At below link under product menu service manuals is available at lower left corner for fair costs, but think you agree its better make some small preparation plan and measure in situation.Thanks Byrtt. Memory is not what it used to be
What are the inductor values thigh I wonder?
Here a link to S45-2 which seems praised as best sonics of the linear phase models https://www.beoworld.org/prod_details.asp?pid=666 .
I'm interested in using this type of crossover for a passive W(TM)W center speaker but I'm worried about horizontal lobing issues. I tried simming it in Boxsim and always see huge horizontal lobing around the W-M crossover point. (but I need to invert woofer polarity in order to get W-M to sum nice so I'm probably missing something) Can the more theoretically minded maybe shed some light on this? Is it inherent to the design because of low order slopes and therefore not suited to a center speaker?
Zoneschimmel,
The polarity of all drivers need to be positive. It is true that you will get more of a significant issue with lobing when used horizontally - in vertical arrangement it is less of an issue as we don't move our heads vertically very much once in a listening position.
If your woofers and mids have a cancellation that requires inversion that would mean that either phase was off 180deg or mid is 2nd order and woofer is 2nd order also. You want woofer to be 2nd order low pass and mid to be 1st order high pass. It's actually very difficult to achieve true 1st order high pass over a large bandwidth. Only possible with a good full range driver and XO frequency needs to be at least 2 octaves above natural roll off of the mid. For a mid like a SS 10F/8424, it starts to naturally roll off below 200Hz. So 2 octaves above means that the mid high pass needs to be no lower than 800Hz for a 1st order filter that reaches to 200hz following a 1st order target function.
The polarity of all drivers need to be positive. It is true that you will get more of a significant issue with lobing when used horizontally - in vertical arrangement it is less of an issue as we don't move our heads vertically very much once in a listening position.
If your woofers and mids have a cancellation that requires inversion that would mean that either phase was off 180deg or mid is 2nd order and woofer is 2nd order also. You want woofer to be 2nd order low pass and mid to be 1st order high pass. It's actually very difficult to achieve true 1st order high pass over a large bandwidth. Only possible with a good full range driver and XO frequency needs to be at least 2 octaves above natural roll off of the mid. For a mid like a SS 10F/8424, it starts to naturally roll off below 200Hz. So 2 octaves above means that the mid high pass needs to be no lower than 800Hz for a 1st order filter that reaches to 200hz following a 1st order target function.
It's inherent to the design - because the woofer and mid are supposed to be aligned to 90degrees out of phase they reach 180degrees (full cancellation) at a much shallower listening angle.
Hmm and for a passive Harsch I would need to recess the mid-tweet at least 20cm, a bit much unless using a horn. Xrk do you know of any other transient perfect topologies which could be used for a center?
Would it be possible to built a XBush like passive center or does it need a lot of delay?
Would it be possible to built a XBush like passive center or does it need a lot of delay?
Just a plain 1st order symmetric XO at 600Hz requires about 3in of setback on the tweeter (or mid in this case) to be properly time aligned. However, even with no setback as shown here and simply flipping the speaker upside down so that the mid/tweeter is below the woofer so that at ear level, there is about 3in extra time-of-flight difference will give you a similar nice response. So basically, listen slightly off axis from mid/tweet (mic placed on axis of woofer for these measurements).
Here is what I did to show you that setback is actually not needed.
http://www.diyaudio.com/forums/full-range/273524-10f-8424-rs225-8-fast-ref-monitor-68.html
XO design:
Simple implementation of XO with fairly low cost caps (50uF BP electrolytic with balance in poly film caps, and small ferrite core coils and small thin film 2w power resistors in parallel:
FR and phase:
Impulse and step response:
For those of you worried about possible high distortion of all that "bleed through" because of a low 1st order filter, it's not so bad at all:
I used BYRTT's trick to make target curves in Rephase, imported them into REW fiddled a bit with levels and delays to get nice summing and imported the frd files into boxsim to get a look at the lobing for a W(TM)W with the following baffle:
For some reason the phase of Bessel low pass at 800 goes way to high towards the treble, tried different settings in Rephase but couldn't find a fix for it.
Horrible horizontal lobing even at 30°
For some reason the phase of Bessel low pass at 800 goes way to high towards the treble, tried different settings in Rephase but couldn't find a fix for it.
Horrible horizontal lobing even at 30°
Last edited:
You definitely can make a passive with two 6.5in woofers using the bookshelf xBush concept. Care and attention had to be paid to where to put the band pass injection ports to properly time align with mid/tweet. Unfortunately it may requires a mockup in foam core or wood as the time delay introduced by the band pass front chamber and ducting to the horn cannot be predicted accurately enough. The mockup will allow the FRD files to be used to determine the virtual acoustic offset. Then the needed delay has to be figured in and the new injection hole location picked out. I know you can predict the mid injection holes very well because the driver's native offset is relatively small and the chamber small.
Nice work with the crossover simulations. I think for B&O Hole Filler the target on the woofer and tweeter is LR2 not BES2 - although may not matter.
Looks like both of the Bessel transfer functions have a delay added - hence the phase increasing continuously as the frequency increases.
The delay-phase relationship is how you can (at least partially) compensate for a time alignment issue using passive filters by use of asymmetric crossover points and/or asymmetric slopes. Of course, having your drivers time aligned at the desired crossover frequencies to begin with makes life a whole lot easier.
The delay-phase relationship is how you can (at least partially) compensate for a time alignment issue using passive filters by use of asymmetric crossover points and/or asymmetric slopes. Of course, having your drivers time aligned at the desired crossover frequencies to begin with makes life a whole lot easier.
Last edited:
I found this article an interesting read: A Unique Loudspeaker Crossover Design with Waveform Fidelity
It describes not only various types of the Baeckgaard crossover but als the Yamanaka crossover with a simular setp up, different Q, and even predating the B&O filler driver concept.
Also, the subtractive way to transient perfect response.
As I found while reading, the subtractive way of filtering is flawed because of the output of the subtractor that will only reach 1st order slope.
It is suggested that allpass filters after the subtractor can correct the phase.
Elsewhere I found a suggestion that another low-pass after the filter can correct the response.
Neither suggestions seem to be transient perfect solutions, and cannot sum to a square wave.
I wonder what happens if you'd subtract the subtractor output again, essentially cascading it through another subtractor. Will that still output a square wave when summed? I've never seen any examples of it. Probably, my thinking is off.
It describes not only various types of the Baeckgaard crossover but als the Yamanaka crossover with a simular setp up, different Q, and even predating the B&O filler driver concept.
Also, the subtractive way to transient perfect response.
As I found while reading, the subtractive way of filtering is flawed because of the output of the subtractor that will only reach 1st order slope.
It is suggested that allpass filters after the subtractor can correct the phase.
Elsewhere I found a suggestion that another low-pass after the filter can correct the response.
Neither suggestions seem to be transient perfect solutions, and cannot sum to a square wave.
I wonder what happens if you'd subtract the subtractor output again, essentially cascading it through another subtractor. Will that still output a square wave when summed? I've never seen any examples of it. Probably, my thinking is off.
The problem with all these x/o's (Baekgaard, Subtractive) is, that they require ideal, texbook drivers with extraordinary bandwith. It can be done, but requires a lot of SPL correcting before it finally works.
Much simpler tools nowadays are available to achieve better results. Rephase is just
one of them
Much simpler tools nowadays are available to achieve better results. Rephase is just
one of them
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.