Can some help me here, in layman's terms!
I currently run my system controlled with a digital dsp enabling me to set delays for minimum phase/time alignment easily, with the help of arta.
Thought I would dabble with analogue again and purchased a Rane 23s. The manual has a chart for "rough" delay settings by crossover frequency and voice coil offset.
My question...why is the amount of delay required different at different frequencies for the same driver offset, which in my case is around 9" as I use a horn and cone mid set up..
Thanks.....
I currently run my system controlled with a digital dsp enabling me to set delays for minimum phase/time alignment easily, with the help of arta.
Thought I would dabble with analogue again and purchased a Rane 23s. The manual has a chart for "rough" delay settings by crossover frequency and voice coil offset.
My question...why is the amount of delay required different at different frequencies for the same driver offset, which in my case is around 9" as I use a horn and cone mid set up..
Thanks.....
It's not as it appears, the effective scale is suggesting otherwise. The setting is actually the same for a given distance over the lower range of frequencies.
The variations at higher frequencies may come down to a device calibration requirement.. for some reason the result varies at higher frequencies but since this is known, you can still reach your goal.
The variations at higher frequencies may come down to a device calibration requirement.. for some reason the result varies at higher frequencies but since this is known, you can still reach your goal.
Not sure I understand your "wavelength" answer....With dsp you can measure the physical distance and just set it because the dsp delays all frequencies by the same amount. I get that, but with a 4th order LR, HP and LP analogue active set to the same frequencies why would you need to set different delays for different crossover frequencies when the voice coil offset hasn't changed ? Speed of sound is a constant? Nope still dont get it but I'm sure gd and phase angle are lurking in the answer somewhere!
the longer the wavelength is the less chance the phase is different over distance, as the frequency increases the wavelengths are smaller and the number of times the phase is likely to be out 180 deg due to the offset increases. i wish i had the proper graphics program to display the same wavelength originating from two different start points to show the relationship.
Last edited:
Sorry, still don't get it completely! Why is that different with digital delay then where you just dial in the physical difference (in mm or ms or microsec) independent of frequency? Which is the focus of my question..why do you take account of crossover frequency with analogue but not digital and why are the "delay" times different.
My impression is the 0-2ms range is a bit of a misnomer and it really means phase shift at crossover, which I guess is what you are saying regarding taking into account the "wavelength"?
Still leaves me a bit confused!!
My impression is the 0-2ms range is a bit of a misnomer and it really means phase shift at crossover, which I guess is what you are saying regarding taking into account the "wavelength"?
Still leaves me a bit confused!!
not easy to grasp and we have not even broached the multitude of partial cancellations that occur at all phase angles between 0 and 180 deg and at all frequencies....(which explains my love of scotch!)
as to a difference between digital and analogue i'm not certain what you mean? whether the device your using is digital(dsp) or analog (still techically a "digital" delay) the only element i see as different is latency of the DSP process.
as to a difference between digital and analogue i'm not certain what you mean? whether the device your using is digital(dsp) or analog (still techically a "digital" delay) the only element i see as different is latency of the DSP process.
I'd join you in one now, only I have to drive in the morning 🙁
It is fairly common for analogue adjustable methods to be more sensitive at one end of the scale, or more accurate in some cases.. especially where a large range of variation is required.
I wouldn't necessarily call it a bad thing either. It is still possible to get an accurate result regardless of what the dial says, and besides, audio types traditionally don't generally have a problem with this level of circuitry.
It is fairly common for analogue adjustable methods to be more sensitive at one end of the scale, or more accurate in some cases.. especially where a large range of variation is required.
I wouldn't necessarily call it a bad thing either. It is still possible to get an accurate result regardless of what the dial says, and besides, audio types traditionally don't generally have a problem with this level of circuitry.
Latency is the same at all outlets. The total delay is determined by the unit's processing.
Ok, this is an actual example. The horn coil is around 9 to 10 inches behind the mid. Arta optimises the response to be flat with a "delay" setting around 229mm...or around .8ish ms. Which makes total sense as it matches the offset of the coils perfectly.
Rane publish a rough guide to the "delay" setting on the 23s. At the same frequency I have chosen to crossover (1k2) Rane suggests it's over the max setting of 2ms on their AC23s. But at lower frequencies, say 800Hz it says to set it to a lower 1.5ms ...
So I still don't get it. Speed of sound is a constant. The distance between the voice coils hasn't changed and nor has the measuring distance ...how can the time it takes for the signal from each source not be the same, whether it's 100hz, 1kh or 10khz? It still travels at 345m/s?
And that's enough for a digital crossover to know...the coils are separated by a fixed amount, the difference in time arrival is determined by how long it takes to reach the measuring mic, and as expected the delay you set is exactly equivalent to the difference in driver offset.
So, no, sorry, I'm still not with this...!! Allenb has given an explanation but not sure I really get that either.
I have contacted Rane but it takes forever to get a reply and usually they answer my query with another question!! Although the AC23a is still available from many sources, Rane don't even list it??
I need the actual physics on this to get a grip on it!
Ok, this is an actual example. The horn coil is around 9 to 10 inches behind the mid. Arta optimises the response to be flat with a "delay" setting around 229mm...or around .8ish ms. Which makes total sense as it matches the offset of the coils perfectly.
