Same ol´question (on phase)

[lasse]

Hello,

I am modelling the Xo:s for the RFL-30L (heard about them before??) and decided to NOT cut the corners this time, but to find out how to do it properly.

It´s about removing excess phase from a measurement. This is what I´ve done (so far):

• Measured the drivers in box from 150 cm distance
• Asked my DCX2496 of the delta AC:s between the drivers
• Pulled the minimum phase from the SPL-curves to 50 Khz
• Started the simulation of the XO:s

The distance 150cm corresponds to 4.4 ms flight time (Baffle/Tw to mic)
The DeltaAC Tw-Wo is 0,18 ms
The DeltaAC Tw-Mid is 0,1 ms

When I model the drivers in SpeakerWorkshop, should I remove 4,4 ms from the Tw, 4,5 ms from Mid and 4,58 ms from Wo?
If yes, should I on top of this model the drivers with their DeltaAC (physical offset) in the XO-modelling?. Doesn´t this mean that this distance (DeltaAC) is taken care of twice?

I know that this topic has been dealt with before but obviously, I have limited skills of perception in this area.

Besr regards//
Lars Rydén
Stockholm, Sweden

[curt_c]

Presuming you measured all three drivers from the same point: To keep the relative phase relationships the same between the drivers, remove the same excess phase from each driver measurement. Ie: 4.4 mS. Removing the difference in AC’s between the drivers would align the AC’s of the drivers in the sim, but they aren’t aligned in real life.

As a sanity check: The summed response of the three drivers w/o removing the excess phase should be the same as the summed response of the three drivers after removing the excess phase.

C

[lasse]

Presuming you measured all three drivers from the same point: To keep the relative phase relationships the same between the drivers, remove the same excess phase from each driver measurement. Ie: 4.4 mS. Removing the difference in AC’s between the drivers would align the AC’s of the drivers in the sim, but they aren’t aligned in real life.

As a sanity check: The summed response of the three drivers w/o removing the excess phase should be the same as the summed response of the three drivers after removing the excess phase.

C

Hello Curt and thanks for the reply. That´s what I thought. Removing, in this case, 4.4 mS will be the way to go.

I just realized that, up until now, I have used near-field data for the woofer and that is what´s in the "raw SPL-curves" in another post on this project. For the Wo with near-field-data, I suppose that the phase-info will have to do as-is? Or should I base this on Wo measurements from 150 cm? (I made that measuremenst also). I suppose that I could "add" 0.18 mS to the near-field data to end up att baffle distance?

Regards//Lasse

[curt_c]

One could theoretically take the nearfield woofer response, subtract its time of flight, and then add back in the .18mS but besides the possible errors in doing this (knowing the exact distance from the mic to the baffle, for instance) the nearfield plot will not capture the diffraction effects necessary for proper summation with the mid and tweeter. –Most notably, the baffle step correction. It will be best to use the 150 cm measurements for all 3 drivers. Since the mic was in the exact same position on all three measurements, the phase differences due to the differences in acoustic centers are already in the measurements. One merely needs to subtract the same time of flight from each measurement to reduce the amount of excess phase displayed. –This is not necessarily a requirement, and a proper crossover can be designed without doing so, but it makes the the phase plots less ‘busy’.

C

[lasse]

One could theoretically take the nearfield woofer response, subtract its time of flight, and then add back in the .18mS but besides the possible errors in doing this (knowing the exact distance from the mic to the baffle, for instance) the nearfield plot will not capture the diffraction effects necessary for proper summation with the mid and tweeter. –Most notably, the baffle step correction. It will be best to use the 150 cm measurements for all 3 drivers. Since the mic was in the exact same position on all three measurements, the phase differences due to the differences in acoustic centers are already in the measurements. One merely needs to subtract the same time of flight from each measurement to reduce the amount of excess phase displayed. –This is not necessarily a requirement, and a proper crossover can be designed without doing so, but it makes the the phase plots less ‘busy’.

C

Hello Curt and thanks again for the feedback.

If you are correct (and quite certainly, you are!) in that the phase info and the delta-AC is "buried" in the measuremens when the mic is in the same position in all three measurements, why are we told to adjust the position in AC in the modelling-SW, e. g. SpeakerWorkshop? Or is it correct NOT to adjust the driver position when the curves come from a measurement as describer here (mic in same position)?

Believe me, I´m learning a lot from this.

Best regards//Lars

[curt_c]

I can’t speak for SW, as I’ve never used it…

Since the difference in phase of each driver is already contained in the measurements, there is no logical reason to add it again.

