O-scope, crossovers, and delay

I just used the o-scope function in REW to diagnose why the hell I could never get my subs and mids to blend properly. I have swapped the RCA's in this whole process and the output from the head unit (went from network mode to standard mode to eliminate HU xovers, two full range outputs now). I found that the subs are somehow delayed by 24 milliseconds. So, since I'm using a MiniDSP C-DSP 8x12, I only have the capability to add 25 milliseconds maximum delay, not enough to compensate for the additional delay plus time alignment. I looped one output back to an input (for the front stage) and added the additional delay to that loop, problem solved. Unexpectedly, there are no ill effects from the loop; and the subs and mids blend perfectly.

Experts please comment on this.>> I'm talking specifically to Justin Zazzi, ErinH, and any other electrical/acoustic engineering types.>> Before I move on, all xover slopes mentioned are acoustic slopes NOT electronic. I can only guess that this additional delay is from the LR24 slopes on either side of the 90hz xover point, one full wavelength delay from each slope adding up to 22.22ms delay, that seems about right. Should we calculate additional delay like this? >"Xover point wavelength in ms" X "number of cycles (360 deg) phase change in slopes (LP+HP)" = "additional delay to add to the high pass side". In my case, 90hz=11.11ms LR24=360deg xover -- (11.11ms)X(1+1)=22.22ms. This math may be different (it likely is different) for Butterworth xovers, but it works for Linkwitz-Riley at least.<< I feel like I rediscovered some math we've all been ignoring for the sake of simplicity. I'm sure this is covered somewhere, point me to it if you can. Or maybe I'm just measuring a single point of group delay, in which case this is fairly pointless.

Has anyone else used an oscillosope for time alignment?

The process is fairly simple:
1) Fire up your laptop
2) Plug in your trusty measurement mic
3) Fire up REW and your DSP software (I'm using REW V5.20 beta 48)
4) Click on the "Generator" button at the top of REW
5) On the "Tones" tab of the "Generator" select "Tone Burst" and set it to whatever frequency your sub-mid xover is
6) Make sure the "Cycles" button on the right is clicked, select 0.5 in that selection box (The half-wave is perfect to show phase and short enough to not clutter the o-scope with unnecessary info.)
7) Make sure "Repeat the burst" is selected
8) Set your level to whatever you normally use
9) Set output to L+R
10) Check your system volume
11) Click the play button
12) Select a sub and mid to align, mute all other speakers
13) Click the "Scope" button at the top of REW
14) Adjust the mV/div and ms/div to get the waves displayed properly (large enough and wide enough, also to show ALL of the waves from the separate speakers) You may also want to mute each speaker to get an idea of what each wave looks like alone, to better approximate the final product.
15) Now adjust T/A and/or phase to get those waves to merge into one coherent, clean wave. ( Compare the first or largest peaks, the horizontal time divisions will give you a very close approximation of how far off your T/A is, if at all) Take notes! Write down your initial delays (just in case this doesn't work for you). Subtract your initial delay from the final delay, now you have the additional delay. The difference in T/A (additional delay) may need to carry over to your tweets also to keep them aligned with your mids.
16) Repeat steps 12 and 15 for the other mid

This could also be used for mid to tweet alignment, but with an appropriate change to the frequency and an adjustment to the "Scope" settings.
I only used this method to determine the additional sub delay in my system, then added that delay (24ms in my case) to the front stage overall delay (in the loop mentioned above), the original tape measure determined delay now works as intended.

Or, you could just do the math.

I thought this might be helpful information for anyone else having T/A or phase problems that they can't diagnose.

Also, sorry about the formatting, don't laugh, I spent several hours on this, it's hard to be coherent when I keep coming back to edit.

I suspect ErinH's video on using a "subwoofer for midbass" actually gives some clues to a solution... since I also suspect this is just inherent in your car's acoustics. Those "acoustic nulls". I'm sure you've already seen his vid.
Tackling those is tough, because you've got a car that's a rigid shape, and you can't really control much of the reflection and absorption patterns without eliminating glass and conversely turning your seats hard.
But I've been rolling it around in my head.

I haven't used my O-scope for time alignment, but you've given me a fun little experiment there.

To me the timing makes sense - in two ways. One, your sub IS further back. Two, I don't know what processing you have, and technically electronics work at the speed of light, but in reality it's not really true. Any dynamic component (inductor, capacitor) can definitely impart a delay on a circuit. If you even have two amps, I wouldn't be surprised. If you have one amp that runs through a processor and the other not... then likely too.

Quote:

---

Originally Posted by geolemon

If you have one amp that runs through a processor and the other not... then likely too.

---

This is especially challenging on the home theater side since any processor or device imparts a minuscule delay, which is easy enough to deal with on the audio side with a processor but gets challenging when needing to sync up with video. Depending on the EQ and processing done, this can also impart slight delays. So for me, I see this as normal tuning where you also need to look at phase to make sure things are lining up “perfectly”. I wouldn’t necessarily use an oscilloscope though that is an option as much as I would look at the phase graphs. When mixing ported and sealed subwoofers, I’ve used all-pass filters to ensure the phase of each subwoofer lines up, which ensures the best sound.

Quote:

---

Originally Posted by geolemon

I suspect ErinH's video on using a "subwoofer for midbass" actually gives some clues to a solution... since I also suspect this is just inherent in your car's acoustics. Those "acoustic nulls". I'm sure you've already seen his vid.
Tackling those is tough, because you've got a car that's a rigid shape, and you can't really control much of the reflection and absorption patterns without eliminating glass and conversely turning your seats hard.
But I've been rolling it around in my head.

