Any time you double the number of drivers performing the same task (like what happens in the middle of an ideal x-over), there is a doubling of sensitivity, so if you build the x-over with a power response goal, there will pretty well always be a ~3dB hump at the x-over as a result of the improved sensitivity there provided the listener is on-axis equa-distant from the drivers in question. I suppose by purposely selecting different slopes for each driver and combining that with driver placement it may be possible to get a mixed breed of power response and frequency response both staying reasonably flat as there are some losses to phase issues. I suspect that such an approach is probably not a good idea as it means that there is likely to be an off-axis peak at x-over somewhere in the room and that could make it's way to the listener as an unwanted reflection. Such a condition might screw up some buzzwords.
I have to ask, is there a stubborn valve amp involved?
I have to ask, is there a stubborn valve amp involved?
Last edited:
It is inherent in the nature of in-phase crossovers that two units will be contributing at the crossover point and there will be a 3dB gain of axial sensitivity over the individual responses. Assuming the units are fairly flat in their power response in the octave around the crossover point, then there will be typically a 3 dB hole at the crossover point. We primarily choose flat axial response and spread the crossover points such that the units are 6dB down at crossover (rather than, say 3 dB down for a quadrature summing approach).
Before everybody goes of debating the merits of flat power vs. flat axial for the next 1000 posts, let me point out that one thing the Toole studies always show quite clearly is that the minor holes in the power response found at the typical (in-phase)crossover point never preclude a speaker from being top ranked in any listening comparison.
David S.
Before everybody goes of debating the merits of flat power vs. flat axial for the next 1000 posts, let me point out that one thing the Toole studies always show quite clearly is that the minor holes in the power response found at the typical (in-phase)crossover point never preclude a speaker from being top ranked in any listening comparison.
David S.
I'd just like to highlight keyser's point - most people assume that in phase crossovers (adjusted for flat on axis response) always have a 3dB hole in the power response.
This is only true when the driver spacing is half a wavelength or more at the crossover frequency. As driver centre to centre spacing is reduced below half a wavelength the power response hole is progressively reduced until by the time you get below about 0.1 wavelengths there more or less isn't a power response hole.
So if you had two drivers spaced say 200mm apart and crossed over at 4Khz they are several wavelengths apart and will have a 3dB power response hole. (In addition to any driver off axis characteristics) The same driver spacing with a 250Hz crossover frequency will have almost no power response hole despite the in phase crossover because they're only a small fraction of a wavelength apart.
As for tilted lobes in a quadrature summing crossover (true odd order crossover) I have to agree. I used to fall on the side of odd order crossovers from the perspective of power response but in the last year my opinion has changed 180 degrees (pardon the pun) to in phase tracking for a number of reasons, including avoiding off axis peaks in the response.
As above though, an odd order crossover will only have off axis peaks / offset lobes if the drivers are half a wavelength or more apart - if they are very close together (<0.1 wavelengths, eg woofer to midrange in most designs) it doesn't really matter whether you use in phase or quadrature summing filters - both will sum flat on axis, flat power response and will have a uniform lobe.
The even order L/R will result in 3dB less power dissipation near the crossover frequency though - as quadrature summing will "waste" some of the power due to incomplete addition.
Yes.
Hmm.. That's always been my thought about coaxials.. that if phase is identical off-axis, why would there be a power response dip? I guess there isn't one, then.
For example, imagine affixing two pieces of string to a baffle, one where a tweeter goes and one at the centre of the woofer placement. Cut one string shorter by a half wavelength at some experimental frequency. Hold the two loose ends together between two fingers and pull them out and up, or out and down until they are both taut, hence delineating the lobe.
Using this technique on a coincident pair of drivers where each string is tied to the same baffle location, shows that in theory they especially need to be brought into phase, and that they will sum at all angles.
Using this technique on a coincident pair of drivers where each string is tied to the same baffle location, shows that in theory they especially need to be brought into phase, and that they will sum at all angles.