The only time I can think of when it is necessary to add in the difference in acoustic centers is when the phase measurements of all three drivers are already minimum phase. (I.e: The Hilbert transform has been applied to the phase of each driver.) The differences in AC then have to be added to the Z axis for each driver.

C

[lhwidget]

Lars, the last time I measured drivers for XO design, I mounted the drivers in the enclosure and measured the tweeter and woofer on axis, from the same distance without changing the test levels.

I removed the excess phase due to time of flight from both measurements (both drivers were flush mounted, I used the distance from the baffle board to the mic). I was using SoundEasy, which adjusts the on-axis responses using a routine which models the drivers as they are mounted in the baffle. I'm not familiar with S-W, measuring both drivers from the same mic position may be the best way for you.

After those two measurements, I took the near field response of the woofer and merged it with its far field measurement. If you're using a port, you could merge its near field response to the woofer's far field response in the same way.

If you can't merge response curves, I'd just use the far fileld measurements for XO design.

[lasse]

Hello and thanks for the feedback,

My measuring-software, the SIRP, doesn´t supply phase-info in the response so after measuring, I extracted (is that the word?) phase, using Frequency Response Modeler 2. I beleive that this uses Hilbert-Bode FFT transform so the phase info should be "minimum phase".

It seems that the correct thing to do is to remove 4.4 mS of flight time from all three responses, using the far field measurement for the woofer, and not bother to correct for the delta AC´s

As a "sanity check", I will do the same thing but also compensate for the extra flight time and see if there is a difference worth bothering about. I will supply the pictures in this thread in a few hours.

Best regards//Lars Rydén
Stockholm, Sweden

[dlr]

Hello and thanks for the feedback,

My measuring-software, the SIRP, doesn´t supply phase-info in the response so after measuring, I extracted (is that the word?) phase, using Frequency Response Modeler 2. I beleive that this uses Hilbert-Bode FFT transform so the phase info should be "minimum phase".

It seems that the correct thing to do is to remove 4.4 mS of flight time from all three responses, using the far field measurement for the woofer, and not bother to correct for the delta AC´s

As a "sanity check", I will do the same thing but also compensate for the extra flight time and see if there is a difference worth bothering about. I will supply the pictures in this thread in a few hours.

Best regards//Lars Rydén
Stockholm, Sweden

Curt is correct, you need to adjust for offset, but in your case with minimum-phase data, you have to add an offset that is the geometric distance from drivers to design point. This includes all axes, x, y and z. If one driver's axis is the design axis, then the geometric distance will be any (x,y,z) offset. That is, the origin of the design axis tweeter is (0,0,0) and the other driver will have horizontal and vertical displacement on the baffle plane plus a z-offset for the acoustic center difference.

You can determine the relative acoustic offset by using the 3-measurement method described at my site. The only difference is that for your case, you have to add the step of using separate software to calculate the HBT phase, whereas the examples of mine use CALSOD that calculates it automatically if using a model rather than measurements.

The reason that you can still use that method is that, as Curt has stated, the phase is included in all measurements. This means that you measure all at the same time/power/mic position. You have to measure the individual drivers and the summed response (all done raw). Be careful of the tweeter, of course.

The summed response will have the phase of the combined drivers. You model the drivers, extract (good terminology) HBT phase of each (only valid for those models, don't change them afterwards), then sum these two in CAD raw (no crossover). They will at first not be close at all to the measured summed response due to invalid excess-phase delta (delta only, absolute is immaterial at this point).

You then leave one driver alone, e.g. tweeter as (0,0,0) if it's the design axis.
This will set the driver AC as the reference point. The mid or woofer is than set initially as (x1,y1,z) where x1 is horizontal offset and y1= vertical offset on the baffle. z is then adjusted until the summed response of the two driver models matches the summed response measured. Phase can be ignored because once the relative offset is correct, the excess-phase delta will be correct.

If you only adjust the z-offset (using a single value for the phase delta), you can use it, but in that case if you want to investigate the off-axis response in CAD software, you'll no longer have a correct phase delta at other and off-axis CAD predictions will be invalid. If you follow the steps above for (x,y,z) settings, then the off-axis response will show the changes for the crossover influence off-axis. I say that because driver directionality and diffraction will still change off-axis, but at least you'll have a much better idea of the off-axis.

Also, if you decide later that you want the design axis to be any other axis, say halfway between drivers, all you need do is set that in CAD. The phase delta will be correct having set the models precisely in space relative to each other.

dlr