Rane publish a rough guide to the "delay" setting on the 23s. At the same frequency I have chosen to crossover (1k2) Rane suggests it's over the max setting of 2ms on their AC23s. But at lower frequencies, say 800Hz it says to set it to a lower 1.5ms ...
So I still don't get it. Speed of sound is a constant. The distance between the voice coils hasn't changed and nor has the measuring distance ...how can the time it takes for the signal from each source not be the same, whether it's 100hz, 1kh or 10khz? It still travels at 345m/s?
And that's enough for a digital crossover to know...the coils are separated by a fixed amount, the difference in time arrival is determined by how long it takes to reach the measuring mic, and as expected the delay you set is exactly equivalent to the difference in driver offset.
So, no, sorry, I'm still not with this...!! Allenb has given an explanation but not sure I really get that either.
I have contacted Rane but it takes forever to get a reply and usually they answer my query with another question!! Although the AC23a is still available from many sources, Rane don't even list it??
I need the actual physics on this to get a grip on it!
too late for me...i'm already a few shots in...
something not mentioned but also important in this scenario is steepness of filter used, as it affects the overlap between the bands we are hoping to time align to make our x-over point as pristine as possible...time for another!
missed that post, and this may not help SoundRight yes the speed of all frequencies is constant but their phase changes at a different rate...and no it's not a perfect relationship but we're only interested in the phenomenon over a limited bandwidth.
something not mentioned but also important in this scenario is steepness of filter used, as it affects the overlap between the bands we are hoping to time align to make our x-over point as pristine as possible...time for another!
missed that post, and this may not help SoundRight yes the speed of all frequencies is constant but their phase changes at a different rate...and no it's not a perfect relationship but we're only interested in the phenomenon over a limited bandwidth.
Last edited:
A delay of 0.75" asks for a setting of 1 at any frequency, as long as it's between 70Hz and 6kHz. A difference of 6" asks for a setting of 2.5 as long as it's between 70Hz and 500Hz.
We are talking about a tenfold difference in frequency needing the same setting. Clearly the delay is not dependent on frequency 'ordinarily'. What we are talking about here is a specific anomaly. It is documented and I'm not certain what contacting Rane would achieve.
How deep do you want to go.. Would you consider the effect of using a linear potentiometer as a volume control as sufficient proof of concept? Do you have a schematic?
We are talking about a tenfold difference in frequency needing the same setting. Clearly the delay is not dependent on frequency 'ordinarily'. What we are talking about here is a specific anomaly. It is documented and I'm not certain what contacting Rane would achieve.
How deep do you want to go.. Would you consider the effect of using a linear potentiometer as a volume control as sufficient proof of concept? Do you have a schematic?
Great, do we know where it is documented? Would love to read about it.
Only lunchtime here....so not started yet!
As for crossover type, 4th order LR as posted for both devices. Obviously different filters will produce different delay characteristics at crossover, but LR is constant for both sections I understand?
Only lunchtime here....so not started yet!
As for crossover type, 4th order LR as posted for both devices. Obviously different filters will produce different delay characteristics at crossover, but LR is constant for both sections I understand?
Can it be in digital the sound is put on hold and analog the sound can not be put on hold and instead go for wave alignment?
Yes, I understand that article, thanks. But it doesn't answer the question as to why, say, "2ms" on the dial is "2ms" at 100hz but not 10khz?
Makes no sense!! Be interesting to see what the "delay" is actually doing because without a "bucket brigade" circuit for actual real time delay my best guess is it is phase shifting? 4th Order LR networks have equal gd in both filter sections if my memory is correct so I imagine they use relative phase shift to bring offset drivers back in phase, but therefore not time aligned of course!
Makes no sense!! Be interesting to see what the "delay" is actually doing because without a "bucket brigade" circuit for actual real time delay my best guess is it is phase shifting? 4th Order LR networks have equal gd in both filter sections if my memory is correct so I imagine they use relative phase shift to bring offset drivers back in phase, but therefore not time aligned of course!
it's more like bringing two sources that are separated by distance or time (they are almost interchangeable in this instance) back to coherence in time i.e. additive and not substractive this is why i mentioned earlier that it's a complex relationship between frequency and distance over time and in real time the clock on all three is running (not to mention that phase of each frequency is varying not merely between 0 and 180 it's 360 so it's far more complex) to wit it's not merely a phase shift it's a rotation and group delay can make this an even bigger mess.
but in fact this is the essence of time alignment all sources capable of producing the same frequencies are "time aligned", no time alignment no phase coherence = potential cancellation (and to make it worse there's the possibility of "additive" lumps, areas where the response is exaggerated )
to hopefully answer your question 2ms is the same at 100hz as it is at 10 khz but it's affect on each (relative phase over time) is different because of wavelength.
but in fact this is the essence of time alignment all sources capable of producing the same frequencies are "time aligned", no time alignment no phase coherence = potential cancellation (and to make it worse there's the possibility of "additive" lumps, areas where the response is exaggerated )
to hopefully answer your question 2ms is the same at 100hz as it is at 10 khz but it's affect on each (relative phase over time) is different because of wavelength.
Last edited: