Feature request? Subwoofer phase measurement with band limited timing reference

[r0b]

tl;dr:
Is there a way to band-limit the acoustic timing reference e.g. to 80-120 Hz so the data makes (relative) sense when crossing a sub to the mains?

Hi there,

quick Intro first:
my name is Robin. I'm from Germany and I was bit by the HiFi bug 20 years back, when an old car battery, car radio and speakers found on the street mounted on a cart were the equivalent of today's bluetooth speaker. Over the years lots of stuff passed though the living rooms. Recently I finished building some new Visaton Studio 2 speakers for my L-C-Rs. L and R are standing each on a side-firing subwoofer equipped with a Visaton GF200 (Visaton btw. has a superb customer service, so I stuck with them after outfitting my third portable speaker with Visaton speakers I bought in our local electronics shop).
Picture attached, room is acoustically rather friendly. LP is free roughly on the apex of an equilateral triangle.

Current situation:
The Studios (passive XO) and the subwoofers are xo'ed with a Monacor DSM 260 digital speaker management system (2 in / 6 out, equivalent to Behringer DCX 2496 at a different quality level). XO point is LR12 db @ 130 Hz for both systems. All is driven by a 6-channel Philips DFR 9000 AVR(Studios bi-amped, woofers direct). Now I am working on the time and phase alignment of woofer to midrange. Yesterday I spent two hours measuring (UMIK-1 available) and getting so frustrated I didn't even save the measurements anymore (a fact I do regret a bit now). There was just no winning in terms of figuring out the phase or even the timing delay between the two sources. Having the woofers sideways doesn't necessarily make it all easier. For example I took measurements at the listening position equally distanced from R and L mains. I measured left subwoofer with left main as timing reference (possible with routing in the DSM) and the same thing of the right. One side says 7 meter delay, the other 3,5 meter or so (physical distance maybe 30 cm). Just all seemed pretty random.

My problem:
I spent about 3-4 hours reasearching every note and thread here, on hts, on avsforums, on german forums concerning time alignment of subwoofers. I failed to find the one concrete tip in evaluating the band-limited impulse response that I get from my subwoofers with respect to my full range sharp peaked mains impulse response.

- I get why acoustic timing is necessary for the UMIK-1.
- I get why it doesn't work with the band limited sub.
- I think I get that I actually need to have the xo's set in the DSM to measure the phase, because otherwise it doesn't make sense.

- I don't necessarily get frequency dependent windowing, but have a feeling near-field measuring and FDW might get me closer (measurement position on straight line from my ears to between mains and midwoofer?).
- I DON'T get how to extract timing or phase information from my system.
- Group-Delay plot? ETC plot? Can't figure out what to make of them, although I read a PDF suggesting using that for alignment.

- Am I on the wrong path asking for a band-limited acoustic timing?


Pointers to any helping threads are very welcome, and I don't fear reading thorugh pages of technical disussions. Trust me though when I say I did google and read at least 5 or 6 detailed walkthrough PDFs/posts concerning Subwoofer alignment (including the ones linkes in the REW intro pdf).

I feel a bit bad for putting out such a long text and I hope it's still concise enough, but I do have the feeling that this here is the playe where I am closest toan answer.

Thanks for reading anyway!

Regards,

Robin

P.S.: I guess this thread is somewhat less useful without measurements. I'll try to get some done these next few days to discuss. Maybe just a pointer as to setup and what to measure will already be enough to get me on the right track ...

 

[r0b]

Ok, this is worth so much more with measurements, so I did some quick ones on my left speaker in near field setup (20 cm from mid-range cone, see pictures.)

The DSM260 is connected with a USB-Audio-Card digitally via coax (no AVR in between). From there to the 6-Channel amp and into the speakers.

I took measurements of the individual chassis both with xo and without, and changed the timing ref around. I hope it is clear from the notes I took.

For all the measurements:
- I only have the midrange with its passive XO playing (called "Main" in the measurements).
- The subwoofer is set to 180° phase in the XO (the midrange is not).
- The Subwoofer has -2db and the Midrange +1,5db in the XO.
- No delays or other relative gains anywhere in the system.
- The mesaurements with sw+main were measured at 6db less output level due to clipping in nearfield. Mic position was the same.

 

[jtalden]

I used your first 2 measurements for analysis.
Below is the phase tracking as measured. The acoustic XO is at about 100 Hz and the XO range from about 70-150 Hz so that is range to consider for phase tracking.

​

A copy of measurement #1 (the SW) was made thus becoming #9. Its impulse was then shifted 3 ms (less delay) to achieve better phase tracking with the Main #2.

The main #2 impulse and the shifted SW-copy #9 impulse were then summed as #10 (A plus B) to view the resulting SPL (the SW SPL was also increased 6 dB for this operation). Measurement #7 was overlaid on this SPL chart for comparison.

This is phase tracking approach is a common target. Other good alignments are possible though. The relative polarities chosen were correct as measured to provide the closest phase tracking given the XO, polarity and phase settings used in the DSM260.

