Planar Speaker Asylum

Welcome! Need support, you got it. Or share you ideas and experiences.

Return to Planar Speaker Asylum

digital crossover points and slopes -again

77.12.214.231
Posted on October 31, 2009 at 05:07:06
play-mate
Audiophile

Posts: 391
Joined: November 21, 2008
 dear inmates,

playing around with slopes and crossover points with a digital crossover is surely a fun and rather time consuming undertaking. that was of course also one of the reasons to implement one in the first place. the main ambition was however to see if one could find a more "ideal" configuration.
especially the reversed mid-driver on my MG3.3R´s has been on my mind, and I´ve been curious to see if I could get all three drivers run with the same polarity.

I have been very reluctant to forward any hastened conclusions, but time is ripe to suggest a setting that I will ask those of you with a digital crossover to try out. (and report back on).

these settings are with all three drivers @ same polarity !!

LP : 24dB at 330Hz
HP: 36dB at 330Hz
LP: 24dB at 1800Hz
HP: 24dB at 1800Hz

all things equal, the phase shift by the XO is reversed by 180° by the reversed slope steepness (as from stock), and countered by the reversed polarity of the mid-driver.

please try it out. I´m keen on comments !


kind regards
are you three way?, posted on November 2, 2009 at 08:09:51
Keith944T
Audiophile

Posts: 127
Location: Western PA
Joined: April 23, 2009
 Or maybe I mean are you tri-amping?
Or maybe the better question is are you setting all three cross points?
I'm digital on the factory inputs...2 and have been messing a bit with the freq's on those. running the high pass on the low side up to 350 and down to 150 on the low pass on the high side.
I haven't noticed any improvement by changing from the factory suggested slopes. But I haven't tried a bunch excepet those that maybe you suggested? about doubling them? I don't remember if that was you. Well, I didn't hear an improvement in fact it was worse, but I don't remember what the "effect" was as that was about a month ago?
But if you tell me specifically a setting that I could do on my bi-amp hookup, I'll give it a shot.

Right now I'm running at 303 hz high pass low side and 200 hz low pass on high side factory suggested slopes 18 & 6 respectively.

RE: are you three way?, posted on November 4, 2009 at 14:29:46
Jim_E
Audiophile

Posts: 100
Location: Seattle
Joined: November 8, 2007
 Keith & Playmate,
The 3.6R and 3.3R crossovers are actually very different animals if the tweaks' reference schematics are correct.

* The 3.3 has a third-order filter on the bass driver, where the 3.6 has a fourth-order filter.

* The ribbon on the 3.3R is a fourth-order filter, where the 3.6 is ribbon is a second order.

I derived the 3.6R bass driver response (blue line in plot) and also modeled 4th & 8th order Butterworth and Bessel Filters at 160Hz to give you an idea of the differences.

regards,
Jim
phase response and XO slopes, posted on November 4, 2009 at 18:02:38
play-mate
Audiophile

Posts: 391
Joined: November 21, 2008
 hey jim,

thanks for your reply.
considering how much space the planar asylum devotes to XO´s for maggies, I was actually counting on more response on this issue..... maybe they´re all still listening ?


playing around with filter slopes and configurations I´m certainly not come under the illusion that this stuff is very straightforward theoretical nor easy to tune in by just listening.....in the contrary !

neither did I want to imply that my findings with my MG3.3R´s are to copy 1:1 for 3.6R´s owners, but my wish was merely to gather response on the theory of doubling slope attenuation and / or to support / break my observations on tweaking the phase response with equal polarity mid-drivers by opposing the XO´s phase response.....

I still think this could be a worthwhile exercise for those with digital XO capabilities.

remember that dipoles are difficult to take out of their context (room).

kind regards
RE: phase response and XO slopes, posted on November 5, 2009 at 09:03:08
Jim_E
Audiophile

Posts: 100
Location: Seattle
Joined: November 8, 2007
 Play-mate,
I was surprised at the number of variations in the 3.xR series cross-overs. I hadn't really spent any time looking at all these models before.

