[richidoo]

In a 1st order loudspeaker crossover with one inductor for the low pass filter, and one capacitor for the high pass filter, are these two filters always in phase with each other (phase coherent) throughout the crossover band regardless of the individual filters' -3dB frequencies?

Put another way, must there be a shared -3dB knee frequency for both the high and low pass filters in order to maintain coherent phase on their output through the crossover band?  Or can the crossing amplitude be shifted away from -3dB in order to affect EQ without losing phase coherence?

If phase coherence is sensitive to crossover frequency and parts tolerance, then could small bypass caps by used to fine tune phase coherence?

Thanks
Rich

[JohnR]

The two filters are not in phase, but the phase shift (and amplitude) is such that the two outputs sum flat and have no net phase shift on-axis.

If you move one of the filters then they won't sum that way. You'd have to move it a fair bit to make a practical difference.

If you're talking about EQ, be aware that (most) EQs will cause phase shift. No free lunch.

[*Scotty*]

In order to achieve transient perfect behavior from the crossover you also have to physically align the acoustic centers of the drivers so that at some distance in front of the speaker the drivers output sums perfectly and a 1kHz square-wave is a least recognizable.
  Before you worry about whether the crossover is phase coherent or not check to see that the drivers have sufficient overlap in the crossover region to allow a true 1st. order crossover. Most of the time the drivers themselves have an acoustical roll off of 6dB/oct or more to which the electrical slope of 6db/oct. must be added which results in a crossover with a slope of more than 6dB/oct., there goes your phase coherent crossover.
Scotty

[Æ]

You have the choice of three crossover network types: linear phase (also know as minimum phase), all pass, and non-all pass. Only one type of crossover fits the minimum phase criterion, the first-order Butterworth. The first-order crossover produces a summed zero phase response which is flat at all frequencies and has a flat magnitude response.

[andyr]

Only one type of crossover fits the minimum phase criterion, the first-order Butterworth. The first-order crossover produces a summed zero phase response which is flat at all frequencies and has a flat magnitude response.

I thought if you used a symmetric, 1st order XO formed by a series inductor on the bass driver and a series cap on the tweeter, which have the same knee frequency ... then you get a 3dB signal boost at this XO frequency?

EG. for an 8ohm driver, if you use a 1.25mH inductor and a 20uF cap then the knee frequency of each filter is 1KHz which sums "above" flat?  So to try and get a flat result, you have to spread the 2 knee frequencies apart slightly.  (If you use something like lspCAD, you will see that you can get a pretty flat summed frequency curve by arranging the -3dB frequencies so that the -6dB point on each filter's roll-off curve is @ 1KHz.)

And is there any other 1st order XO except a Butterworth? 

Regards,

Andy

[*Scotty*]

This link has more information on filter types and behavior. Apparently the first order filter does not have the 3dB hump at the cross-over center frequency.
http://sound.westhost.com/articles.htm
http://sound.westhost.com/articles.htm
Scotty

[richidoo]

I'm interested in the electrical phase only at this point, but good points made about acoustic interference, driver resonance and impedance affecting the overall crossover design and perceived phase coherence.

I know Linkwitz-Riley passive crossovers demand -6dB crossover amplitude as part of the recipe that yields electrical phase coherence through the filter band, and so I suspect that the 1st Butterworth is also sensitive in electrical phase to the crossing amplitude, since they still use caps and coils.

By phase coherency I mean that the electrical output of the low pass lines up on a scope in perfect phase with the electrical output of the high pass for all frequencies through the filter band. I know they both shift 90 degrees together.  I think that for any given coil there is only one value cap that can make a phase coherent crossover with it, assuming equal, flat loads. Is this true?

Thanks for the link to Elliott pages, Scotty, I'll read those (again.) It's more fun to ask here. 

[*Scotty*]

richidoo,the acoustic response we listen to and measure is the result of the crossover's electrical characteristics and the drivers mechanical behavior. The crossover can be electrically correct and sum properly when the electrical output is viewed on a scope and yet when the speaker's acoustic output is measured it does not have a flat response in the crossover region and it does not sum correctly at the crossover point.
 This has occurred because the drivers had insufficient overlap on either side of the crossover point and the combination of the crossover's electrical slope and the drivers acoustical rolloff resulted in an acoustical slope of greater than 6dB/oct.. The resulting phase shift of more than 90 degrees guarantees that there is no way that a phase coherent crossover can be achieved.
 In fact the only way the crossover can now work is if the combined electrical slope and driver rolloff results in a 12dB/oct. slope which allows the drivers or the tweeter in the case of a two-way loudspeaker to be connected in reverse phase which put the drivers back in phase at the crossover point.
Scotty

[richidoo]

Thanks Scotty.