The advantage of a brickwall filter is the attenuation slope. Phase problems can arise when IIR/analog slopes are steep.
... but a brickwall filter also cuts the fourier components of the signal. Those parts of the signal get removed above and below the target frequency. Because of this, a ripple appears in the impulse response. This can't be avoided. You could get the on axis to sum correctly seeing as the ripples would appear on both sides of the target (and if identical they will cancel) but off axis they will/can be audible. How audible depends on the target frequency. Lower targets might be masked by the wavelength ... higher targets not so much. The steeper the slope, the longer the ringing lasts.
Thank you Puppet, so as i understand it except in certain particular cases (higher efficiency needed - P.A./ live act- or problematic's driver behavior - for eliminate breakup e.g.) no real and definitive advantage upon a 'classic' filter case (LR or JMLC kind) if drivers are good upon the bands they're dedicated to.
I suppose the audibility of the ringing off axis could be 'nasty' as narrow band, and audibly modify the radiated power pattern of loudspeakers.
Still a pick your poison situation if i understand correctly...
Is there an advantage of brickwall considering lobing in a three way with aligned drivers versus a classical filter configuration like say a LR?
Last edited:
Easily? I guess you live out in the highlands... 60 to 65dB is more like it.
Last edited:
Lobing occurs as a direct result of two things, the first is the interference pattern created when two or more drivers cover the same band of frequencies and the second is due to any off axis characteristics that would modify this and or be responsible for any lobing in it entirety.
Lets say you take a standard two way loudspeaker with a 4" mid/bass, a 1" dome tweeter, place them 30" apart and now cross it at 1500Hz with a 4th order filter. At and around the xover frequency the 4" will not be beaming but both it and the tweeter will be reproducing the same band of frequencies. If you go vertically off axis, due to the large driver separation you will get an interference pattern created by the two drivers going in and out of phase with one another, this is standard lobing. If you were to cross these two drivers to one another with brick wall filters, then theoretically only one driver would reproduce a given frequency at any one time, so all lobing around the xover frequency would disappear.
If you keep going up in frequency, until the tweeter starts to beam, then the primary listening lobe would start to narrow. If you were using a driver of a suitably large diameter, or were crossing the two drivers over at say 6kHz, where the 4" will have already started to beam, then the lobing pattern will be modified due to its response starting to droop. In this instance if you were to use a brick wall filter then you'd still get the primary lobe narrowing as the mid/bass starts to beam and then widen as you crossover to the tweeter, before it narrows again.
Lets say you take a standard two way loudspeaker with a 4" mid/bass, a 1" dome tweeter, place them 30" apart and now cross it at 1500Hz with a 4th order filter. At and around the xover frequency the 4" will not be beaming but both it and the tweeter will be reproducing the same band of frequencies. If you go vertically off axis, due to the large driver separation you will get an interference pattern created by the two drivers going in and out of phase with one another, this is standard lobing. If you were to cross these two drivers to one another with brick wall filters, then theoretically only one driver would reproduce a given frequency at any one time, so all lobing around the xover frequency would disappear.
If you keep going up in frequency, until the tweeter starts to beam, then the primary listening lobe would start to narrow. If you were using a driver of a suitably large diameter, or were crossing the two drivers over at say 6kHz, where the 4" will have already started to beam, then the lobing pattern will be modified due to its response starting to droop. In this instance if you were to use a brick wall filter then you'd still get the primary lobe narrowing as the mid/bass starts to beam and then widen as you crossover to the tweeter, before it narrows again.
5th Element, Thank you for your answer about lobing, it clarify certains things i had not entirely understood.
In fact now i understand why taking measurement about directivity behavior of each drivers in box are important to determine for optimal results.
For what i understand from the second part of your answer one of the main concern in choosing the cutoff point of Xover is the 'continuity' of directivity pattern between adjacent drivers.
So it raise another question for me: for best results which general directivity pattern must be choosen ( in case of drivers without horn)? I would say a natural narrowing of directivity dicted by medium/tweeter transition (e.g. if at 6khz medium as directivity of 60° -6db, tweet should be the same).
In fact now i understand why taking measurement about directivity behavior of each drivers in box are important to determine for optimal results.
For what i understand from the second part of your answer one of the main concern in choosing the cutoff point of Xover is the 'continuity' of directivity pattern between adjacent drivers.
So it raise another question for me: for best results which general directivity pattern must be choosen ( in case of drivers without horn)? I would say a natural narrowing of directivity dicted by medium/tweeter transition (e.g. if at 6khz medium as directivity of 60° -6db, tweet should be the same).
That makes perfect sense of course, except - I remember reading an old Altec technical bulletin in which they actually measured cones of different sizes and profiles. IIRC, they found only the diameter have have a significant effect. There was a graph with the tech note showing beaming vs cone size. I'll see if I can dig it up.
Could you be more precise Puppet please? I don't understand the difference between dispersion pattern and off-axis response, in fact the way i see it (and what i've got in mind in my precedent comment) is to take measurement of frequency response using a sine sweep of usable area of driver (plus/minus one octave) at different angle to define dispersion pattern.
E.g.: one measurement on axis then at 15° off axis, then 30°, then 45°, etc,etc.
And repeat this same process for each driver in Loudspeaker.
This should give all nescessary informations no?
Bare, interesting though about section geometry. In my case drivers are 'flat': honeycomb disk aluminium drivers from Technics. A 'big' Three way with 15" woofer 3" medium and 1" tweeter, and by nature all drivers are time aligned.
Pano i'm curious to see this document if you can find it.
Here it is, Altec Technical Letter #237
http://alteclansingunofficial.nlenet.net/publications/techletters/TL_237.pdf
Note what is said in the second paragraph
http://alteclansingunofficial.nlenet.net/publications/techletters/TL_237.pdf
Note what is said in the second paragraph
ANALOG ACTIVE CROSSOVER UPDATE
I tried designing active crossovers with LTSpice this weekend, by simulating the circuit together with the potential amplifier circuits. While I can achieve good performance in AC analysis, FFT was horrible.
We usually work very hard to get a good performing amplifier. Now with the active crossover front end, the task is even harder!
Using analog active crossover is like running a good speaker with bad amplifier
I tried designing active crossovers with LTSpice this weekend, by simulating the circuit together with the potential amplifier circuits. While I can achieve good performance in AC analysis, FFT was horrible.
We usually work very hard to get a good performing amplifier. Now with the active crossover front end, the task is even harder!
Using analog active crossover is like running a good speaker with bad amplifier
"Horrible" means the low THD.
I tried 2 types of crossovers (Actually 3 but I quickly gave up using gyrator due to phase issues). Discrete and opamp based. For discrete I made a circuit based on Kaneda, but of course I used my own JFETs and BJTs. For opamp based I used 4 opamps per driver: 3 for Bainter Notch filter and 1 for LP/HP (It is perfect for outdoor system where THD is not critical).
For LP I aimed to use OPA111, for HP I used faster opamps where the faster the opamps the better the performance.
It is no secret. Just build an amplifier with opamp input and you will find in either simulation, measurement or listening test that the performance is a function of the opamp quality, and in my experience is far from top quality. And here we have to deal with more than one opamps...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.