Housekeeping note.
It seems like the information contained in this thread would apply to pro sound coaxial drivers. One could measure the HF driver and woofer offsets using an impulse test (difference in time of flight). Next choose the quasi-optimal crossover that places the crossover frequency at an appropriate frequency for the drivers and their physical offset. I wonder if this is similar to the type of crossover Danley uses in his coaxial Synergy speakers. One of the projects I've been wanting to attempt is a baby version of the Danley SM100 using the Faital 6HX150.
Reverse formula? Try to simulate with Jmlc excel spreadsheet. : http://www.acommeaudio.fr/files/filtre_alignement_synthese.xls
You must delay highs with physical offset if using a passive crossover or digital delay with active dsp one ( dcx / bss ect. ) in order to lower and linéarise global group delay.
crd
You must delay highs with physical offset if using a passive crossover or digital delay with active dsp one ( dcx / bss ect. ) in order to lower and linéarise global group delay.
crd
Last edited:
I will play around with the spreadsheet and see...but what I have done with it hasn't really clicked yet.
I was under the impression that these alignments were based on the HF horn having an offset behind that of the woofer. And was wondering if the woofer was behind the HF horn, if you would reverse the filter types and multipliers that was listed a few pages back?
I was under the impression that these alignments were based on the HF horn having an offset behind that of the woofer. And was wondering if the woofer was behind the HF horn, if you would reverse the filter types and multipliers that was listed a few pages back?
I have read through this post a few time and need help understanding the driver offset. Is it correct to say;? The LF driver is delayed by its crossover so the HF driver needs to be behind it physically by the distance given by O.22c/Fc for the 3rd order butterworth jmlc for proper time alignment.
Thanks
Cort
Thanks
Cort
Crossover creates group delay increasing at low fréquences.
The tweeter offset required for a "quasi optimal" operates the "form of équivalence" between true delay and group delay to balance the over all group delay.
The tweeter offset required for a "quasi optimal" operates the "form of équivalence" between true delay and group delay to balance the over all group delay.
An externally hosted image should be here but it was not working when we last tested it.
Most excellent, Thank you for showing the picture in my brain, can you describe what you hear when you get the offset right? I am sure many have done the experiment. what are the charecteristics of the sound when the offset is to small or large. The extremes are obvious, but what is the window? 0.1 ms?
Hi!Ha, what I have in mind won't fit into a car.
Any news about 8(6)HX150?..
(Ordered sale pair 8HX150 5 minutes ago
)
When the offset is about right; IF individial driver's response show BOTH acoustic slopes AND phase targets respected, then you'll end up with a pretty flat summed response.
Which is not the only thing we are able to feel by subjectively listening to the result, but is a part of something balanced sonically for sure.
In real world the phase responses relative to each other will never give you perfection, because of imperfection of the drivesr even if your designed crossover is techincally excellent by itself.
So, assuming you have acoustics targets very near to the goal (slopes) you defined, and acoustic real world phases derivating a little from the goal, then varying (teaking) the acoustic offset to get the most flat response will probably be the most sonically satisfying in the end.
You'll find the inverting point of the non-unfolded phase response to happen between drivers, if it's technically close to the original thoeritical design in the excel spreadsheet, then I would say you're a lucky guy
Best,
nAr
Revival?
I found this old thread when I worked out more-or-less the same result as Marco, decided to try a "6th order Bessel" Low Pass and searched on that.
Some of my ideas have already been discussed but I hope the participants are still interested because there's a few details I still don't have clear.
How to determine the required filter order if I have a specific offset.
For instance, to use round numbers and worst case, if I had an offset of 50 cm from my woofer to the mid driver and a 666 Hz crossover, that's a 1.0*c/Fc
What order will be adequate to produce this offset?
The total Group delay does not seem to increase linearly with the filter order and I don't know any universal formula to calculate or estimate it.
There is a simple formula for phase shift per order, can this be useful?
I have considered cascaded Bessel filters, similar to the way L-R uses cascaded Butterworth, and it looks effective.
A Bessel filter plus an all-pass for extra time delay also looks practicable. Comments?
Heyser points out that Group Delay is not necessarily time delay.
Does this make a difference in this situation?
Hopefully
David
I found this old thread when I worked out more-or-less the same result as Marco, decided to try a "6th order Bessel" Low Pass and searched on that.
Some of my ideas have already been discussed but I hope the participants are still interested because there's a few details I still don't have clear.
How to determine the required filter order if I have a specific offset.
For instance, to use round numbers and worst case, if I had an offset of 50 cm from my woofer to the mid driver and a 666 Hz crossover, that's a 1.0*c/Fc
What order will be adequate to produce this offset?
The total Group delay does not seem to increase linearly with the filter order and I don't know any universal formula to calculate or estimate it.
There is a simple formula for phase shift per order, can this be useful?
I have considered cascaded Bessel filters, similar to the way L-R uses cascaded Butterworth, and it looks effective.
A Bessel filter plus an all-pass for extra time delay also looks practicable. Comments?
Heyser points out that Group Delay is not necessarily time delay.
Does this make a difference in this situation?
Hopefully
David
Last edited:
I am trying to help you, Dave.
Each lo- or hipass of first order results in a group delay at DC equalling its time constant, which is 1/(capacitance*resistance) or resistance/inductance. For example, a first-order lopass with -3dB corner f=1000Hz causes t=0.16ms at DC. A lo- or hipass of second order with the same character frequency, in explanation, where phase is half way thru, and which is 1/(2*pi*(LC)^(1/2)), causes twice that t, for example a second-order lo- or hipass with f=1000Hz causes t=0.32ms.
For a non-ringing lo- or hipass, in explanation with Q<=(1/2), group delay at f is half of its DC value. I guess, that for a second-order lopass, not only relative amplitude but also relative group delay at f is Q times the value at DC.
Each lo- or hipass of first order results in a group delay at DC equalling its time constant, which is 1/(capacitance*resistance) or resistance/inductance. For example, a first-order lopass with -3dB corner f=1000Hz causes t=0.16ms at DC. A lo- or hipass of second order with the same character frequency, in explanation, where phase is half way thru, and which is 1/(2*pi*(LC)^(1/2)), causes twice that t, for example a second-order lo- or hipass with f=1000Hz causes t=0.32ms.
For a non-ringing lo- or hipass, in explanation with Q<=(1/2), group delay at f is half of its DC value. I guess, that for a second-order lopass, not only relative amplitude but also relative group delay at f is Q times the value at DC.
Wouldn't it we want to get low pass delay to match the physical offset between the drivers ? If so, only the low pass crossover exhibits time delay, and the high pass doesn’t ??? ... Or at least, it was my understanding of the way Q_O would work, as of today
Best,
nAr
I meant, only the low pass crossover exhibits significant time delay in order to make proper time alignment.
The high pass has its own delay, but seems negligible regards to the physical offset alignment.
Best,
nAr
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.