I am re-working a xover I built some time back that is for a 3-way design.
My design is essentially this:
https://sound-au.com/project81.htm
While my design is a 3-way, it is literally just a single adaptation of the above. All I have done is taken the signal that comes off the high-pass filter block and fed it back into another 2-way network.
-> The low pass from the initial network goes directly to the low frequency amplifier.
-> The high pass from the initial netowork goes into the SECOND 2-way network
-> The midrange is fed through the SECOND low pass out
-> The tweeter is fed through the SECOND high pass out
All this is done w/ active circuitry - this is NOT a passive crossover.
So now I am wondering how the phase shift is going to play out...
The two outputs from the initial filter are each 90 degrees out-of-phase from the input - so 180 total.
The outputs from the second filter are also each 90 degrees out-of-phase from the output of the first filter. So correct me if I am wrong but...
1) Tweeter and midrange will be 180 out of phase from each-other
2) Midrange and woofer will be 90 out of phase from each other, no?
In the second filter, the midrange is delayed another 90 degrees - bringing it back down to where the original signal should be - and since the woofer was delayed 90 degrees from the original signal to start, we have a total of 90 degrees difference between woofer and mid - correct?
So there is the first question - am I understanding the phase shift correctly?
Question 2: Is this 90 degree going to be much of an issue? I suppose that the driver VC offsets can affect things much more than this 90 degree offset imposed at the xover.
Question 3: I suppose I could use any arbitrary 1st-order filters back-to-back to provide the 90-degree phase shift needed to bring the woofer back into relative phase with the tween and woofer network, no? Any RC or CR network set that is well above the woofer xover frequency would basically each provide 45 degrees - giving me 90 total.
Now for the one I have been wanting to ask for YEARS....
Question 4: I am not a math junkie and never got anything beyond my high school education in math. So forgive this ignorance but..
I understand how you can delay a signal by 45/90/etc degrees. That makes perfect sense. Energy gets stored in the network and released a tiny fraction of time later - delaying the signal that is output.
But how on earth does the high-pass signal get "moved forward"?!?!?! Forgive me here - but even with advances in quantum mechanics, we have not fully developed time travel in actual practice, so... How can a signal be shifted 90 degrees FORWARD? I am - ovbiously - thinking about this in the time domain. So to shift a signal FORWARD, you have to move the entire time scale BACKWARDS. What am I missing here?
My design is essentially this:
https://sound-au.com/project81.htm
While my design is a 3-way, it is literally just a single adaptation of the above. All I have done is taken the signal that comes off the high-pass filter block and fed it back into another 2-way network.
-> The low pass from the initial network goes directly to the low frequency amplifier.
-> The high pass from the initial netowork goes into the SECOND 2-way network
-> The midrange is fed through the SECOND low pass out
-> The tweeter is fed through the SECOND high pass out
All this is done w/ active circuitry - this is NOT a passive crossover.
So now I am wondering how the phase shift is going to play out...
The two outputs from the initial filter are each 90 degrees out-of-phase from the input - so 180 total.
The outputs from the second filter are also each 90 degrees out-of-phase from the output of the first filter. So correct me if I am wrong but...
1) Tweeter and midrange will be 180 out of phase from each-other
2) Midrange and woofer will be 90 out of phase from each other, no?
In the second filter, the midrange is delayed another 90 degrees - bringing it back down to where the original signal should be - and since the woofer was delayed 90 degrees from the original signal to start, we have a total of 90 degrees difference between woofer and mid - correct?
So there is the first question - am I understanding the phase shift correctly?
Question 2: Is this 90 degree going to be much of an issue? I suppose that the driver VC offsets can affect things much more than this 90 degree offset imposed at the xover.
Question 3: I suppose I could use any arbitrary 1st-order filters back-to-back to provide the 90-degree phase shift needed to bring the woofer back into relative phase with the tween and woofer network, no? Any RC or CR network set that is well above the woofer xover frequency would basically each provide 45 degrees - giving me 90 total.
Now for the one I have been wanting to ask for YEARS....
Question 4: I am not a math junkie and never got anything beyond my high school education in math. So forgive this ignorance but..