You are really in an ideal situation by tri-amping, and being and being fully able to select all the cross-over points. My suggestion would be to try 24 dB/octave cross-overs for each driver, and adjust the driver lead polarity to all be the same.

There is some comprise between the steepness of the crossover slope and the overall performance. The larger the crossover order, the more the impulse response is "smeared" or prolonged. (This is true for either analog or digital implementations.) My thought is that 4th-order (24 dB/Octave) is a good comprise between an adequately steep crossover, and maintaining a good impulse response.

Most commercial digital crossovers units (and your software?) allow a choice between Butterworth, L-R, and Bessel filters. Butterworth filters are traditional since they can actually be built with analog components easily. The L-R filters are really a cascaded Butterworth that have better phase and magnitude characteristics.

For ease of use, the L-R filters are a good starting point as you can use the same frequency for each high/low pass pair.

I'm an advocate of Bessel filters, but they are somewhat more involved to setup properly. Bessel filters offer the best phase response of all three filter types, but are somewhat less steep. If both the LP and HP are set to the same frequency, you will have a slight dip in the magnitude response. This can be compensated for by adding a slight "overlap" of the crossover points. For fourth-order use a factor of 1.53 for the low pass, and a factor of 0.65 for the high-pass.

Presented with the almost infinite choices of three filter types, multiple slopes, and variable frequencies it can be quite difficult to determine the "ideal setup". If you start with the 24 dB/decade slopes, and L-R filters, than you can than concentrate on just cross-over frequencies.

Using the default factory values is a good starting point. I would suggest keeping reasonably close to them, since the "drivers" may have their own characteristic roll-offs just outside the cross-over region.

regards,
Jim

RE: phase response and XO slopes, posted on November 4, 2009 at 18:28:28
Neuro
Audiophile

Posts: 625
Location: Santa Cruz, CA
Joined: June 29, 2001
 I unfortunately don't have access to a digital crossover, but I've definitely played with lots of configurations with my 4-way Marchand XM44.

If you're liking the sound of your current setup, then keep it by all means. :) Based on the 3.3 reference schematic, it looks like the midrange is wired with reversed polarity, while the woofer and ribbon are both wired in phase.

I would expect a full LR4 network to perform better than the one you're using. I've tried 8th order Linkwitz-Riley filters, and I felt there was too much separation between the drivers. Shallower slopes can bring back some coherency and seamlessness but often at the expense of lost detail.

Magnepan obviously chose their whack-ass crossover networks for a reason (because they work very well) but It definitely makes the job harder for us active-xo people. :) IIRC, the DEQX HDP-3 can measure timing delays for EACH driver and adjust its digital crossover accordingly. It's almost like you could let it measure your driver's performance and produce an optimized crossover. How awesome would that be? Or you could just bribe Wendell to give you the active XO settings they used during speaker dev. ;)

-- Nils

RE: phase response and XO slopes, posted on November 5, 2009 at 02:54:53
play-mate
Audiophile

Posts: 391
Joined: November 21, 2008
 hey nils,

I DO envy the DEQX feature of time / phase aligning automatic, but that apart, my computer music server has all the other capabilities.
unfortunately I have not yet figured out how to integrate the ispCAD program with it yet, but it is certainly my ambition to achieve the same minute control of the maggie drivers, as is possible with the DEQX.
-at 1/8 of the cost BTW !!

working with this kind of powerful tools makes only optimal sense if you take all the parameters into the equation. i.e. reflections, delays and dispersion etc.

while it is also my opinion that magnepan did their homework when constructing their XO networks, I bet there was lots of compromises. the reversed mid driver is just one example !


kind regards
RE: phase response and XO slopes, posted on November 5, 2009 at 09:49:25
Neuro
Audiophile

Posts: 625
Location: Santa Cruz, CA
Joined: June 29, 2001
 play-mate,

Best of luck. Just curious -- have you experimented with your speaker toe-in and position with each crossover setting?