- I don't necessarily get frequency dependent windowing, but have a feeling near-field measuring and FDW might get me closer (measurement position on straight line from my ears to between mains and midwoofer?).

FDW was not necessary due to your close measurement distance. The direct sound phase traces were relatively easy to read. I usually have mic placed on line to the LP about 1 - 1.5 m from the 2 drivers. This would make no significant difference however in this case since the XO is at a low100 Hz. It would be more of an issue for a 2 kHz XO.

- I DON'T get how to extract timing or phase information from my system.

The overlay phase chart provides the phase comparison the 2 measurements as shown. If the 2 impulses are very far from 0 ms the phase chart is difficult/impossible to read. In that case both impulses need to be shifted the same XX.XXms to place then near 0 ms. In this case no shift was needed.

- Group-Delay plot? ETC plot? Can't figure out what to make of them, although I read a PDF suggesting using that for alignment.

Several other good methods are available to determine delays. This is the only one I have much experience with.
​

 

[r0b]

Hi jtalden,

thanks for the quick input on my measurements. I played the example through just now on my PC, just shifting the mains 3ms the other way (cause that's what I'd do on my DSM). It did end up the same, yay!

View attachment 12579
A copy of measurement #1 (the SW) was made thus becoming #9. Its impulse was then shifted 3 ms (less delay) to achieve better phase tracking with the Main #2.

What I meant by "extracting information" is: how did you come up with the 3ms delay? For me it is not clear from the impulse or the step graph.



Also, here it looks as though the phase is (wrongly) inverted on the SW. Or does the direction of the first impulse not correlate with with being out of phase?

Much appreciated,

Robin


Edit: I think the barrier in my head is the following: REW creates an "offset" for the timing data of my SW relative to mains (acoustic ref), right? This offset is somewhat arbitrary due to the very band limited data REW has to work with. Isn't the phase information just random then as well?! Or are these phase graphs actually (somewhat) true to the real world phase of my system? (Please say yes )

 

[jtalden]

Yep, +3ms delay on the mains is the same result!

Arbitrary Timing?
The REW phase response is correct for all measurements and is referenced to 0 ms.

> With loopback and acoustic timing disabled REW will automatically adjust the impulse back near 0 dB. It is helpful for several of the charts; phase, group delay, spectrogram, etc, to have the impulse placed at 0 ms. This is not useful for delay timing analysis between driver measurements however. The relative timing delay between 2 different driver measurement is lost. This mode can be thought of as arbitrary for timing as the amount REW shifts each measurement can differ based on several factors.

> With loopback timing active the reference channel fixes 0 ms when REW sends the measurement signal. Thus 2 driver measurements will retain the same relative timing. They will both be shifted away from 0 ms due to the time of flight and any measuring system delays such as and DSP.

> With acoustic timing is active REW will set 0 ms when the reference sound is detected. If the same reference speaker is used, 2 driver measurements will again have the same relative timing, but may be shifted away from 0 ms due to any difference in time of flight between the arriving measurement sound and the reference sound.

The shift away from 0 ms can make the phase difficult/impossible to read to shifting both measurements back to near 0 ms is necessary. To keep the relative timing, both need to be shifted the same amount.

Extracting Information?
It is possible to extract the information is several ways. I only use the phase tracking method for delay timing as I feel it provide more detailed information as to what is going on throughout the entire XO. It is also probably more confusing to use than some of the other methods. I will let others explain the alternate methods of delay timing.

In this phase tracking analysis, given that it is a ~100 Hz XO, I knew that 1 ms increments is a good offset to use for trial and error impulse adjustments. It is a good balance of efficiency and accuracy. That is, we can find good tracking with a minimum of trial offsets. The process is just to move the SW impulse in 1 ms increments until the phase of the 2 are aligned in the overlay chart. Invert the SW impulse in REW to see if the helps or hurts the tracking situation. With experience this can go very quickly. It's

hard to explain all the​

little tricks to get there quickly in this Post. It just takes some experience. One tip is; don't try to keep track of the how many impulse shifts or the increments used to get the best tracking. We can just use the impulse overlay chart of the original SW and the SW-copy to measure the amount that was shifted. Control-Shift with a right mouse drag works well on Windows to measure the offset,

Wrong Phase?
Yes, the polarity of the SW is very likely inverted based on the initial negative rise in the step response. This is the best choice in this case. It achieves the closest phase tracking. It is the normal situation to choose the best polarity relationship based on measured/modeled results. When you indicated the DSM260 was set for 180° phase, I did not know if that meant the polarity was inverted, or if that was a phase shift control of some type. Regardless, +3 ms mains delay with this relative polarity is correct for the given measurements. If significant changes are made to the XO point, polarity, or other XO settings the timing will be affected.

If you prefer, very good results can also be achieved by inverting the SW to positive polarity and adjusting the mains delay timing roughly +8ms instead of +3 ms. Adjust the timing so the phase crosses at the XO frequency. We could do this analysis as above to determine a more accurate timing change.

 

[jtalden]

Per the tip, here is how REW can measure the final change in impulse timing.