ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 [2] 3   Go Down

Author Topic: Thinking outside the greybox  (Read 14819 times)

Langston Holland

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 82
Re: Notched Crossovers
« Reply #10 on: March 18, 2012, 11:18:01 am »

Hi Uwe:

I italicized and footnoted the portion of the quote from my post that you reversed in meaning to make it clear to less careful readers that I didn't say that. It's not a problem at all BTW and it allowed my first opportunity as mighty moderator to change someone's post. Now I feel drunk with power! :)

The 18 sound driver's woofer peak at 2.5kHz is interesting and may very well be happening with the MW12's woofer as well - I don't know yet. My point is that it simply doesn't matter since the low pass filter skirt is already reducing the woofer's output in that area in excess of 40dB. Adding a notch filter to remove that peak in this situation is like kicking a dead horse.

I hearby add a new term to the vocabulary of filters: Anyone using EQ to correct a portion of the passband that is too low in level to affect the loudspeaker's response in anyway is using an emofilter.
Logged
God bless you and your precious family - Langston

Hayden J. Nebus

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 637
  • Richmond, VA
Re: Notched Crossovers
« Reply #11 on: March 18, 2012, 12:29:17 pm »

only some minor corrections  :)

compare this http://www.eighteensound.it/index.aspx?mainMenu=view_product_simple&pid=248 ( Curves )
it is probably not the same driver but, but maybe similar

Uwe

That's interesting, and IF that curve looks very similar at 2k, which it probably does, answers my question... kinda.The filter likely corrects for the peak in driver response along the LPF skirt.

I think I see here how someone, in this case the esteemed Mr. Thiele, might go about discovering/developing the notched crossover. One has a cone driver that exhibits breakup behaviour along its transition through a LPF into the stop band. One then adds a notch to make driver response more linear in the transition band. Looking at the combined response of the notch and LPF, one would see that they have effectively increased the slope for an octave above the LPF corner frequency while maintaining a near identical phase response. Why not apply that symetrically on the next passband? Who doesn't want steeper attenuation with less group delay? Kinda like this "blue" preset: who doesn't want maximum GBF in half space conditions from a WEDGE?



Logged

Ivan Beaver

  • Hero Member
  • *****
  • Online Online
  • Posts: 9032
  • Atlanta GA
Re: Notched Crossovers
« Reply #12 on: March 18, 2012, 12:36:53 pm »

The idea of a notching out "out of band" signals is nothing new.  I was first exposed to it back in 2000 (I believe) at one of John Murrys classes.  I am sure it had been around for quite awhile before then-I just hadn't been exposed to it.

It can be used in several different ways-one of which is being discussed (a steeper slope without screwing up the phase) and by getting rid of "pesky" peaks above and below crossover freq, helps the overal response of the crossover.

We do a good bit of it in our active alignments.  Nothing wrong with "throwing away" stuff you have to much of.
Logged
A complex question is easily answered by a simple-easy to understand WRONG answer!

Ivan Beaver
Danley Sound Labs

PHYSICS- NOT FADS!

Uwe Riemer

  • Newbie
  • *
  • Offline Offline
  • Posts: 43
Re: Notched Crossovers
« Reply #13 on: March 18, 2012, 07:22:19 pm »

Hi Uwe:

I italicized and footnoted the portion of the quote from my post that you reversed in meaning to make it clear to less careful readers that I didn't say that. It's not a problem at all BTW and it allowed my first opportunity as mighty moderator to change someone's post. Now I feel drunk with power! :)

The 18 sound driver's woofer peak at 2.5kHz is interesting and may very well be happening with the MW12's woofer as well - I don't know yet. My point is that it simply doesn't matter since the low pass filter skirt is already reducing the woofer's output in that area in excess of 40dB. Adding a notch filter to remove that peak in this situation is like kicking a dead horse.

I hearby add a new term to the vocabulary of filters: Anyone using EQ to correct a portion of the passband that is too low in level to affect the loudspeaker's response in anyway is using an emofilter.

Hi Langston,

Im glad you took it the intended way, though you missed two other minor "corrections"  :)

Now to the emofilter:
the peaks around 2.5kHz in the 12CX800 example are narrow, and could exceed 12dB measured with less smoothing ( Assumption ! ), which would render the -40dB to less than -30dB

and narrow peaks usually come along with strong ringing in the time domain

in Haydens graph I can see some deviations above and below the 2.5k notch, which could be
a) measurement error
b) some FIR trickery
if b) , why should EAW bother with some trickery at -40dB and below, if it does nothing

May the power be with you  :)


Uwe

« Last Edit: March 18, 2012, 07:25:58 pm by Uwe Riemer »
Logged

Langston Holland

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 82
Re: Notched Crossovers
« Reply #14 on: March 18, 2012, 08:16:25 pm »

Quote
May the power be with you  :)

And to you, young skywalker!

I've decided to pursue this thing to the end, meaning I'm going to find a MW12 from somewhere(1) and do all the necessary acoustic measurements as well as the electrical. I will publicly eat crow(2) if that 2.5kHz notch filter does anything useful.

