Phase tracking, wich one is better?

[requiem]

This



or this



the crossover is a 4th order LR at about 2300 Hz for both graphs

[TacoD]

I try to match the two lines, so the first picture should be better. There is less delay between the two drivers (smaller horizontal gap). If you switch polarity of one of the drivers you should see the graph moving. Also the frequency response nulls at the x-over point and you can expect a deeper null when phase is more aligned.

[requiem]

I see a much deeper reversed polarity null (more than 40 dB) when the woofer's and the tweeter's phase are *not* exactly aligned (second graph). When both phase-shift lines are closer or exactly aligned the reverse null is only 15dB below the target response.
Is this a normal behavior or its due to some simulation/measurement error?

[TacoD]

Maybe the FR is not symetric anymore, so summation fails. You need a symetric filter slopes to get the textbook nulling.

[requiem]

Maybe the FR is not symetric anymore, so summation fails. You need a symetric filter slopes to get the textbook nulling.

That was it, I modified a bit the xover to have symmetric LP and HP roll-offs and now the deepest null occurs when the phase is more aligned. Thanks TacoD :T

This is how it looks now



What else could be improved in this regard?

[Jed]

Don't get all hung up on phase characteristics, small variances are not as important as other design parameters. For example, final voicing and listening etc. Sometimes symmetrical slopes don't sound the best either and can create a peak in the power response. I agree that it is important to get things to look as good as posible on paper before the final tweaking however.

[requiem]

Thanks Jed I’ll keep that in mind. I just want to understand the most.
I guess shipping cost to foreign countries makes me a simulation-paranoid, as purchasing xover components can't be done in separate shippings.

[JonP]

I have a question on this... I've read that phase matching near the crossover is important, too. In other words, the part of the spectrum on either side of the crossover frequency.

If that's important, would you say that the first response was better, with a nearer woofer/tweeter phase match over a wide range, even though the crossover matching is a bit off?

I know this is just one part of many things that all need to be optimized, I'm just trying to get a feel for which parts are more important...

[Jed]

I have a question on this... I've read that phase matching near the crossover is important, too. In other words, the part of the spectrum on either side of the crossover frequency.

If that's important, would you say that the first response was better, with a nearer woofer/tweeter phase match over a wide range, even though the crossover matching is a bit off?

I know this is just one part of many things that all need to be optimized, I'm just trying to get a feel for which parts are more important...

Do you have FR and impedance/phase graphs you could show us?

[JonP]

Uh, I'm referring to the ones Requiem has posted in this thread...

[Jed]

whoops, shouldn't have quoted you, the question is actually directed to the thread starter.

[requiem]

I cannot say witch phase response yelds to better sound as I haven't build the xover or taken any real measurements, the phase data shown here was taken from existing measurements made by Bjorn Idland at the seas fabric for the CA18RNX woofer and 27TDFC tweeter. It's all simulated data and theory so far.

I would think that phase matching at the xover frequency is more important than at the rest of the spectrum since is at this frequency that both, the tweeter and the woofer, are playing loud and combining SPL for the final sound wave. At the rest of the spectrum one of the two drivers is attenuated and therefore it's contribution to the SPL is not so important.

I have read that a speaker can be a bit more detailed if the xover is phase optimised, but also that there are much more important specs to take care of as flat frequency response to begin with. However, if the frequency response looks as good when phase is matched as it does when its not, I see no reason not to match both drivers' phase as good as possible and see what happens.

Of course I might be wrong.

Cheers

[JonP]

No problem Jed...

I've noticed that with a design I'm trying to simulate, sometimes you don't have the flattest response coinciding with the closest phase match. (of the crosspoint) I'm guessing that's due to the filter slopes not being textbook perfect.

I have read that having the phase of the two drivers matching over an octave above and below the crossover point is a desirable thing... I was wondering if the wide range matching is more or less important than a perfect intersecting crossover point. Having both would be best, but if faced with one or the other, which way to lean?

Still hoping that one of the guru's might drop in and comment on that.

[Paul H]

...
I have read that having the phase of the two drivers matching over an octave above and below the crossover point is a desirable thing... I was wondering if the wide range matching is more or less important than a perfect intersecting crossover point. Having both would be best, but if faced with one or the other, which way to lean?

Still hoping that one of the guru's might drop in and comment on that.

No gurus here yet, so I'll add my 2 cents ...

Wider range matching is much more important - having the same phase at a single point is not important.

What you're looking to do is have the drivers work together (in phase) in the crossover frequency range. When the drivers are in phase your on-and-off-axis responses will be smoother. Drivers working out of phase will create negative interference (sound cancellation of each other) which can result in spotty and uneven distribution of sound at various angles from the speaker on-axis direction.

How much of the frequency range must be in phase (or close) is a matter of opinion - an octave either side of the crossover point may be a good target, but may be tough to achieve. Filters with steeper slopes have a smaller crossover range, so the phase-matched range may be smaller too, all else being equal.

[TacoD]

Off-axis behaviour is another thing, but my experience is that with most setups a more flat response sound better than great phase tracking.

Requiem, I am working on a 2.5-way with Seas CA18RNX & 27TDC (same tweeter without ferrofluid). Also somewhere I have a filter for a 2-way CA18RNX and 27TDC tweeter. So if you have questions you can contact me.

[requiem]

Thanks TacoD, actually there's one thing that keeps me wondering...

