X-over and phase confusion, help needed

[complet]

Im rather new to the DIY speaker world. I have buildt an 2-way floorstander from scratch, and learned a lot

But there is a few things about phase that confuses me, and I hope some of you could learn me som more.

I have worked with an xo that is 1.order plus zobel on the woofer and 2.order plus L-pad on the tweeter.
I can adjust the values to 2. order acoustic slopes, and get a good reverse null, measured at 1 metre.
What happens at 2, 3 and 4 metres? Are the phase still correct?

If I use a 1.order plus zobel at both low and highpass, should I still aming for a good reverse null?
A theoretical 1.order network should be 90 degr out of phase, right?
How can I check that my xo is correct then?

I have some more questions but hopefully someone could answer this first.
Thanks!

[rc white]

The results at one meter should apply for larger distances, neglecting floor and ceiling reflections.
A thing to keep in mind is that so called first order crossovers are really third order crossovers because the drivers have second order acoustic characteristics themselves, and the overall electrical plus acoustic response is what matters. Exactly how third order they are depends upon how close to the drivers natural roll off frequencies your crossover frequencies are and you need an overlap of at least two octaves to ensure that the crossovers resemble true first order characteristics, it is this that will determine overall performance.

[cjd]

Phase relates to acoustic slope (that's why the Hilbert transform works...)

And I'll step up and disagree about the slopes inheriting the driver's natural roll-off. That's only going to be true if you're crossing close enough to the edge of driver response to inherit this. And on anything other than tweeters, the roll-off is hardly ever anything resembling a 2nd order slope. Well, the bottom end roll-off of any driver in a box (that includes chambered tweeters) will be 2nd order. That's why you can sometimes get lower order electrical circuits to hit your target slope in the high-pass side.

Best way I know to check a crossover is with measurements.

C

[complet]

Phase relates to acoustic slope (that's why the Hilbert transform works...)

And I'll step up and disagree about the slopes inheriting the driver's natural roll-off. That's only going to be true if you're crossing close enough to the edge of driver response to inherit this. And on anything other than tweeters, the roll-off is hardly ever anything resembling a 2nd order slope. Well, the bottom end roll-off of any driver in a box (that includes chambered tweeters) will be 2nd order. That's why you can sometimes get lower order electrical circuits to hit your target slope in the high-pass side.

Best way I know to check a crossover is with measurements.

C

Could you please explain a little more how phase relates to acoustic slopes, and what that means.

And I'm talking about measuring, but is there only with a true 2.order xo it should have a reverse null?
With a first order network the drivers should be 90 degr out of phase, and give a peak of 3dB at the xo frequency, but will it be the same with the tweeter connection reversed? Or is there a better way to check if the xo is correct?
And this is only true if the driver are phase aligned?

[cjd]

Could you please explain a little more how phase relates to acoustic slopes, and what that means.

I could write a paper! This technically applies to a minimum phase system only, but all audio systems can be reduced to this (I think that's right). Apply the Hilbert transform to the response to get minimum phase. Apply physical offset (x, y, and z) relative to other transducers (you do this to ALL the drivers in a system), and apply your crossover, and you'll get the same results (if you get the acoustic offset correct) as using live data. Wikipedia has the math behind both the Hilbert and more generically on minimum phase. Probably a lot more. So I'm not going to try to reproduce that work.

And I'm talking about measuring, but is there only with a true 2.order xo it should have a reverse null?
With a first order network the drivers should be 90 degr out of phase, and give a peak of 3dB at the xo frequency, but will it be the same with the tweeter connection reversed? Or is there a better way to check if the xo is correct?
And this is only true if the driver are phase aligned?

Not just 2nd order will have the reverse null. Any system where acoustic phase is aligned (or should be) through the transfer region is going to have a null when you invert one signal. A third order BW has the signal 90 degrees out of phase, so it will not exhibit this kind of null.

There are a lot of things you can look at to determine how effective a crossover is (hitting an idealized slope, by the way, is probably the least important aspect in reality... ) The important targets will include a nominally flat (relatively speaking of course) or near-target flat response (for those that have a mild peak at crossover, you'd want to hit something near this). However, you can reach this many ways. So the next important aspect is that you don't have any *negative* summing - i.e. the final response is always equal to or greater than the response of any driver (or driver set).

From here, you start looking at phase tracking and off-axis response patterns. We've discussed phase tracking and, in fact, a lot of this other stuff, elsewhere here. Sometimes multiple times. (I think it may be in JonW's build thread, most recently?) So dig around, you'l likely find a lot more answers.

