Hi, some help please! I am running a three way active system with a horn top end. Obviously the cd driver is a distance behind the mid.
When implementing a dsp crossover it is a relatively simple matter to adjust for the time delay caused by the physical offset of the driver voice coils..in this case around 225mm which results in approximately 600micro seconds delay to the mid.
Using arta and Omnimic one can monitor the frequency response at the selected crossover frequency and watch the effect of the time delay dialling in an amount to achieve the flattest response?
But my question is this!! The dsp also has a phase control option in 10deg steps from zero to reverse (180 of course) . The filters are set to 4th order LR which have zero degrees polar tilt at crossover. Is there any requirement for Phase Control adjustment with 4th order LR once the time delay to account for driver offset has been set? I understand the signal can be adjusted to "exit" at the "same time" but may still not be "in phase"?
Is there any way of checking one is both minimum phase and time aligned? Both arta and omnimic will plot minimum phase against FR...but I'm not sure what I'm looking for in regard to this phase plot!!
Anybody help with this!
When implementing a dsp crossover it is a relatively simple matter to adjust for the time delay caused by the physical offset of the driver voice coils..in this case around 225mm which results in approximately 600micro seconds delay to the mid.
Using arta and Omnimic one can monitor the frequency response at the selected crossover frequency and watch the effect of the time delay dialling in an amount to achieve the flattest response?
But my question is this!! The dsp also has a phase control option in 10deg steps from zero to reverse (180 of course) . The filters are set to 4th order LR which have zero degrees polar tilt at crossover. Is there any requirement for Phase Control adjustment with 4th order LR once the time delay to account for driver offset has been set? I understand the signal can be adjusted to "exit" at the "same time" but may still not be "in phase"?
Is there any way of checking one is both minimum phase and time aligned? Both arta and omnimic will plot minimum phase against FR...but I'm not sure what I'm looking for in regard to this phase plot!!
Anybody help with this!
Hi, 4th order filters won't probably yield 4th order acoustic xo perhaps.
Here is how this is handled in the modern way, in short 😀
1. Measure all drivers individually with reference timing ( not sure if possible with omnimic )
2. process the measured responses with same windowing, preserving the timings
3. input the measurements into crossover design software like VituixCAD
4. tune the crossover in simulator, all of this included to see exaclty how it all plays out, Acoustically
See VituixCAD measurement manual for ARTA or REW https://kimmosaunisto.net/
In short yeah some delay on the mid is probably needed and good but it is hard to get it exactly spot on without the real measured phase. You could estimate it playing sound for the whole system and tuning the delay, see where it peaks. Remember it peaks +/- 1/4 wavelength at crossover. 1kHz is ~34cm long so +/-8cm would be close enough accuracy for the delay. If you have 2kHz xo, +/-4cm would be fine.
I believe passive xo systems would use different xo slopes to compensate the delay but with DSP we can just use delay.
Here is how this is handled in the modern way, in short 😀
1. Measure all drivers individually with reference timing ( not sure if possible with omnimic )
2. process the measured responses with same windowing, preserving the timings
3. input the measurements into crossover design software like VituixCAD
4. tune the crossover in simulator, all of this included to see exaclty how it all plays out, Acoustically
See VituixCAD measurement manual for ARTA or REW https://kimmosaunisto.net/
In short yeah some delay on the mid is probably needed and good but it is hard to get it exactly spot on without the real measured phase. You could estimate it playing sound for the whole system and tuning the delay, see where it peaks. Remember it peaks +/- 1/4 wavelength at crossover. 1kHz is ~34cm long so +/-8cm would be close enough accuracy for the delay. If you have 2kHz xo, +/-4cm would be fine.
I believe passive xo systems would use different xo slopes to compensate the delay but with DSP we can just use delay.
Last edited:
Hi,
Thanks...comprehensive explanation but didn't really understand all of it!
I thought for the delay setting one just measured the offset of the voice coils...in this case about 225mm, and applied the time delay calculated from that. Using Arta in FR mode, small adjustments of delay could be seen to effect the FR, the flattest response through
x- over being achieved at around 600microseconds which confirmed the physical offset of the voice coils as being around 225mm? It was also possible to see the two impulse response spikes become coincident at this setting?
Am I right in assuming the LR 4th order has the same delay in each leg so in theory Odeg phase setting would be the correct setting as long as the final acoustic response was 4th order from each driver?
Is it as "simple" as unwrapping the phase in Arta or reading off the phase angle from the "mag + phase" plot for each driver and applying the "difference"?
Thanks again for your help....
this might not be the exact delay, depending on your slopes. The delay is right when there is biggest hump, most constructive interference, and depending on the crossover and other features of your system this might be dip or hump or flat or what ever. It is easy to spot in a simulator when the phase info is preserved in measurements as per my previous post.
What is your crossover frequency? Reasoning from your quite a big waveguide I suspect it is < 2kHz in which case using the voice coil distance might be close enough approximation for the delay as I briefly mentioned in previous post you could have it within ~+/-4cm for about the same outcome, roughly +/-100us
Measuring drivers individually and comparing results is not going to work without timing reference, USB systems have delay which your computer OS can change at will which makes the phase / delay info between measurements unreliable. REW allows you to use acoustic reference which would work with the omnimic, not sure if ARTA has anything like it. ARTA users usually use two channel sound interface where the other channel is used as loopback for reference and the other has the mic connected. However, you can play both drivers simultaneously and look at the graph there. Or just use the 600us you've already figured out, it is probably close enough.
Hope it helps 🙂
Hi,
Yes, crossover is around 800Hz on a 90 x 40 horn.
Thank you for your detailed response, its been very helpful..
Yes, crossover is around 800Hz on a 90 x 40 horn.
Thank you for your detailed response, its been very helpful..