This topic does find use in a wide variety of situations where the circumstances get complicated. It's good to talk about ideal filters, also about specific applied cases. Even about specific common aspects.. but generalisations can get cumbersome.
Are you suggesting that L&R made the case for anything more than a filter with two signals being added? Ie, is the discovery of lobing the result of the analysis of an applied situation, a side-effect, or the original intention?
Are you suggesting that L&R made the case for anything more than a filter with two signals being added? Ie, is the discovery of lobing the result of the analysis of an applied situation, a side-effect, or the original intention?
Last edited:
I don't understand the question and it's been many years since I read Linkwitz's AES article, but as far as I recall, it was about a two way loudspeaker with the tweeter above the woofer. The Linkwitz-Riley filter ensured that the main lobe was horizontal and that there was no +3.01 dB bump on axis. The summed response was an all-pass response.
Basically the compromises he accepted were the phase shift of the all-pass response and the -3.01 dB dip in the power response (power integrated over all directions). He argued that the all-pass response was no big deal, because all-pass filters are only audible when their quality factor is very high.
Basically the compromises he accepted were the phase shift of the all-pass response and the -3.01 dB dip in the power response (power integrated over all directions). He argued that the all-pass response was no big deal, because all-pass filters are only audible when their quality factor is very high.
Marcel, I was only trying to understand your point about whether crossing subs qualifies.. by checking what makes an LR filter.
Anyway regarding your latest post, it could also be said that the Butterworth is an all-pass crossover.
Anyway regarding your latest post, it could also be said that the Butterworth is an all-pass crossover.
Magicos use some sort "elliptical" filter, as far as I know. Which might behave quasi the same AS LR with regards to the crossover frequency (range), as far as vertical lobing (or tilt of the main lobe) is concerned. That would depend on whether each filter's branch is down by 3 dB (then it WOULD NOT behave like LR) or by 6 db (then it would behave like it) at the point of crossover.
However, the deficit in power response of a LR crossover manifests itself NOT in measurements of the horizontal dispersion but in the total power == dispersion above and below the (normal) listening axis (assuming mids and highs drivers are aligned vertically). Thus all (or almost all.. I may not no all that is out there) measurements of Magicos available (is there anything else at all but from Stereophile?) don't really show the quality of (total) power response. They show at best a relatively good horizontal dispersion.
Concerning Revels: IF they actually use regular LR crossovers, they are bound to have a deficit in power response. (Unless there is a model with a coaxial driver?) And this will show in all measurements which include vertical angles.
So... I can conclude that horisontal dispersion, is pretty good with LR filters, and that I get a compromise in the vertical dispersion, which I also clearly see on the measurements on Heissmanns homepage 👍 But chances of me moving my head 50cm up and down, are less likely than 50cm left or right. Which should make the LR filters fine 🤞😊
True, but you do get some colouration of the reverberation when you listen in a reverberant room.
Exactly, that's in most cases the point why an uneven power response - as a result of "lobing" in the vertical plane - is seen as (let's say) "not perfect". Otherwise it really wouldn't matter much, if at all ...
And again, we have much more coloration from horizontal reflection that differ from the direct sound, then floor and ceiling bounce, which are easier to tame with carpets and absorption. In the end, I'd rather have a smooth off-axis response that looks like the on-axis, than a smooth vertical response, if I should choose between the two.
I wrote horizontal, but I meant slightly displaced in the front-back direction if needed to get the acoustic centres aligned.
Fair enough. I'm just pointing out what the theoretical features and limitations are, how important or uninportant you find those is up to you.
I don’t think you are comparing the right things. The horizontal off axis response of both topologies is the same. The power dip manifests itself in a small bandwidth. Tackling it with absorption on the floor or ceiling isn’t that easy.
I'm not saying that anything you write is wrong or unnecessary to think about. I merely have not yet tried to hear a speaker where I could say that it sounded better, because a given filter was used. I mostly find an even response and smooth off-axis to be the one that I can pick out as the better sounding speaker.
I have a Coax that I bought, which I want to try again. It should have an even dispersion, but the trick is to make it have an even response. Almost like I end up fighting the age old challenge of all loudspeakers - which is - they are all some kind of compromise 😀
I thought that the vertical reflections could be ignored to a bigger degree, since they hit both ears more equally. Whereas the horizontal reflections will have a much bigger impact, since they reach our ears at different timings, especially since our ears are on each side of our head. Is this theory somehow flawed, or again just a worthwhile compromise?
Imagine what it sounds like coming down at you from the ceiling with a big hole in the response near the crossover.. You know a hole makes the treble stand out by itself like some noise.
Nulls are the same with even-order Butterworth and Linkwitz-Riley. With odd-order Butterworth, the radiation pattern gets tilted, so you still have nulls, but at different (vertical) angles.
So we are back to the basics? less or no filters in the most sensitive range, from around 700 - 7000Hz, is the best solution. Something like Geddes' speakers, where we cross around 700Hz with lower order filters and minimum corrections.