The biggest improvement I noticed going to fully active XOs was when I added mid-tweeter phase correction. Sure, you can accomplish much the same thing by physically moving the drivers, but my designs end up with a higher SAF when I keep my boxes looking conventional. She who must be obeyed is happier and my boxes are easier to build. sounds like a win-win to me.
While I agree that the active XO must be designed for the specific application, my feeble mind finds it easier to design active baffle step compensation, phase correction/all pass filters and notch filters than passive ones. I now own/have built more amplifiers than I can ever use, so that cost is not an issue for me.
While I agree that the active XO must be designed for the specific application, my feeble mind finds it easier to design active baffle step compensation, phase correction/all pass filters and notch filters than passive ones. I now own/have built more amplifiers than I can ever use, so that cost is not an issue for me.
BobEllis said:
Actually those are two separate things.
The phase correction/time delay actually just alters the phase of the signal in the same way a normal xover does but it doesnt change the TIME of the signal. Both the tweeter and the midrange will still both produce sound at the same time but the phase of the situation will be sorted out which is mostly what matters.
Time aligning still requires moving the drivers into physical alignment. So yes when you need to alter the phase you can do it physically by moving the driver but you will also add in time alignment also. The active circuit only alters the phase. For the best of both worlds you want to time align the drivers physically and then use a phase circuit to give you correct phase for the slopes/topology or the high and low pass.
Somewhere else it has been noted that time aligning is not that critical i.e we cannot hear the time difference between the small difference from mid to tweeters. Also moving the tweeter back can add in deffraction effects you may wish not to have. I think I have written this before oh well.
Fifth,
We may be differing on semantics, but as I understand it you are using time alignment to refer to physically aligning the acoustic centers of the drivers. I agree that this is not the ultimate solution, but helps get close. IMHO this is really an attempt at phase alignment that is often called time alignment - probably a carry over from marketing lingo that the masses could understand.
On Sigfried Linkwitz' site, in discussing phase compensation, he says that he uses it to compensate for the driver's high pass response AND the physical distance between the drivers. The goal is to have a smooth phase transition between drivers at the crossover frequency rather than absolute zero phase.
This phase change can be accomplished electrically or by physically moving from the time aligned position by an amount that shifts the phase of the signal received by the listener the proper amount at the crossover frequency. This phase alignment is only valid at the crossover frequency, since at other frequencies the physical distance between drivers is a different portion of a wavelength. But then, the same is true of the electrical solution.
As you noted, there is plenty of evidence to show that similar signals arriving at the listener's ear fractions of a millisecond apart are indistinguishable as seperate events, but that does not mean that the character of the perceived sound will be the same as the relative phase of the two signals change.
I agree with you on the impact of diffraction effects of using a recessed tweeter. That is why if I try for a physical alignment I use a sloped baffle or a bent one like that used by JMLab. Now you need to design for off axis response of at least one driver and maybe rebuild a cabinet if you get it wrong. It's much easier to change a resistor in the phase compensation filter, hence my active XO preference.
We may be differing on semantics, but as I understand it you are using time alignment to refer to physically aligning the acoustic centers of the drivers. I agree that this is not the ultimate solution, but helps get close. IMHO this is really an attempt at phase alignment that is often called time alignment - probably a carry over from marketing lingo that the masses could understand.
On Sigfried Linkwitz' site, in discussing phase compensation, he says that he uses it to compensate for the driver's high pass response AND the physical distance between the drivers. The goal is to have a smooth phase transition between drivers at the crossover frequency rather than absolute zero phase.
This phase change can be accomplished electrically or by physically moving from the time aligned position by an amount that shifts the phase of the signal received by the listener the proper amount at the crossover frequency. This phase alignment is only valid at the crossover frequency, since at other frequencies the physical distance between drivers is a different portion of a wavelength. But then, the same is true of the electrical solution.
As you noted, there is plenty of evidence to show that similar signals arriving at the listener's ear fractions of a millisecond apart are indistinguishable as seperate events, but that does not mean that the character of the perceived sound will be the same as the relative phase of the two signals change.
I agree with you on the impact of diffraction effects of using a recessed tweeter. That is why if I try for a physical alignment I use a sloped baffle or a bent one like that used by JMLab. Now you need to design for off axis response of at least one driver and maybe rebuild a cabinet if you get it wrong. It's much easier to change a resistor in the phase compensation filter, hence my active XO preference.