First, I've been using two different existing measured data for the seas CA18RNX, one provided by Bjorn Idland, the other traced from zaphaudio.com site. After summing my own baffle simulation using FRD consortium tools I get a different response using both source measurements. The main difference is the baffle step compensation required to achieve a flat response. With zaph's measurements I have to use a 4mH inductor for the low pass filter, while with Bjorn's data only 2.2mH. I don't think I made mistakes with FRC tools because when I recreate zaph's Waveguide design the simulation matches just fine (using zaph's traced data and in the bass section only of course) with the measured response shown at the site.
The baffle i want to use has a width o 22cm and the xover is a 4th order LR at 2000Hz, the box is 19L ported tuned to 45Hz.


Which xover is more correct given your experience with this driver?

[Dennis H]

After summing my own baffle simulation using FRD consortium tools I get a different response using both source measurements.

I think that's to be expected. I think Zaph measures with a quasi infinite baffle, in a room, on the driver axis. Bjorn measures in an anechoic chamber, driver in a box, with the mic on the tweeter axis. So, both baffle step and phase will be different for the two methods. I'd expect Bjorn's measurements to be more accurate given all the gear at his disposal but you do need to read his measurement conditions carefully and possibly adjust for them when you do your model.

SEAS FABRIKKER AS
15.11.2004

The files in the sub-folders are frequency response and impedance data stored in .txt files.
The data is tabulator separated and listed in the following order:
Frequency Magnitude Phase

These measurements are made in a box that can be used as a 2-way speaker. It is important that you note this difference if you compare these measurements to the datasheets for our drivers. The boundry conditions are different, especially on the tweeter measurements, so here you will see diffraction effects.

The object of this project is to give our customers access to measured data that can be used to simulate a crossover. While using LspCAD for some time I have found a very good correlation between simulated and measured end-result, so I hope this will make it easier for you to construct your own speaker.

Measurement conditions
The speaker is placed on a turntable-stand in a 4-pi (free-field) anechoic chamber.
Applied voltage is 2.83 volts. Measurement system is ATB precision. All measurements are done with a stepped sine at 250 frequency points. No gating. Smoothing at 1 (lowest setting- approx. 1/12 oct.)

The microphone (B&K 4133) was placed 1 meter from the front of the speaker, on the tweeter axis. The microphone was not moved during the whole measurement session. The woofer measurements are also done with the microphone centered on the tweeter axis. The reason for doing this is that the relative phase difference between woofer and tweeter will be accurately stored in the measurements.

To verify your simulation setup you can do the following (I'm referring to LspCAD):
Go to the General menu and import the following file in the "Target SPL curve"
L18RNX_P & 27TFFC in paralell-SPL.txt
This file shows a measurement from 100Hz to 20kHz with a woofer and tweeter in paralell without any crossover components attached.
Set the measurement distance to 1 meter.
Then import the IMP and SPL_on axis files for the same drivers located in it's respective sub folder.
L18RNX_P H1224-08_IMP.txt
L18RNX_P H1224-08_SPL_on axis.txt
27TFFC H881-06_IMP.txt
27TFFC H881-06_SPL_on axis.txt

The simulated sum of these two drivers without any crossover components and connected in paralell should match the "Target SPL curve" above 100Hz.

Off-axis measurments is also included in the folders. To get a reliable result in simulation you should always import both woofer and tweeter files for the same angle. Again, this is to get the phase difference correct. All off-axis measurements are in the horisontal-plane. Now you have a chance to see both effects of driver directivity and diffraction in your simulation of an off-axis response.


Box
The box outside dimensions:
Front baffle: 20cm wide and 49.6cm high
Box depth: 29.6cm
Edges are trimmed with a roundover-bit r=12mm.

Driver placement:
Both drivers are centered on the vertical axis.
The distance between the drivers centers are 158.5mm
The distance from the top of the baffle to the tweeter centre: 65mm
The distance from the top of the baffle to the woofer centre: 223.5mm

The box has a removable tweeter baffle. This way we can measure all drivers flush mounted in the same box.
The box is approximately 18 liters and the port tuning is approximately 40Hz. All woofers are measured like this, regardless of what the T/S calculations suggest.

Good luck with your speaker building.
Remember to check that the drivers are available at your local distributor.

Bjorn Magne Idland
Lab technician
Seas Fabrikker AS
bidland@seas.no

[requiem]

Thanks Dennis, I checked the settings in Baffle Diffraction Simulator and forgot to adjust the on axis distance to 10cm for my baffle, witch should suit zaph's near field infinite baffle measurement. Now both responses look quite similar except for the bump at 670 Hz that I get with Bjorn's measurements.

I think i'll give this design a try as it is and see how precise this simulation results when its finished.

This is the predicted response using zaph's measurements, now the baffle step compensation needed is the same as with Bjorn's measurements:



As for the phase goes, I tried to keep it aligned at xover frequency, and as close as possible 1 octave up and down from that frequency, probably won't make any difference anyway :B



Any comments welcome.

TacoD it would be good to see your xover design as well.

Thanks for all the help.

[TacoD]

I believe the measurements form Bjorn Idland they are in agreement with my own measurements (only with a lot more resolution ). Measurements in cabinets are more useable than tracing plots and using BS sims. Almost all 2-way loudspeakers crossed between 1500 and 2500 Hz use something between 1.5 mH and 2.5 mH as series conductor. An inductor of 4 mH is way off imho.

In my setup I use 1.8 mH + 13.3 uF and a zobel (10 uF + 10 Ohm) on the woofer. My x-over point was ~1800Hz. I use a different tweeter, but I used someting like 10uF + 0.56 mH. The Seas is works at a lot of x-over points.