I understand how you can delay a signal by 45/90/etc degrees. That makes perfect sense. Energy gets stored in the network and released a tiny fraction of time later - delaying the signal that is output.
But how on earth does the high-pass signal get "moved forward"?!?!?! Forgive me here - but even with advances in quantum mechanics, we have not fully developed time travel in actual practice, so... How can a signal be shifted 90 degrees FORWARD? I am - ovbiously - thinking about this in the time domain. So to shift a signal FORWARD, you have to move the entire time scale BACKWARDS. What am I missing here?
Phase shift doesn't imply time travel... 90 degree advance is the same as 270 degree retardation, no implication of a time shift at all. You look at the step response or impulse response when thinking time domain, not phase shift at a single frequency. The plot of phase delay against frequency is also useful here.
In principle you need to include an all-pass section in a three-way Linkwitz-Riley filter, see https://www.diyaudio.com/community/...-crossover-please-comment.394499/post-7235230 For 12 dB/octave, it should be a simple first-order all-pass.
So my curiosity got the best of me last night and I decided to scope it out.
Hooked channel 1 up to the mid-out, and channel 2 to the low out. I then fed a signal at the low-to-mid xover frequeny and... the two were exactly 180 out of phase.
Same when feeding a signal at the mid-to-high freq and swichting channel 2 of the scope over to the high out... Also exactly 180 out of phase from the mid.
So reverse the midrange after the amplifier. Simple enough.
I imagine the additional 90 degrees of phase shift between the extra high/low section -v- the low output of the previous network is just not enough to notice? I put the two signals from each output exactly over each other on the scope to maximize my chance of seeing the difference between the two signals and... they look exactly 180 from each other.
I got some special cables set up and am now playing around with plotting the responses w/ REW (line out -> xover -> line in)
Hooked channel 1 up to the mid-out, and channel 2 to the low out. I then fed a signal at the low-to-mid xover frequeny and... the two were exactly 180 out of phase.
Same when feeding a signal at the mid-to-high freq and swichting channel 2 of the scope over to the high out... Also exactly 180 out of phase from the mid.
So reverse the midrange after the amplifier. Simple enough.
I imagine the additional 90 degrees of phase shift between the extra high/low section -v- the low output of the previous network is just not enough to notice? I put the two signals from each output exactly over each other on the scope to maximize my chance of seeing the difference between the two signals and... they look exactly 180 from each other.
I got some special cables set up and am now playing around with plotting the responses w/ REW (line out -> xover -> line in)
Gentlemen...
So tonight I spent some time and decided to scope it out a bit more.
Before I begin, let me just say... I am not a professional and am still learning my way around the scope.
Here is what I have done:
1) Connected the xover circuit and fed it a 190hz signal (the xover point of the low-to-mid/high section).
2) Connected one channel of the scope to the low-pass out
3) Connected the other channel to the mid band-pass out
4) Inverted the low-pass channel from the scope
Here is what it looks like:
So yeah - there IS a bit of a lag on the low-pass -vs- the mid band-pass.
But that does not look 90 degrees out-of-phase to me.
My initial thoughts here are: that does not look THAT bad and there is likely going to be MORE phase-alignment errors in the speaker VC-spacing than what I see here, no?
The high-pass section is around 3900 - is this the ratio which is mentioned previously? 190-to-3900?
Seems like I may not need to worry about building an all-pass, no?
-Dean
So tonight I spent some time and decided to scope it out a bit more.
Before I begin, let me just say... I am not a professional and am still learning my way around the scope.
Here is what I have done:
1) Connected the xover circuit and fed it a 190hz signal (the xover point of the low-to-mid/high section).
2) Connected one channel of the scope to the low-pass out
3) Connected the other channel to the mid band-pass out
4) Inverted the low-pass channel from the scope
Here is what it looks like:
So yeah - there IS a bit of a lag on the low-pass -vs- the mid band-pass.
But that does not look 90 degrees out-of-phase to me.
My initial thoughts here are: that does not look THAT bad and there is likely going to be MORE phase-alignment errors in the speaker VC-spacing than what I see here, no?
The high-pass section is around 3900 - is this the ratio which is mentioned previously? 190-to-3900?
Seems like I may not need to worry about building an all-pass, no?
-Dean