Sorry about that. 300 is a typo. It should have been 3000Hz.150 and 300 Hz ?
That doesn't make much sense for a 3 way.
Linkwitz Riley just has a filter Q of .5
Far as the very generic filter topology of a 2nd order 3 way
View attachment 1029634
To answer your first question, yes the topology of the filter is the same.
And yes the inductor and capacitor values would slightly change.
To change the Q of the filter cutoff.
Linkwitz Riley Filter has a Q of .5
Butterworth Filter has a Q of .707
Additional components could include padding resistors to match driver levels
Impedance compensation, Baffle compensation, notch filters etc etc.
Every driver is different.
Generic filter calculators usually don't yield optimal results
Might be more useful to describe your project
and what drivers your using.
How much? I see many good measuring loudspeaker like Magico and Revel using LR4 and having pretty good power response. So with the right baffle and driver placement, it should be possible - right?Ideally, Butterworth crossover filters keep the sum of the powers from the drivers constant. That causes a peak in the on-axis response with even-order filters and depending on the driver placement, a rotated main lobe with odd-order filters.
By design, Linkwitz-Riley filters keep the magnitude of the sum of the sound pressures on axis constant, ideally anyway, but cause a dip in the sum of the powers.
So, when I see a small dip around the crossover frequency, and a small bump off-axis - that's the LR4 filter in action?The dip is -10 dB * log10(2), which is about -3.01 dB. The peak in the on-axis response with even-order Butterworth filters is +10 dB * log10(2) ~= +3.01 dB. Both are obviously theoretical figures, that you get when there are no other imperfections.
https://heissmann-acoustics.de/dxt-mon-182/
Has become? Alison is the first i know of that really started pushing the idea.Probably not generally known, but since we understand the importance of the power response (thx Toole e.a.) this has become quite relevant in speaker design.
This is one of the issues with BSC filters. They raise the on-axis to flat, raising the power response up above flat.
dave
Yep, since ‘80s… 😊
Earlier. https://www.audioholics.com/news/a-tribute-to-roy-f.-allison
Allison 1 came out in 1976.

dave
So, when I see a small dip around the crossover frequency, and a small bump off-axis - that's the LR4 filter in action?
https://heissmann-acoustics.de/dxt-mon-182/
Theoretically, the response with a Linkwitz-Riley filter should be flat on axis and have a dip vertically off axis when the drivers are mounted above each other. That off-axis dip should be quite small when the difference between the distances to the drivers is well below a quarter of a wavelength. When you power average the responses in all directions, then the dip should be -3.01 dB.
If you can get the drivers within quarter wavelenght you are likely dealing with a WAW not a traditional multi-way.That off-axis dip should be quite small when the difference between the distances to the drivers is well below a quarter of a wavelength
Once they get that close the drivers become essentialy coincident and spacing related lobing goes away.
dave
Last edited:
Thanks for the info but they both look the same so is the difference in the values of the caps and inductors?
Yes. The topology is the same, but the behaviour is determined by the values of caps and inductors.
I mean the difference between the distances to the listener or measuring microphone. That's only equal to the distance between the drivers when you are (vertically) 90 degrees off axis.If you can get the drivers within quarter wavelenght you are likely dealing with a WAW not a traditional multi-way.
Once they get that close the drivers become essentialy coincident and spacing related lobing goes away.
dave
Last edited by a moderator:
I think it is the vertical distance between the drivers times the sine of the angle that you are vertically off axis, assuming that you are listening at a large distance and that the drivers are point sources that don't have any phase shift of their own and that are placed straight above one another. (I did write theoretically in post #29, didn't I?)
Using the end of the triangle wher eyour ears are is fraught with the head in a vise issue.
Physically within a forth wavelength, lobing disappears everywhere.
dave
Physically within a forth wavelength, lobing disappears everywhere.
dave
If the driver signals were 90 degrees out of phase, like with an odd-order Butterworth set-up, you could get a deep notch when the difference between the distances to your ears is a quarter of a wavelength (or a 6.02 dB bump, depending on whether you get constructive or destructive interference, which depends on whether the tweeter or the woofer is closer to your ears, among other things). With Linkwitz-Riley, you get -3.01 dB at a quarter of a wavelength difference between the distances, no matter whether you are closer to the tweeter or to the woofer.
More generally, if the phase difference you get due to path length differences is alpha, you get a factor of cos(alpha/2) in the amplitude of the sum when the drivers are driven in phase and a factor of cos(pi/4 +/- alpha/2) when they are driven 90 degrees out of phase. That makes the magnitude response more sensitive to small path length differences when you use odd-order Butterworth than for Linkwitz-Riley or even-order Butterworth (which has a 3.01 dB bump on axis).
More generally, if the phase difference you get due to path length differences is alpha, you get a factor of cos(alpha/2) in the amplitude of the sum when the drivers are driven in phase and a factor of cos(pi/4 +/- alpha/2) when they are driven 90 degrees out of phase. That makes the magnitude response more sensitive to small path length differences when you use odd-order Butterworth than for Linkwitz-Riley or even-order Butterworth (which has a 3.01 dB bump on axis).
Butterworth is maximally flat amplitdue and Bessel in maximally linear delay. LR is somewhere in between.
That 6.02 dB should be 3.01 dB.If the driver signals were 90 degrees out of phase, like with an odd-order Butterworth set-up, you could get a deep notch when the difference between the distances to your ears is a quarter of a wavelength (or a 6.02 dB bump, depending on whether you get constructive or destructive interference, which depends on whether the tweeter or the woofer is closer to your ears, among other things).
Are you suggesting we look at this with the condition that the drivers are a similar distance from the listening position? I'd think that's the only way to do it unless otherwise stated.and have a dip vertically off axis when the drivers are mounted above each other. That off-axis dip should be quite small when the difference between the distances to the drivers is well below a quarter of a wavelength.
Or am I missing the point?
Yes. I assume a conventional set-up where the tweeter is placed a short distance above or below the woofer, slightly horizontally displaced to align the acoustic centres if needed, and the ears of the listener are at about the same height as the drivers and at a distance that's much greater than the vertical distance between the woofer and the tweeter. The path lengths from the woofer to the listener and from the tweeter to the listener are then almost equal.
If there is a large difference between the path lengths from the drivers to the listener, like when one of them is a subwoofer placed at the other side of the room, then the assumptions of Linkwitz and Riley don't hold. I don't see any particular advantage to using Linkwitz-Riley filters under those conditions.
If there is a large difference between the path lengths from the drivers to the listener, like when one of them is a subwoofer placed at the other side of the room, then the assumptions of Linkwitz and Riley don't hold. I don't see any particular advantage to using Linkwitz-Riley filters under those conditions.
- Home
- Loudspeakers
- Multi-Way
- what is the difference between linkwitz-riley crossovers and butterworth crossovers