4th order Digital XO with Time delay - Question

[Docks]

Assuming I have the correct drivers and everything in place. Why would someone choose a digital 1st order XO over a 4th or 8th order L-R with time delay?
Thanks!!

[CharlieLaub]

Originally Posted by Docks

Assuming I have the correct drivers and everything in place. Why would someone choose a digital 1st order XO over a 4th or 8th order L-R with time delay?
Thanks!!

I think that the reasons are all in the time domain response:

1. A 4th order LR has a small group delay peak near the crossover frequency. This becomes pretty significant for 8th order LR. For first order, there is no group delay at all I think...
2. Also, few people realize that both HP and LP crossovers of 2nd order and generate group delay, and that this always occurs below the crossover frequency. So, a LP filter has a passband (frequencies lower than the crossover frequency) with delay and a stop band without delay, while a HP filter has a stop band with delay and a passband with no delay. The result is a differing amount of group delay above and below the crossover point in the system response. This "distorts" the time response of the input, even if the system frequency response is perfectly flat. Luckily, the ear is not all that sensitive to this, up to total delays of around 1-2 mS. Note that total delay also includes delay from driver offset.

-Charlie

[Pano]

As Charlie says, plus my on observations:

Remember that you choice 1st order electrical it could be a much higher order acoustic slope.
My current crossover is 4th order Butterworth at 800Hz acoustical. But electrically it's 3rd order and spread a lot to achieve the acoustic target.

By choosing 1st electrical order you could be at 2nd, 3rd or even higher acoustically. That can be OK, but 1st order crossover don't attenuate much distortion out of band. So although the response may drop fast, there can still be a lot of distortion where the driver is getting signal, but not making much output other than distortion. This is easy to hear and measure with some 1st order filters. Going to a steeper filter can keep the driver from being excited where it outputs mostly distortion.

So for me, choosing 1st order means you have drivers that are (very) well behaved far away from the crossover point. If they aren't, then you may achieve your target slopes, but have a lot of distortion. In passive crossover they are a quick and easy choice that can sometimes work well. First order would be great, if drivers were better behaved.

[Docks]

I REALLY want to create a phase linear speaker. IIR seems to currently provide that without hassle.
1st order on the dcx seems also to provide this option. How do you feel about 4th order LR with delay?
I am also worried about potential damage and distortion to the tweeter.
I want to integrate scanspeak 6600 or 6620 with a mid Please help!

[Davey]

Docks,

I hope you're not confusing phase-coherent with phase-linear.

With anything but 1st-order acoustic it's not possible to create a linear-phase result using analog or IIR. You can achieve a phase-'coherent' response using the DCX2496 because it allows pure delays to the drivers. Or you can achieve a phase-coherent result in the analog domain by using all-pass filters.

But, you can't achieve a phase-linear result using IIR digital boxes or analog crossovers with anything but 1st-order acoustic crossovers. (Actually there are a few exceptions, but that's generally a true statement.)

A "4th order digital crossover with time delay" would be phase-coherent, not phase-linear.

If I were you, I would abandon the linear-phase goal.

Cheers,

Dave.

[oublie]

you can to a certain extent achieve what you want.

This would mean routing all your sources through a computer and using a vst plugin to actively cross and equalise your drivers. This tries to acomplish linear phase and also allows the use of other plugins for time delay etc.

Heres one I made earlier

[whgeiger]

Driver bandwidth/distortion and efficiency are inversely related. More of the former will be at the expense of less of the latter. “Well behaved” means trading driver higher efficiency and lower distortion for shallow stop-band slopes in the c/o. The purpose of the crossover is to keep signal out of the driver that it cannot handle. I.E., those low frequencies that cause diaphragm excursions that exceed low distortion displacement limits, or those that go to heat in the voice coil because the driver is not responding to higher signal frequencies. The fact that signal filtering required to meet this mission introduces signal delays beyond those introduced by the drivers themselves, is of little merit, when steeper stop-band slopes will permit use of drivers optimized to be operated efficiently and at lower distortion over a narrower bandwidths. When DSP is used, particularly in a DIY setting, optimization of system performance is readily obtainable without material compromise in signal phasing or driver performance, not to mention the experimentation it facilitates during the design tweaking process.

Regards,

WHG