TOPIC:

AIUI - please tell me if my understanding is skewed - there is always delay involved in a conventional IIR XO:
* 6dB slopes result in the 'lower' driver being delayed 90 deg, compared to the 'upper' driver.
* 12dB slopes produce a 180 deg delay - so one of the drivers is generally connected in reverse polarity, to compensate for this
* 18dB slopes produce a 270 deg delay and
* 24dB slopes produce a 360 deg delay - ie. a whole cycle ... so the drivers can be connected with the same polarity.

But what if we use an active digital XO - like say, a miniDSP 10x10HD - to provide the IIR filters?

This offers delay - my question is ... should this delay be used, to deliver a better result?

For instance, let's take a 24dB XO between mids & tweeter @ 3kHz:
* for simplicity, let's assume the speed of sound is 1000fps
* wavelength at 3000Hz is 1000/3300 = 0.33ft
* so the mid driver is delayed 0.33ms behind the tweeter.

Should we therefore set the delay on the tweeter in the 10x10HD plug-in to be 0.33ms, so as to time-align the tweeters with the mids?

Andy
 

If you have a physical offset of the acoustic centers, then an added delay is appropriate.
However, rather than computing a value, I would implement it empirically using the reverse-polarity method and adjusting for the deepest null at the crossover frequency.
That works for even-order crossovers......which most seem to be these days.

Dave.

Thanks, Dave.

Yes, I'm aware of a delay used to counter a physical offset (that's why I currently have the 'mains' delayed by 3.13ms).

Your suggestion to use "the reverse-polarity method and adjust for the deepest null at the crossover frequency" (rather than simply calculating) is interesting - I will try this when I next get my mic & REW out.  

And, yes, I'm using even-order XOs.

Andy
 

Thanks, Dave,

That was very neat; using the reverse polarity method; the REW graph showed more dip when I delayed the tweeter.
(Unfortunately, I don't seem to be able to add an image here.)


Thanks for your help, Dave.
 

Yeah, more dip would be expected.
But the task is to adjust the delay higher/lower to yield the 'maximum' dip at the crossover frequency.  Then, when returning the polarity to normal, you have the best phase match across the crossover region.

Remember, this only works for even-order crossovers.

Dave.

I didn't know this. Thank for the info.