C

[dlneubec]

I don't believe that phase measured and designed at 1 meter will match the same way at 2 or espcially 3 meters. This has not held true with my designs and measurements and does not make sense physically. Assuming, for example, that you ear height is even with the tweeter and the woofer is below the listening axis. The three form a tiangle with the distance between the woofer and tweeter the short leg, the tweeter to ear, the long leg and the woofer and the ear the hypotenuse. The hypotenuse is always longer, but as you get out to 3 meters, the hypotenuse and the long leg becomes much closer to the same. In other words as you get out to 3 meters, there is very little difference between the distances from the two drivers to your ear, but at 1 meter, there can be a much more significant distance, IMO.

[fjhuerta]

I believe that 1st order crossovers (supposedly) "need" more distance in order to integrate coherently.

Then again, I've never seen a "true" 1st order crossover...

[Dennis H]

Yeah, what Dan said. If you want to measure an MT at 1m, measuring both drivers from the same point, you should measure a bit below the tweeter axis to compensate for the geometry. If you want to measure a tall WWMT tower from one point at 1m, good luck. Either measure each driver on its axis at 1m like JonM does or measure from one point at a bigger distance like CJD does.

[rc white]

The above is the frequency and phase plot of a first order filter used with a woofer and tweeter with a one octave overlap and Butterworth roll of's..




This is the same with one driver connected reverse phase, as can be seen frequency response is somewhat smoother. As can be seen the phase in the second case is also not as smooth and the overall response is far from all pass in both cases.
As far as cancellation goes the line of complete cancellation is straight at a particular angle from the baffle, i.e. the angles of reinforcement and cancellation are the same regardless of distance.

[complet]

I don't believe that phase measured and designed at 1 meter will match the same way at 2 or espcially 3 meters. This has not held true with my designs and measurements and does not make sense physically. Assuming, for example, that you ear height is even with the tweeter and the woofer is below the listening axis. The three form a tiangle with the distance between the woofer and tweeter the short leg, the tweeter to ear, the long leg and the woofer and the ear the hypotenuse. The hypotenuse is always longer, but as you get out to 3 meters, the hypotenuse and the long leg becomes much closer to the same. In other words as you get out to 3 meters, there is very little difference between the distances from the two drivers to your ear, but at 1 meter, there can be a much more significant distance, IMO.

This is exactly what I'm been thinking also. Thank you!
When measuring in a normal living room, its difficult to get a good measurement on a greater distance than 1 metre, because of reflection from floor and ceiling, thats why I asked.

[complet]

Thank you all for your answers!
I will read other build thread here also, but I try a couple more question:

When I measure a 2-way system at 1 metre, at tweeter axis, I'll get acoustic phase. (who looks like a mess). Is there any reason to get the minimum phase to work with, or is it ok to use it as is, in xo simulation?

When talking about 1.order network, and thinking we have drivers that will works well with 6dB roll off, will this work as it shoul, and sound good, without phase alligned drivers? (both drivers on a flat baffle)

Sorry about my english, it could be better, I know!

[dlneubec]

Here is what I've taken to doing. When doing in room, I measure at 1 meter to get a good read on the frequency response, but I also measure at 3 meters (or whatever the actual listening distance will be) to check and tweak the crossover for the best phase aligment using a narrower gated window. I don't just do measurments and simulations but mock up the crossover and do as built measurements to better tweak the FR and phase. I do check the FR at the listening position, but using a very narrow window to reduce the floor and ceiling bounce effects. Using a combination of these methods works for me.

[cjd]

When I measure a 2-way system at 1 metre, at tweeter axis, I'll get acoustic phase. (who looks like a mess). Is there any reason to get the minimum phase to work with, or is it ok to use it as is, in xo simulation?

If you are measuring all the drivers without moving the speaker or the mic or anything else, there is no reason to do anything other than work with the raw data. In fact, there may be reason to use this over a minimum phase model since you don't have to then set acoustic offset data into the software.

I usually end up dealing with both.

Also, I do tend to measure at various distances but end up using the 1.5 or 2M data on the 3-ways. You can tweak z-axis on the various drivers to get some idea of how things change phase-wise as you get further away. The trick is (as you've alluded to) finding the point where you get good data and not so much of the other noise contribution that exists from room reflections, etc. Measuring is its own confusing game, often frustrating.

C