I found that with a fully symmetric LR4 network (240hz, 1700hz points), the speakers sound best with ribbons inside and toed in more or less to be aimed at the listening position; on the other hand, the stock slopes with their intended timing delays sound best with tweeters out and with less toe-in.

FWIW, reversed polarity of drivers is pretty much a given for any network with 180° of phase difference (e.g. 1st order + 3rd order, or a standard LR2 network).

For what it's worth, I'm going to stick with the Marchand unit for a while. Phil is a pleasure to work with, and unfortunately I don't have a Windows machine with which to configure a DEQX.

-- Nils

RE: are you three way?, posted on November 4, 2009 at 13:29:18
Jim_E
Audiophile

Posts: 100
Location: Seattle
Joined: November 8, 2007
 Keith & Playmate,
The 3.6R and 3.3R crossovers are actually very different animals if the tweaks' reference schematics are correct.

* The 3.3 has a third-order filter on the bass driver, where the 3.6 has a fourth-order filter.

* The ribbon on the 3.3R is a fourth-order filter, where the 3.6 is ribbon is a second order.

I derived the 3.6R bass driver response (blue line in plot) and also modeled 4th & 8th order Butterworth and Bessel Filters at 160Hz to give you an idea of the differences.

regards,
Jim
RE: are you three way?, posted on November 4, 2009 at 13:52:18
Neuro
Audiophile

Posts: 625
Location: Santa Cruz, CA
Joined: June 29, 2001
 Furthermore, the stock passive crossover for the 3.6 effects a placement where the tweeter should be farther from your ear. The asymmetric slopes result in the tweeter leading the midrange leading the woofer, IIRC.

In many rooms, I've found that Magnepans sound best with tweeters out and minimal (i.e. < 20°) toe-in, and this placement is conducive to such toe-in.

With symmetric crossover configurations (especially second- and fourth-order) you really should be aiming the panels directly at (or very close to) the listening position. For some room configurations, this leads to lots of toe-in.

*shrug*

-- Nils

RE: are you three way?, posted on November 2, 2009 at 17:52:47
play-mate
Audiophile

Posts: 391
Joined: November 21, 2008
 hey keith944T

well, of all the possible settings of a digital XO......99% are not right.

any manufacturer of loudspeakers with passive XO´s will have to judge the right compromise of equations. running things active with digital XO will, (all things equal) most likely offer a closer-to-the-ideal configuration.

I´m surprised that you can´t hear any difference.
-I would accept that you can´t decide which setting is "better", -but "no difference" at all ???

but yes : I did claim that doubling the slopes seemed to keep the phase issues in check.
and yes : I do run my system in tri-amp mode.

I´m not sure what model maggies you are having, but it seems you´re having a multi-channel theatre system, which does of course "delude" phase and timing in a bewildering way.

I´m of the firm belief that two channel stereo, is best with only two speakers.....

kind regards
I didn't say, posted on November 2, 2009 at 19:45:00
Keith944T
Audiophile

Posts: 127
Location: Western PA
Joined: April 23, 2009
 I didn't hear a difference, I said it wasn't an improvement.

I have 3.6's, running a 1000 w/chnl ICE Class D on the the low pass and 150 w/chnl Monoblock Tubes on the high side.
Running through a Behringer dcx2496, I know it's not optimal, but I want to learn what "might sound" better.
Plus I like all those buttons !

I can see how tri-amping might be the way to go, but alas, I'm not up to the task of "surgery on the Maggies". Not too crazy about finding a place for yet ANOTHER amp!

I listen to 2-channel a lot and mostly in fact. I have a tube Pre-amp that has a pass through for the AVR for movies and multichannel SACD.
But most of the time it's just 2 channell with the 3.6's.
Source is usually a Cambridge Audio Azur 640H music server/CD player.
I run a Cambridge Audio DVD 99 as the SACD player through the analog/2 channell system as well for the high SACD bit rate.

The 1.6's are my surrounds and are fed from the AVR to a Rotel 1080B.

Page processed in 0.034 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2009, Audio Asylum, all rights reserved ] [ Go Green ] [ Take5 ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]