Driver phase difference at crossover point.

[Boscoe]

Just heard a few things about phase differences at the crossover point just wondered if this is the case as I've looked into a few simulations. Can we ignore this problem with active crosssovers?

Thanks
Boscoe.

[Pano]

Sure, you can ignore it in passive crossovers, too.

Do you want to?

[cocolino]

Originally Posted by Boscoe

Just heard a few things about phase differences at the crossover point just wondered if this is the case as I've looked into a few simulations. Can we ignore this problem with active crosssovers?

Thanks
Boscoe.

Whatever phase difference "problem" arise with passive crossovers, the same issues happen with active x-overs.
"Active" has a few advantages over "passive" but that doesn`t mean that "active" voids the usual and inevitable frequency / phase response relationship of passive filters (unless not so common "substraction" type active crossovers are involved).

[DBMandrake]

Even with a subtraction filter phase is important because drivers may not be (probably are not in most cases) time aligned, introducing additional phase shift.

A crossover can't really be designed without measuring the actual phase shift of each driver on the design axis so that all contributing factors towards phase (crossover response, driver response, time alignement) are taken into account.

Ignore the phase response and chances of getting the wanted response are slim...

[Boscoe]

How do you go about obtaining the phase difference? What's the best way of minimising the difference? Is it only remedied by delays?

[weltersys]

Originally Posted by Boscoe

How do you go about obtaining the phase difference? What's the best way of minimising the difference? Is it only remedied by delays?

The phase response of drivers and filters can be seen with dual FFT test gear like Smaart or TEF.
Physical offset, digital delay, all pass filters, using different filter types and slopes for the lower and upper crossover, in and out of band EQ may be used to achieve the desired acoustic crossover and phase alignment.

The "best way" would be a matter of opinion and available options with the gear chosen, and the reproduction use. There are many ways to skin the phase cat.

Art

[Boscoe]

I intend to use two HiVi M6a's with a Fountek NeoCD30 I have found the phase response for the M6a but not the Fountek could someone help me there? Can I assume that the phase will be the same in the plot in free air as in a ported box at 2.2KHz?

Thanks

[theSuede]

You HAVE to stop thinking "phase", and start to think in units of "time".

Phase does, with the addition of a few more parameters GIVE you "time", but it is not equal to "time".

The 6" driver is deeper set, the sound at ~2kHz may eminate from a point well behind the speaker baffle surface.
The Fountek has a radiation center that's about 1/2" behind the baffle surface.

At 2200Hz, one inch difference is the same as a 60º phase shift.

The easiest way to optimize phase shift in reality is to adjust the slopes of the x-over until you see that the individual drivers sum POSITIVE at all points, and that the addition at the exact x-over point is as close to 6dB higher than the individual drivers as possible.

It's also important to remember that the electrical phase angle has to be combined with the driver's natural phase angle to make a meaningful simulation. The RESULT of x-over + driver response gives you the phase.

This is especially important when you cross a tweeter close to its' own fall-off point. Most tweeters have a 2'nd order slope with a certain resonance point and Q. When you combine this with a 2'nd order x-over you get a 4'th order system.

[speaker dave]

Originally Posted by Boscoe

How do you go about obtaining the phase difference? What's the best way of minimising the difference? Is it only remedied by delays?

You need to measure the driver and crossover response in a way that shows the phase as well as amplitude. Something like Holm impulse will do that. When you see the phase curves of the various sections (woofer, tweeter) you will see how close together or far apart they are.

Phase is just signal arrival time expressed in terms of degrees of a circle. The two wavefronts (woofer/tweeter) can arrive in phase (0 degrees, 360 degrees, etc. apart) in which case they will add constructively. If they are more than 90 degrees apart they will cancel leaving a hole in the response in the crossover region (although they may add okay at some other angle).

The phase shift (or rotation) is inherent in any rolloff due to passive filter, active filter, or even the natural rolloff of the drivers. Also, because the woofer usually has more depth than the tweeter, it will have considerable extra phase shift.

The objective of speaker design, in a nutshell, is to manipulate the crossover response so that driver and crossover together have the right amplitude and phase shift to add together seamlessly.

David

[wintermute]

For a visual example (I need to see pictures ) have a look at my measurements below.

The original measurements were taken using holm impulse exported and imported into speaker workshop. measurements for both mid basses and tweeter were taken with the mic in the same place. Holm was set to zero-locked after taking the tweeter measurement (this ensures that the time zero (and hence phase) stays constant relative to each driver.

The crossover was then modelled in speaker workshop using impedance measurements and the spl measurements.

The very first crossover did not take into consideration phase (in a serious way), and I was not happy with the results. the second one did and the results are in the second attachment (and I'm very happy both subjectively and objectively).

What you see in the second attachment is I believe the sort of thing to aim for (the phase is the dotted lines) note how that track closely to each other from around 1.3K to a bit below 5K. No it wasn't easy getting that, but I think it was worth the effort. The difference with this new crossover compared to the first is like chalk and cheese!

How much of it is down the the phase, and how much due to other factors I'm not sure, but I do think that the matching of phase through the crossover region (crossover point 2.8Khz) is more than likely one of the reasons that this crossover is subjectively much better than my first effort.

Now that was with a passive crossover. How do you think that those drivers will sum with an active crossover that for example, simply applies a specific electrical 4th order crossover at 2.8Khz? I strongly suspect it won't sound anywhere near as good

Tony.