This is interesting then because I can use the delay circuit to alter the phase such that most, if not all, orders or slope and frequencies yield the correct phase response for a good null on reverse polarity. However all those filters will not require (or require more) the amount of delay necessary to correctly "Time Align" the two drivers.
Another point (im not speaking fact here im just thinking out loud) if you use an inverting circuit it will completely invert the signal without adding in any delay, yes it "flips" it thru 180 degrees but it doesnt actually mess with the time. Correct me if I am wrong here. Now I am of the understanding that with the all pass network you just alter the number of degrees you can shift the signal by but not actually alter it in time ie make the signal take longer to reach the tweeter.
With a time aligned speaker isnt the whole point to have the drivers radiating in the same Z axis so that the fronts of the sound waves arrive at your ears at the same time? (It seems a trivial thing considering it only works for one angle vertically you will be listening at.) To manage this you will need to actually put the drivers in physical alignment, then apply an all pass network to alter the phase of the system for the desired filter you choose.
I am saying this because the tweeter will always require say 10us of delay to bring it into the same Z plane as the woofer. But some filters (highpass lowpass configuration) may require 5us or 20us to correctly phase align them, what happens now? you have the system phase aligned so that you achieve destructive interference on reversed polarity, but you have more/less delay then required to bring the tweeter into the same Z axis as the woofer. What do you do, alter the delay to the 10us? that would time align but mess up the phase alignment.
If the delay all pass network didnt work in time but just altered the phase of the signal then you could achieve both, you physically align the drivers so that the signal always radiates from both drivers in the Z axis. Then you apply the correct amount of phase shift to phase align the xover you have chosen.
Is this not how it all works?
Correct me if I am wrong please, thats how I understand it, I dont wanna go around with some wrong idea in my head.
Another point (im not speaking fact here im just thinking out loud) if you use an inverting circuit it will completely invert the signal without adding in any delay, yes it "flips" it thru 180 degrees but it doesnt actually mess with the time. Correct me if I am wrong here. Now I am of the understanding that with the all pass network you just alter the number of degrees you can shift the signal by but not actually alter it in time ie make the signal take longer to reach the tweeter.
With a time aligned speaker isnt the whole point to have the drivers radiating in the same Z axis so that the fronts of the sound waves arrive at your ears at the same time? (It seems a trivial thing considering it only works for one angle vertically you will be listening at.) To manage this you will need to actually put the drivers in physical alignment, then apply an all pass network to alter the phase of the system for the desired filter you choose.
I am saying this because the tweeter will always require say 10us of delay to bring it into the same Z plane as the woofer. But some filters (highpass lowpass configuration) may require 5us or 20us to correctly phase align them, what happens now? you have the system phase aligned so that you achieve destructive interference on reversed polarity, but you have more/less delay then required to bring the tweeter into the same Z axis as the woofer. What do you do, alter the delay to the 10us? that would time align but mess up the phase alignment.
If the delay all pass network didnt work in time but just altered the phase of the signal then you could achieve both, you physically align the drivers so that the signal always radiates from both drivers in the Z axis. Then you apply the correct amount of phase shift to phase align the xover you have chosen.
Is this not how it all works?
Correct me if I am wrong please, thats how I understand it, I dont wanna go around with some wrong idea in my head.
Yes that's a good one but I don't really consider it a "commercial" product - more a kit for hobbyists like the stuff available from Elliott or Linkwitz. I was referring to the typical analog pro XOs like the Behringer mentioned here or, for that matter, the cheaper Marchands which don't have all the extras. The XM-44 isn't plug-n-play if you want to use all its features. It will require measurement and customization for the particular speaker.
Time and phase are two different ways of looking at the same thing. As 5th says, you need to compensate for both the physical alignment of the acoustic centers and the group delay of the driver/filter combination. Linkwitz uses allpass networks on the tweeter to do both and get a good phase match through the XO region. The test is to reverse the polarity of the tweeter. You should get a deep null if you've done it right.
Right although usually passive crossovers fudge a bit from a textbook LR, either by changing Fc, the Q or the slope of the filters, to get good phase alignment. But with an active analog filter you can easily do a straight time delay with an allpass so that's another arrow in the designer's quiver. And of course, straight digital delays are common with DSP boxes. Some, like Dan Wesnor over on the Mad board, like doing passive allpass stages but it adds to complexity and not many people are doing them.
- Status
- Not open for further replies.