I have started to look into the details of a crossover for my subwoofers to main speakers but this turns out to be less obvious than expected.
The mains have JBL 2226 woofers in 120 litre sealed boxes (~ 4 cu ft).
The box size was chosen to result in a system Q = 0.5 so the acoustic roll-off would naturally be Linkwitz/Riley 2nd Order, with the system 6 dB down at ~66 Hz.
Seemed like it should be easy to do the sub crossover, just add a 66 Hz LR 2nd order hi frequency roll off.
Of course the sub has a low frequency roll off too and this impacts the phase of the system around the hi frequency roll-off.
What I had not considered, because I've not seen it discussed much, was just how much this messes up simplistic solutions.
The result was that when I went in to Vituix to "fine tune" the crossover details I found I needed a completely different set up.
Is there some site or paper that analyses this that I never noticed?
I don't mean just blanket "the crossover needs to take the acoustic roll-off into account" statements but some sharper advice or recommendations.
At this point I would prefer to avoid a DSP solution and I am not sure I learn much from simplistic "just deconvolve the impulse response" software.
But if there's an informative DSP paper then I am ready to learn.
Here's the woofer details.
And here's the sub.
I can show the crossover details if anyone is actually interested.
David
The mains have JBL 2226 woofers in 120 litre sealed boxes (~ 4 cu ft).
The box size was chosen to result in a system Q = 0.5 so the acoustic roll-off would naturally be Linkwitz/Riley 2nd Order, with the system 6 dB down at ~66 Hz.
Seemed like it should be easy to do the sub crossover, just add a 66 Hz LR 2nd order hi frequency roll off.
Of course the sub has a low frequency roll off too and this impacts the phase of the system around the hi frequency roll-off.
What I had not considered, because I've not seen it discussed much, was just how much this messes up simplistic solutions.
The result was that when I went in to Vituix to "fine tune" the crossover details I found I needed a completely different set up.
Is there some site or paper that analyses this that I never noticed?
I don't mean just blanket "the crossover needs to take the acoustic roll-off into account" statements but some sharper advice or recommendations.
At this point I would prefer to avoid a DSP solution and I am not sure I learn much from simplistic "just deconvolve the impulse response" software.
But if there's an informative DSP paper then I am ready to learn.
Here's the woofer details.
And here's the sub.
I can show the crossover details if anyone is actually interested.
David
Attachments
One solution is to add the sub's phase response to the one of the mains. One way to do this is by using a highpass with the same rolloff as the sub on the main's signal path. The added benefit would be a reduction in power and excursion.
Regards
Charles
Regards
Charles
Hi Charles
Another possibility is to add all-pass phase correction to the mains.
I tried this because it makes use of more of the main's woofer output.
The 2226 woofer is already better than most so-called sub-woofers so it seems a waste not to use it as much as possible.
The sealed box naturally limits the excursion, it was chosen partly for this reason.
It works rather well, with a 66 Hz crossover the excursion is limited to a reasonable value just as the speaker hits it's thermal limit.
And -6 dB @ 66 Hz means it's a beast😉
Best wishes
David
Hi Dave
Yes, an allpass does of course work. Keep in mind that a highpass would only reduce the power of the 2226 where it's response has already fallen sifnificantly. IMO it is a matter of taste which one to use.
Regards
Charles
Yes, an allpass does of course work. Keep in mind that a highpass would only reduce the power of the 2226 where it's response has already fallen sifnificantly. IMO it is a matter of taste which one to use.
Regards
Charles
Yes, there are sensible reasons for either way.
Because the 2226 is so capable I think it can make a useful contribution in this case, even if it's response does fall.
The response of a vented box falls even quicker, so by 20 Hz they are back to about even.
They can add in phase so that means an extra 6 dB at 20 Hz, quite substantial.
I can post screen clips of some of my trials and you can see how they look, perhaps come up with a better solution.
Do you have Vituix?
It's excellent and already has the 2226 and 2245 in the driver database so it's easy to duplicate my work.
Or I can post the project files.
Best wishes
David
No, I don't use Vituix so far. The choice is a subjective one anyway. What does have more benefit - an increase of 6 dB below cutoff or the improved IM behaviour of your 2226 ?
It could be even made switchable so you can choose depending on source material !
Regards
Charles
It could be even made switchable so you can choose depending on source material !
Regards
Charles
The increase of 6 dB @ 20 Hz does help flatten the frequency response.
I think the evidence from Floyd Toole and Earl is that the flatter response is probably more important than a decrease in IM that will only be detectable at ear-bleed SPL.
But near 20 Hz the room modes are so dominant that flatness is a fantasy anyway, so it hardly matters.
I will probably do whatever is the neatest to implement.
But you made me think, so I tried to split the phase correction across a filter and an all-pass!
A first order all-pass combined with a 2nd order Butterworth does perform as expected, not quite as flat as the pure all pass but still reduces the excursion of the woofer.
It seems quite a happy compromise, thank you😉
Best wishes
David