ALL,
So I'm learning that IIR filters cause an electrical phase shift equal to their analog counterpart and you can compensate for that using delay...right? If that be the case, then some questions based on this system: sub woofer= LR24dB Low pass @ 90Hz___woofer=LR24dB HP@90Hz & LR24dB LP@650Hz___planar mid/tweet=LR24dB HP@650Hz
1) If I have a LR 24dB Lo Pass 90Hz on my sub and a LR 24dB Hi Pass 90Hz to my woofer, that would be a 180deg shift that would delay the signal @ 90Hz by 5.5ms...right? So I'd add 5.5ms delay to woofer, yea?
2) A 180deg phase shift would delay the woofer signal by .75ms at 650Hz...if I've delayed the woofer's signal by 5.5ms to "meet" the sub, is that cumulative to the tweeter's overall delay for a total of 6.25ms?
3) On top of all this, is the delay for time alignment added in as well?
For reference, my sub is a down firing 18 in an aperiodic enclosure. The OB 15" woofer Voice Coil is about 5" forward of the center of the 18. The OB planar mid/tweets are about 3" in front of the 15 VC.
Thanks for the lessons to come!
So I'm learning that IIR filters cause an electrical phase shift equal to their analog counterpart and you can compensate for that using delay...right? If that be the case, then some questions based on this system: sub woofer= LR24dB Low pass @ 90Hz___woofer=LR24dB HP@90Hz & LR24dB LP@650Hz___planar mid/tweet=LR24dB HP@650Hz
1) If I have a LR 24dB Lo Pass 90Hz on my sub and a LR 24dB Hi Pass 90Hz to my woofer, that would be a 180deg shift that would delay the signal @ 90Hz by 5.5ms...right? So I'd add 5.5ms delay to woofer, yea?
2) A 180deg phase shift would delay the woofer signal by .75ms at 650Hz...if I've delayed the woofer's signal by 5.5ms to "meet" the sub, is that cumulative to the tweeter's overall delay for a total of 6.25ms?
3) On top of all this, is the delay for time alignment added in as well?
For reference, my sub is a down firing 18 in an aperiodic enclosure. The OB 15" woofer Voice Coil is about 5" forward of the center of the 18. The OB planar mid/tweets are about 3" in front of the 15 VC.
Thanks for the lessons to come!
Attachments
Member
Joined 2003
Your head is in the right place, but all these assumptions you are making are just that, assumptions. Thinking of phase shift only electrically and alignment based on a guess of voice coil location will have limited reliability. Required delay will vary by driver, location and acoustic frequency response, so should be determined by acoustic measurement. For miniDSP there is much application data on their website, I would start here:
https://www.minidsp.com/applications/rew/measuring-time-delay
To help with your questions on delay and crossover order, etc. it can be simply shown using VituixCAD.
1. LR24 includes 180 degree phase shift at the knee. High pass and low pass at LR24 at the same frequency are perfectly in-phase at the crossover frequency. Not taking into account acoustic delay between drivers, the pair of LR24 filters will sum flat.
2 and 3. Generally one would set up delay values relative to the tweeter, so tweeter include 0ms delay, any delay of other drivers would be entered relative to the tweeter. Required delay will vary by driver, location and acoustic frequency response, so should be determined by acoustic measurement. It is not as simple as taking a ruler and measuring from one driver VC to the other.
https://www.minidsp.com/applications/rew/measuring-time-delay
To help with your questions on delay and crossover order, etc. it can be simply shown using VituixCAD.
1. LR24 includes 180 degree phase shift at the knee. High pass and low pass at LR24 at the same frequency are perfectly in-phase at the crossover frequency. Not taking into account acoustic delay between drivers, the pair of LR24 filters will sum flat.
2 and 3. Generally one would set up delay values relative to the tweeter, so tweeter include 0ms delay, any delay of other drivers would be entered relative to the tweeter. Required delay will vary by driver, location and acoustic frequency response, so should be determined by acoustic measurement. It is not as simple as taking a ruler and measuring from one driver VC to the other.
Dcibel, thanks for the response on this. I will definitely be doing some measurements to verify everything, I just wanted to see if my thought process/reasoning was solid. I think I understand; even though there is an electrical phase shift at crossover, it's not really inducing a delay of a full cycle...JUST an electrical phase shift, correct? Thanks again.
Member
Joined 2003
Phase is derived from the frequency response. Electrical frequency response is directly transferred to the acoustic output, which is often referred to as the "transfer function". That is to say that a phase change in the electrical function is also a phase change in the acoustic output.
LR24 high pass will be 180 degree shift at the knee, but LR24 low pass will also be 180 degree shift at the knee. So even though there is 180 degrees of phase shift, the two filters remain in phase and sum flat, which I've shown above. Given this perfect theoretical example, no delay in either driver is required. This of course doesn't take into account driver location and actual driver frequencyt response, which is why real acoustic measurement is required for any accuracy, and you'll find that these basic "textbook" filter transfer functions are rarely the correct solution.
I mentioned previously, all the DSP functions and resulting response can be easily evaluated in VituixCAD using the active filter blocks, just set miniDSP as the DSP system in the options. The blocks can be directly transferred to your DSP with accuracy when doing this.
LR24 high pass will be 180 degree shift at the knee, but LR24 low pass will also be 180 degree shift at the knee. So even though there is 180 degrees of phase shift, the two filters remain in phase and sum flat, which I've shown above. Given this perfect theoretical example, no delay in either driver is required. This of course doesn't take into account driver location and actual driver frequencyt response, which is why real acoustic measurement is required for any accuracy, and you'll find that these basic "textbook" filter transfer functions are rarely the correct solution.
I mentioned previously, all the DSP functions and resulting response can be easily evaluated in VituixCAD using the active filter blocks, just set miniDSP as the DSP system in the options. The blocks can be directly transferred to your DSP with accuracy when doing this.
Sum flat but 360° phase difference between the channels. Audible or not audible, that is the question.
That’s exactly the part I’m trying to figure out…is a full cycle behind audible and is even worth trying to align the drivers on the same cycle and phase by delaying by a cycle…plus physical offset. Like I said, I will hopefully get some free time to test these out, this is all just me trying think it thru so I understand the measurements when taken. Thanks guys!
Yeah if the GD of the low pass is 5.5ms
And you add 5.5ms to the mains (not sub) so the mains are in time with the sub (before the knee) it would sound linear phase , except the crossover wouldn’t have phase tracking and would be way out of time
The goal is to delay the sub at the same rate the mains are starting to delay because of the crossover delay
that’s the phase tracking properties of the crossover, thought out by the mathematicians that came up with it
the summed result is a all pass….
And you add 5.5ms to the mains (not sub) so the mains are in time with the sub (before the knee) it would sound linear phase , except the crossover wouldn’t have phase tracking and would be way out of time
The goal is to delay the sub at the same rate the mains are starting to delay because of the crossover delay
that’s the phase tracking properties of the crossover, thought out by the mathematicians that came up with it
the summed result is a all pass….