Timing info question

[Greg Dunn]

This is such a great tool that I've donated twice already.

When attempting to time-align my system, I've noticed a couple of things which I'm not sure I'm analysing correctly. I'm running my mains (Magneplanar Tympani 1-C) through a 31-band EQ and then into my power amps. The sub is also EQed but then there is a miniDSP before the power amp used for delay because the sub is several feet from the mains due to space issues. I've attached the info screens for the left, right, and sub sweeps.

First of all, the right speaker has a delay equivalent to several inches of distance (I can move the mic and the number doesn't change much at all). I know a large planar speaker doesn't have a single location that can be pinpointed as a source, but I've measured the distance to the LP from each of the 6 panels and the differences are much less than 4 inches left to right. The impulse from each speaker is pretty clean and in the measurement, so I'm not sure how REW is calculating the relative distances? It's not an issue audibly, more of a curiosity.

Secondly, the sub has a delay set in the miniDSP, which is 4 mS based on prior measurements and physical distance. I can hardly believe that I got a value that is accurate to 3 decimal places! I know that the impulse from the sub is very much less crisp than the ones from the mains, but it seems that REW would therefore have a much looser tolerance on what it can calculate as a delay. Is my analysis wrong? One reason I'm extra curious about this is that when measuring the sub without the miniDSP delay enabled, I got a delay of about 6.6 mS - so the miniDSP native processing delay of about 2 mS was probably still active.

 

[John Mulcahy]

You'd need to post the mdat with the measurements for folk to comment. Subwoofer impulse responses are broad and slow to change compared to main speakers, commensurate with their limited bandwidth, so delay estimation is based on where the response appears to start. Aligning subs is more easily done by comparing their phase with the phase of the main speakers through the crossover region, the Alignment tool in beta 29 helps with adjusting delays for that.

 

[Greg Dunn]

Here's the mdat file. I'm pretty comfortable with the absolute phase of the speakers, but if anyone can comment on the timing values, it would help me understand what's going on with that.

 

[jtalden]

The timing handoff from SW to L, R Mains is very good. The chart attached shows the phase at the 60 Hz XO target is the same. The phase slope from the SW to the Mains is also about the same so there is good SPL support throughout the XO range.

Additional info FYI:
Normally it is better to sweep the SW and Mains the same frequency range rather than only sweeping above and below the XO. That way we can see the overlap range of the SPL and the phase tracking through that overlap range. I would suggest an absolute minimum range of ±2 octaves from the XO. For a 60 Hz XO the minimum sweep range would be 15-240 Hz. I normally just recommend a full-range sweep of 20-20k Hz - whatever is the normal overall response sweep. Most other setup analysis measurement are full-range so it is then not necessary to keep changing the sweep range.

If the sweep range was overlapped a suggested, my experience suggests we would have seen that above 60 Hz the SW phase trace curls up a little further away from the Mains phase trace. I would expect that another maybe 1 ms (1 ft) of delay to the SW would be needed to provide the ideal phase tracking. This is a trivial difference that would no impact to SPL, or have any other significant impact to sound quality. You are good to go as is.

 

[Greg Dunn]

Yes, I understand that the time delay is approximately correct - that's audible as well as measurable. My question was about why REW gave me a significant offset (about 4 mS) from one main speaker to the next, and a completely 0.000 offset from the sub to the left main. When I measured full range on the mains and 10-300 Hz on the sub, I got about 6 mS between mains and about 3 mS from the sub to the left main (after DSP delay has been added). It seems like the time delays are heavily dependent on the measurement sweep range of the individual transducers, and the difference between the mains makes no sense - there is clearly not 6 inches of differential distance between each main and the LP. So REW is calculating these in a way which makes the delay values not useful for my system and apparently the phase plot is the only way to accurately determine them.

 

[John Mulcahy]

The REW estimate looks pretty close, about 81 mm offset lines them up well. Restricting the range of the measurement affects the timing calculations, so measure the mains full range, not from 60 Hz. It's fine to use zero as the start frequency.

 

[Greg Dunn]

Thanks John, I figured it was related to how the timing calculations were done!