, interestingly enough EAW put a - 4dB PEQ @ 6K73 . Rightly or wrongly I always assumed that was there to disguise that problem. It did make them sound sweeter.

The other thing to consider is at what frequency the ringing occurs – if it above 15K who really cares.

I used two boxes for the plot – but essentially it’s the results of one box at about 6 feet as they don’t interact that much.

You are absolutely correct about how it effects the interaction between cabinets. This was evident when we used them with the Lake for the first time. The program I developed in the work shop was almost spot on, normally when I developed programs for the 750s like that they needed quite a bit of modification.

Regards


Peter

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73224 is a reply to message #72981 ]

Sat, 03 September 2005 07:49

Fred Blatt Messages: 45 Registered: April 2004

Should Get Out More

I'll agree with Jon on this one. That impulse response graph is useless as a comparison without a corresponding frequency response graph to go along with it. You can make a linear phase impulse response have sidelobes as small as you want if the frequency response transitions aren't very steep. It would also be instructive to have the impulse response graph shown with the amplitude in dB instead of linear (i.e. volts).

Fred

[Updated on: Sat, 03 September 2005 08:19]

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73284 is a reply to message #73224 ]

Sat, 03 September 2005 18:32

John Roberts {JR} Messages: 2967 Registered: April 2004 Location: MS

Has No Life

FreddyEddy wrote on Sat, 03 September 2005 07:49

I'll agree with Jon on this one. That impulse response graph is useless as a comparison without a corresponding frequency response graph to go along with it. You can make a linear phase impulse response have sidelobes as small as you want if the frequency response transitions aren't very steep. It would also be instructive to have the impulse response graph shown with the amplitude in dB instead of linear (i.e. volts).

Fred

This may be another case where we need to trust our ears instead of our eyes.

It might be useful to look at "Gibbs" phenomenon (the fancy name for pre-ring etc) for the specific case of square waves. Fourier teaches that a square wave is the sum of a series of (odd) harmonically related sine-waves. If we look at this a bit backwards (one of my specialties) a band limited or LPF'd square wave will be this sum of harmonics but with the higher overtones (sine waves) scraped off. Subtracting the upper overtone will "look" like a negative version of that missing higher frequency sine wave superimposed on a perfect (fully loaded) square wave. The lower the order or frequency of these missing overtone the larger the apparent "ringing" or deviation from sharp edged square wave.

While the whole concept of step or impulse response from the low or mid outputs of a crossover is worthy of more inspection, it's easy to visualize that a few hundred Hz bandpass would leave huge gaps in an actual square wave or step. The apparent ringing may just be the ideal result of the limited bandwidth, and the visual error is just the missing higher frequencies that shouldn't be there anyhow.

Of course I could be wrong....

JR

---

Resotune at http://circularscience.com/

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73305 is a reply to message #73284 ]

Sat, 03 September 2005 20:53

Peter James Messages: 179 Registered: April 2004

Has No Life

Unfortunately as I said before I’m not really in a situation to comment FIR algorithms etc.

The Lake does sound better; I did measure the impulse response and thought it was better than I was getting with IRR filters.

I guess the answer is to measure the Lake with an FIR and IRR filter and compare the results. When I have time I will try to do that and post the results.

I must also apologize, I checked the filter slopes - @ 1.5K its about 50dB @ 170 Hz its only about 15dB!!! What’s interesting is at 1K3 its about 80 dB as you go higher it drops to 50 then as you go higher it gradually increases to 80 the drops again etc.

You can also implement 24dB LR linear phase filters if you like.

Peter

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73418 is a reply to message #73117 ]

Sun, 04 September 2005 19:44

Peter James Messages: 179 Registered: April 2004

Has No Life

found it -

http://www.silcom.com/~aludwig/Signal_processing/Digital_fil ters.htm#Impulse_response

http://www.silcom.com/~aludwig/images/Digital_filter_impulse _compare.gif

Peter

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73435 is a reply to message #72395 ]

Sun, 04 September 2005 23:25

Sven Haferkamp Messages: 1 Registered: September 2005

Newbie

HEy TT, the Processor you are searching for is the DSC28 by AD-Systems (or also Klein+Hummel ProC28). It corrects the Phase REsponse of the speaker and also the frequency respsonse if it is your wish.

First Solution - Controller was found
Second Problem - we are not able to produce this controller any longer because of some Motorola Processors that are not longer available. But....on the next Prolight & Sound 2006 in March, Frankfurt, Germany, we will show a new amplifier with integrated FIR Controlling Concept, including LAN REmote, AUdio over Ethernet, 14 Parametric EQ for ROom Equalization, and more.

If you are intersted- send me a mail !

Best Regards from Germany

Sven Haferkamp

Report message to a moderator

Re: Anyone knows a Pro FIR filter Speakerprocessor? [message #73556 is a reply to message #73418 ]

Mon, 05 September 2005 20:02

Fred Blatt Messages: 45 Registered: April 2004

Should Get Out More

Peter wrote on Sun, 04 September 2005 20:44

found it -
http://www.silcom.com/~aludwig/Signal_processing/Digital_fil ters.htm#Impulse_response
http://www.silcom.com/~aludwig/images/Digital_filter_impulse _compare.gif

As I expected, it's not a fair comparison. The blue line is about an 11th-order (66dB/octave!) Butterworth filter - of course it's going to have lots of time-domain ringing! You can clearly see that the blue line's frequency response stays much flatter before dropping off quickly compared to the other 3 lines. The 4th-order (24dB/oct) Linkwitz-Riley has roughly the same frequency response as the two FIR examples until about 10dB down (a point above which the crossover slope would likely be most meaningful), and it has roughly the same amplitude of time-domain ringing.



I'm not disagreeing with your opinion or offering my own opinion on IIR versus FIR crossover filters. I'm just pointing out that the example time-domain graph you chose was not a fair one to use as a comparison between the two types.

Fred

• Attachment: Digital_filter_compare.gif
  (Size: 81.05KB, Downloaded 112 time(s))

Report message to a moderator

Pages (2): [ «  <  1  2]

Previous Topic:	Focusrite Got Burglarized
Next Topic:	Does anyone use REV500?

Goto Forum: - SR: LAB  LAB: The Classic Live Audio Board  LAB Subwoofer  Lighting  audioseminars.com  LAB Marketplace- SR: LAB Lounge  LAB Lounge- SR: Product Reviews  Product Reviews: Sound Reinforcement- SR: Installed Sound  Installed Sound/Contracting- SR: Church Sound  Church Sound  H.O.W. AV- SR: Pro AV  Pro AV- SR: The Basement  The Basement- SR: Testing and Feedback  TestForum  

-=] Back to Top [=-

Current Time: Sun Apr 16 08:08:48 CDT 2006

Total time taken to generate the page: 0.04257 seconds

.:: Contact PSW :: Advertise on PSW :: Your Privacy :: Feedback :: Forum Home :: PSW Home ::.

© copyright 2006 ProSoundWeb.com
169 Beulah Street, San Francisco, CA, 94117 USA
Voice: 415 387 4009  |  Fax: 415 752 8144
Send comments or questions to forums@prosoundweb.com

Powered by: FUDforum 2.7.4 Copyright ©2001-2004 Advanced Internet Designs Inc.

fud_msg_focus("msg_73556");

[라디오]

http://srforums.prosoundweb.com/index.php/m/73556/0#msg_73556