Coaxials are a great example of what I'm talking about. Because the two drivers are concentric the centre to centre spacing is zero, thus no power response hole due to using an in phase crossover.
Of course the driver itself - the low frequency portion, may start beaming if the crossover frequency is too high causing a power response dip, but that's due to the driver directivity, not the crossover phasing.
I'm not sure that your string analogy says that the two coincident / coaxial drivers must be driven in phase. All it says is the relative phase won't shift as you go off axis. (Ignoring phase shift off axis due to beaming of the larger driver causing high frequency rolloff...)
If the two coincident drivers cross over at 0 degrees and -6dB or 90 degrees and -3dB the on axis result will still be flat and the off axis response between the two crossovers will be nearly the same as well since the only variation in phase off axis will be a result of driver beaming.
(Because the 0 and 90 degree crossovers will have different slopes, there will be a slight difference in the way that the beaming of the larger driver affects the summed response off axis)
I hadn't thought much about drivers closer than a half wavelength although I think the point is valid. In practice it becomes hard to achieve, so it is more theoretical than real.
Using a coax as an example is oddly the one case were phasing between drivers doesn't matter. Quadrature summing would remain fixed for all angles and would therefore have no downside (At least in theory. Most drivers will have incidental phase shift off axis from directivity).
David S
Using a coax as an example is oddly the one case were phasing between drivers doesn't matter. Quadrature summing would remain fixed for all angles and would therefore have no downside (At least in theory. Most drivers will have incidental phase shift off axis from directivity).
David S
So does all this mean that I'm ok with my old Tannoys?
The xover point is where the woofer beams at 90deg and the tweeter has a 90deg conical horn.
I'm running them active and my xover allows for 360deg relative phase adjustment between bands but I have never heard great differences when playing with it.
However I also use a supertweeter mounted above the DualConcentric and that is very sensitive to playing with the phase.
PS: Slopes are 24dB L-R.
The xover point is where the woofer beams at 90deg and the tweeter has a 90deg conical horn.
I'm running them active and my xover allows for 360deg relative phase adjustment between bands but I have never heard great differences when playing with it.
However I also use a supertweeter mounted above the DualConcentric and that is very sensitive to playing with the phase.
PS: Slopes are 24dB L-R.
Not that hard. Our MTM achieves driver spacing on the order of 1/4 w/l
http://www.diyaudio.com/forums/css/176790-el166-mtm-ml-tl.html
dave
Also not that hard to achieve in a woofer to midrange crossover, even with physically quite large drivers.
For example a 12" woofer and an 8" midrange spaced fairly close together allowing say 5cm clearance between the frames would be a centre to centre spacing of 30cm, which at a 250Hz crossover would be about 0.22 wavelengths.
Naturally smaller more closely spaced drivers would be much less again. So I think for woofer to midrange crossovers the power response dip of in phase crossovers is going to somewhat less than 3dB in the typical case since driver spacing is quite small in wavelengths.
Midrange to tweeter crossovers on the other hand are going to tend to have well over half a wavelength spacing in most instances, in some cases several wavelengths if the crossover frequency is quite high, so here there will certainly be a power response dip for non-coincident drivers using in phase crossovers.
On the other hand there is some reason to believe that a power response dip at higher frequencies is far less audible than it would be at lower frequencies like 250Hz where we're perceiving the room response more than the on axis response...
That's the ideal setup for a concentric design yes - cross it over at the frequency where the woofer directivity reduces to 90 degrees, and as the woofer cone itself acts as a kind of a 90 degree waveguide for the tweeter the directivity of the two will be matched at the crossover frequency.
If the drivers are perfectly concentric if you vary the phasing between the two drivers you'll just be altering the overall amplitude response near the crossover frequency (which should still be audible mind you) but not the shape and direction of lobes as would be the case with non-coincident drivers.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.