ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 [2]  All   Go Down

Author Topic: More thoughts (and pictures) on crossovers, phase, and delay  (Read 1102 times)

Luke Geis

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1518
    • Owner of Endever Music Production's
Re: More thoughts (and pictures) on crossovers, phase, and delay
« Reply #10 on: October 31, 2018, 03:36:33 pm »

It has been my understanding that the crossover corner frequency is where the bulk of the phase issue arise. We oversimplify the filters effects by simply saying that a 2 pole or 4 pole filter has 180* and 360* of phase shift respectively. We tend to forget phase lead and phase lag around the corner frequency and that the phase of the upper passband starts at +45* per pole and settles to 0* of phase while the lower passband starts at 0* of phase shift and shifts to its final relative phase. This means that there is a fair number of frequencies in which the phase of two different drivers are never going to properly align no matter what we do unless we make a frequency dependent phase inverting filter.

In the case of a 24db per octave filter ( a 4th order filter ), we have 360* of phase shift occurring between the hi and lo band. The upper band would start at 180* and settle to 0* while the low band would start at 0* and shift 180* out of phase where it would eventually settle at. The 360* of phase shift only actually exists around the crossover point, but as we get lower and higher in frequency relative to the crossover point the phase shift is actually only 180*.

This is where all-pass filters help. A first-order allpass filter provides 180* of phase shift regardless of frequency. This means that when used on the upper passband that we could have the lows and hi's in proper phase outside of the crossover region ( both now lagging by 180* ), but it doesn't address the phase shift within the crossover region. Keep in mind that the upper passband would now start at 360* and then settle to 180*. Within the crossover region, there will always be some amount of phase mismatch, but at least the relative phase beyond the crossover region would be in proper phase. This doesn't really address the issues of frequency choice ( for the crossovers ) and the physical placement and resultant time-based issues. 

We simplify the group delay utilizing the IR and or FFT information and then simply align the phase slopes till they best match. Honestly, this is probably the best way. If we were to widen the window and average less for a more realistic outcome, we would probably find that we can never truly align the subs and the mains at all frequencies that pass between them.

Lead-Lag Compensators do exist but are not implemented in audio as of yet. It may be the next step in the correction of crossover networks. I think with this form of technology we can get better results. The problem isn't so much the resultant phase difference as much as the altering phase directions within a crossover network.
Logged
I don't understand how you can't hear your self

Mark Wilkinson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 766
Re: More thoughts (and pictures) on crossovers, phase, and delay
« Reply #11 on: October 31, 2018, 03:51:54 pm »

Luke, what about complementary crossovers...e.g. linkwitz-riley...that share the same phase trace throughout summation region?

 
Logged

Luke Geis

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1518
    • Owner of Endever Music Production's
Re: More thoughts (and pictures) on crossovers, phase, and delay
« Reply #12 on: October 31, 2018, 06:45:25 pm »

LR crossovers do solve the phase issue through the crossover region, yes, but they still exhibit the total phase shift from the two passbands.

An LR filter is simply a cascaded BW filter. An LR filter is a 2x summation of two BW filters. So two 6db BW filters cascaded will add up to a 12db LR filter. It will have an equal phase shift at all frequencies through the crossover region, but it will still exhibit the total phase shift of the pole count. In the case of a 12db LR filter that would be a 180* phase shift. You still have to play god in other words.

The biggest issue is still the physical difference in time and space of the drivers in interest. Electrically we can add them up and alter things to be more in line, but the acoustic outcome is what matters. The big question is how do we measure it and what is the best way to make it align? As we know if we move any number of feet up or down, left or right, we will change the time-space point of incident and thus the phase and acoustic summation result. What yields the best outcome?
Logged
I don't understand how you can't hear your self

Mark Wilkinson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 766
Re: More thoughts (and pictures) on crossovers, phase, and delay
« Reply #13 on: October 31, 2018, 07:15:29 pm »

LR crossovers do solve the phase issue through the crossover region, yes, but they still exhibit the total phase shift from the two passbands.

An LR filter is simply a cascaded BW filter. An LR filter is a 2x summation of two BW filters. So two 6db BW filters cascaded will add up to a 12db LR filter. It will have an equal phase shift at all frequencies through the crossover region, but it will still exhibit the total phase shift of the pole count. In the case of a 12db LR filter that would be a 180* phase shift. You still have to play god in other words.

The biggest issue is still the physical difference in time and space of the drivers in interest. Electrically we can add them up and alter things to be more in line, but the acoustic outcome is what matters. The big question is how do we measure it and what is the best way to make it align? As we know if we move any number of feet up or down, left or right, we will change the time-space point of incident and thus the phase and acoustic summation result. What yields the best outcome?

Yep on the butterworths...I've tried to attach a chart on how the orders act, for any following along...(if upload is working yet)

And sure, acoustic summation is all that finally counts...and measured to what position?  But those are issues I think beyond what's being discussed...

So far, we've kinda just been focusing on the relationship between phase and impulse response....

Nope..uploader still full when tried to post...
Here's the chart without good formatting..still should make sense i hope

      Bu crossovers   
         
              deg    dB gain
order pol     @ xover   @ xover
1   equal   90   0
1   opposed   270   0
2   equal   180   -total
2   opposed   0   3
3   equal   270   0
3   opposed   90   0
4   equal   0   3
4   opposed   180   -total
5   equal   270   0
5   opposed   90   0
6   equal   90   -total
6   opposed   180   3
7   equal   90   0
7   opposed   270   0
8   equal   0   3
8   opposed   180   -total

Logged
Pages: 1 [2]  All   Go Up
 


Page created in 0.04 seconds with 18 queries.