PDA

View Full Version : A New Crossover Theory



Ian Mackenzie
02-21-2005, 07:42 PM
A new Crossover Theory
Below is an extract of a post from diyaudio.com that may prove interesting for those involved in the crossover design, particularly if you have access to a sophisticated digital crossover network.

Improoved Sound Stage and 3D-Imaging Post #265 (http://www.diyaudio.com/forums/showthread.php?postid=579241#post579241)Dear folks out there,
I would like you to participate at one of the most interesting filter designs I ever tested on my Behringer.

Some time ago a French gui called Thierry Martin mailed me some information about a new kind of filter design for speaker crossovers engineered my Mr. Le Cleach. The material was ONLY in French so it was quite hard for me to follow his approach. Finally I found a French person which translates the article to German (uuh, what an ugly German!). After that I started translating the article to real German and also to English for you. The whole material you’ll find here (8am to 10pm german time):

Document and Speadsheet (doc, xls): http://freerider.dyndns.org/anlage/LeCleach1.zip (http://freerider.dyndns.org/anlage/LeCleach1.zip)
Theory of operation (ppt): http://freerider.dyndns.org/anlage/LeCleach2.zip (http://freerider.dyndns.org/anlage/LeCleach2.zip)


Well, in-between I tested the approach of Le Cleach with my DCX-2496 and got so astonishing results that I thought I had to let you know about his theory. Believe me or not: With his filter design you get a soundstaging and imaging you never heard before on your system!!

So what has to be done?

First adjust your DCX-2496 the way you have a flat frequency response and time aligned speakers which can be achieved by the auto align function of the DCX 2496 (yes, you need a microphone this is VERY important!). I guess you did it before ;-)

Now read the paper (.doc-file) I submitted in the ZIP, open the spreadsheet and enter the values according to your current DCX-2496 configuration. Do not enter the current time delay of your DCX in the spreadsheet! Now in the spreadsheet modify the values as described in the paper. If you got a good result in the spreadsheet adjust your values in the DCX-2496 so they match the spreadsheet.
A bit care has to be taken at the delay. The auto align function of the DCX calculates values for the delay. To these values the values from the spreadsheet have to be added. This means the distance between the chassis will become bigger. Le Cleach says the speakers (bass, mid) have to be moved towards the listener. With the DCX-2496 you have to move the mid and tweeter away from the listener which is the same!! The distance gets bigger.

Now save this adjustment under a new preset of your DCX, sit down and listen while switching between your old preset and the new one. What you will hear is really astonishing. The whole characteristic of the sound will stay the same (same frequency response) but with the new settings the soundstage expands and you will get a never heard 3D-imaging.

A nice test for the new setup is the SACD “Scared Love” from Sting. In my opinion it is a bit over instrumented and it is sometimes difficult to differentiate between the voices and the instruments. With the new filter parameters the whole soundstage unfolds and ….. you know, I now really love this SACD.

And best of all: The filter modification is for free just try :-)))

Earl K
07-27-2006, 05:11 PM
Something to look at :)

- These studies use the Le Cleach spreadsheet to illustrate how one can reduce the buildup of energy at Fc ( that accompanies many of the filter choices used in passive crossovers ) .

- Mostly, the hipass and lowpass F3 points have been chosen to split apart from each other .

- The relevancy ? Well, JBLs' network designers are masters at using unconventional filter types and asymetrical slopes to achieve smoother transitions through the crossover region.

PS ; Oh yeh,, The TRACE is the axial reponse of the drivers mixed together .

The TRACE is what J.M. LeCleach calls "coincidental" energy that can occur from "in-phase" reflections .

:)

Earl K
07-27-2006, 05:11 PM
(i) These studies show a fairly radical mix of filters.

(ii) The last two show hat happens when a LR-24 type is used on the lowpass and is mated to a 12 db Butterworth acting on the Hipass .
- The small dip ( or trough ) at Fc doesn't look too impresive until one literally turns up the woofer section by a couple of db. Then one achieves a very nice curve .


I would encourage anyone interested in playing around with filter types to download this spreadsheet ( that Ian kindly posted a link to ) .

You may not see the relevancy now or even for some time / but it is a useful tool to spend some time with .



:)

boputnam
07-27-2006, 05:58 PM
(Some other intersting) links which might explain it:

http://www.maxim-ic.com/appnotes.cfm/appnote_number/733

Then there is the Rane primer:

http://www.rane.com/note160.html.