1. Anyone willing to ship a MW12 to zip code 32571 would be appreciated. I'll pay return shipping. For the interest of science. :)
2. Eat crow: informal colloquialism; to be humiliated by having to admit one's defeats or mistakes.
Logged
God bless you and your precious family - Langston

Mac Kerr

  • Old enough to know better
  • Administrator
  • Hero Member
  • *****
  • Offline Offline
  • Posts: 6829
  • Audio Plumber
Re: Notched Crossovers
« Reply #15 on: March 18, 2012, 09:13:50 pm »

I've decided to pursue this thing to the end, meaning I'm going to find a MW12 from somewhere(1) and do all the necessary acoustic measurements as well as the electrical. I will publicly eat crow(2) if that 2.5kHz notch filter does anything useful.

I'm pretty sure I remember Dave Gunness talking about this very point in one of his presentations at the LAB Expo. IIRC he was talking about using out of band filters to do phase correction. I'm sure an email to Dave will get you an answer.

Mac
Logged

Jay Barracato

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2025
  • Solomons, MD
Re: Notched Crossovers
« Reply #16 on: March 19, 2012, 10:26:57 am »

I'm pretty sure I remember Dave Gunness talking about this very point in one of his presentations at the LAB Expo. IIRC he was talking about using out of band filters to do phase correction. I'm sure an email to Dave will get you an answer.

Mac

I made a post about this a while ago when I was trying to wrap my head the same thing. The problem I kept bumping into is that the width of the effect in the time domain matches the width in the amplitude domain.
Logged
Jay Barracato

Hayden J. Nebus

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 637
  • Richmond, VA
Re: Notched Crossovers
« Reply #17 on: March 19, 2012, 03:09:01 pm »

Here is the axial LF driver response.



And here is the processing in the blue background trace




Logged

Langston Holland

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 82
Re: Notched Crossovers
« Reply #18 on: March 20, 2012, 02:20:23 am »

Crow is not on the menu.

Yet.

EAW does not include the emofilter in their Lake presets for the MW12. I'm sure this isn't an oversight. More on this in a bit.

Even though I've mentioned it a couple of times, I'm not sure some folks understand what I'm saying about out of band EQ, notches, etc. They are VERY useful on many occasions, but not when they are so far out of band that no part of their transfer function contributes anything to the loudspeaker's output. I'll quit now. :)



I do not yet have the acoustic measurements of a MW12 to conclude this discussion, but should have within a couple of weeks.

For now I've made the electrical measurements of the MW12 "blue" preset in both the UX8800 and Lake processors. Much can be learned by studying these.

First, let's look at the UX output. You'll notice a great deal of narrow bandwidth EQ correction on both high and low passbands. Presumably these corrections address a large average of MW12's both on and off axis. I don't have the experience to know if such small perturbations are consistent across hundreds of coaxial drivers over the expected lifecycle of this product, but I assume so or EAW wouldn't have done it.

Notice that the unwrapped LF passband phase trace (red) is a normal downward sloping affair, whereas the HF passband phase trace (cyan) slopes upward, which is the opposite of how nature works. This is a by product of the focusing that results in a flat phase response when combined with the downward sloping acoustic output of the HF driver.

UX8800 MW12 Blue Preset (electrical):



Since no focusing is involved with the Lake preset, you'll notice both phase traces are minimum phase, thus slope downward.

Lake MW12 Blue Preset (electrical):



A comparison of the UX and Lake LF passband follows. Notice that the phase traces overlap throughout the region where there is usable output from the woofer. This is one proof that the UX does not employ focusing in the LF passband. They did use the filter software to create the coefficients responsible for the dozens of minor EQ corrections for both passbands, but avoided the huge latency penalty that would have been involved with correcting the time domain of the LF.

Notice that the emofilter is missing from the Lake preset. The fancy filter design software that EAW (and others) use have adjustable frequency boundaries that you set to tell the software where to optimize and where to leave things alone. That included the 2.5kHz acoustic bump that the LF driver obviously has in the MW12, but EAW didn't bother with the manual implementation of same in their Lake presets.

Lake vs UX MW12 LF Blue Preset (electrical):



I'll finish this up once I get my hands on a MW12.

Edit: Just confirmed a demo of a MW12 - thanks Troy. :)
« Last Edit: July 19, 2012, 02:56:49 am by Langston Holland »
Logged
God bless you and your precious family - Langston

Jim McKeveny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1091
Re: Notched Crossovers
« Reply #19 on: March 20, 2012, 08:17:08 am »

Ceteris paribus A.N. Thiele gave us the benefits of higher-order slopes without the audible negatives. I was late to the party, only discovering his methods by reading that 2006 Voice Coil issue. I suspect that much of his work is uncredited (uncompensated?) in many "proprietary" DSP presets...
Logged

ProSoundWeb Community

Re: Notched Crossovers
« Reply #19 on: March 20, 2012, 08:17:08 am »


Pages: 1 [2] 3   Go Up
 



Page created in 0.06 seconds with 25 queries.