I haven't used my O-scope for time alignment, but you've given me a fun little experiment there.

To me the timing makes sense - in two ways. One, your sub IS further back. Two, I don't know what processing you have, and technically electronics work at the speed of light, but in reality it's not really true. Any dynamic component (inductor, capacitor) can definitely impart a delay on a circuit. If you even have two amps, I wouldn't be surprised. If you have one amp that runs through a processor and the other not... then likely too.

---

MiniDSP C-DSP 8x12 for processing, 2 JL Audio XD 400/4's for mids and highs, 2 Wolfram Audio W3000.1's on subs, Pioneer DEH-80PRS HU in standard mode with no xovers or T/A.
This was just a check that opened up a can of worms (questions). I'm just trying to figure out where 24 ms of additional delay came from; xover induced, processing delay, vehicle acoustics, or all of the above.

Quote:

---

Originally Posted by dgage

This is especially challenging on the home theater side since any processor or device imparts a minuscule delay, which is easy enough to deal with on the audio side with a processor but gets challenging when needing to sync up with video. Depending on the EQ and processing done, this can also impart slight delays. So for me, I see this as normal tuning where you also need to look at phase to make sure things are lining up “perfectly”. I wouldn’t necessarily use an oscilloscope though that is an option as much as I would look at the phase graphs. When mixing ported and sealed subwoofers, I’ve used all-pass filters to ensure the phase of each subwoofer lines up, which ensures the best sound.

---

No need for a dedicated o-scope, just REW's o-scope function. I tried all pass filters, phase inversion, and extra delay on the mids and highs, but none of those or any combination had the expected effect. I was always left with imperfect blending. Finding that 24 ms delay explained the non-textbook behavior, now I just want to know where all of that delay gets created. The Wolfram amps do have a phase adjustment on them, so there's a possible cause that I haven't investigated yet.

Quote:

---

Originally Posted by ckirocz28

I looped one output back to an input (for the front stage) and added the additional delay to that loop, problem solved. Unexpectedly, there are no ill effects from the loop; and the subs and mids blend perfectly.

---

Yep that's a neat trick!

Quote:

---

Originally Posted by ckirocz28

Experts please comment on this.>> I'm talking specifically to Justin Zazzi, ErinH, and any other electrical/acoustic engineering types.>> Before I move on, all xover slopes mentioned are acoustic slopes NOT electronic. I can only guess that this additional delay is from the LR24 slopes on either side of the 90hz xover point, one full wavelength delay from each slope adding up to 22.22ms delay, that seems about right. Should we calculate additional delay like this? >"Xover point wavelength in ms" X "number of cycles (360 deg) phase change in slopes (LP+HP)" = "additional delay to add to the high pass side".

---

I think there is some utility to this idea, however you would be time aligning at one particular frequency at the expense of all others. Ideally you want to time align at all frequencies at the same time. I've had the best success with creating an acoustic response that matches a textbook filter such as a 4th-rder high-pass Linkwitz-Riley. If the acoustic response matches the target, then the phase response should be predictable. If you do this with both the high-pass on your midbass and the low-pass on your subwoofer then they will be a predictable 360 degrees out of phase at all frequencies, meaning they are in phase at all frequencies.

I do not think they are time aligned at all frequencies, however, since 360 degrees of phase delay is still one cycle of delay at every frequency. The problem with trying to time align them is if you adjust the crossover frequency to be exactly time aligned such as using the oscilloscope method, then every other frequency is now both not time aligned and also not in phase either. SO if you line up a half-cycle burst at 80hz on the o-scope precisely, then when you change to a 50hz burst they will not be aligned anymore.

I admit I have not tried it, but this is how I understand it would work. I would love to see you give it a try and share what you learned!

Quote:

---

Originally Posted by Justin Zazzi

If you do this with both the high-pass on your midbass and the low-pass on your subwoofer then they will be a predictable 360 degrees out of phase at all frequencies, meaning they are in phase at all frequencies.

I do not think they are time aligned at all frequencies, however, since 360 degrees of phase delay is still one cycle of delay at every frequency.

---

Okay hang on. Here's an idea. I would love chirocz28 to give this a try and see what happens.

So I don't think your method would work because you are using a burst that is one half cycle of a wave. If you line up the peak from two speakers to the same point in time, then that's where I think this idea would fall apart.

But what if .... what if you line up the peaks of two speakers to be exactly one cycle delayed from eachother? If the peak from one speaker is delayed exactly one wavelength compared to the peak from the other speaker, then you have 360 degrees of phase difference at a single frequency. If you have 360 degrees of phase different at one frequency, then in theory you should have 360 degrees of phase difference at all frequencies for a 4th-order filter.

So instead of using a burst of one half cycle, what if you use a burst of 2 cycles long? Then you can clearly delay one speaker by one wavelength and you will have one wavelength of overlap between the two speakers to do your time alignment with. Something like this:


Attachment 10583

I think the middle cycle where the two speakers overlap will be twice as tall because they should sum together.
At least, this is how it might look in a textbook before real-life makes it messy.

OMG you HAVE to try this!


Attachment 10584

Ok, after some modeling, I've learned that the xovers only add group delay (same group delay for LP and HP) and the processing time is under 1ms in most cases, so totally ignore the above mentioned math.
After some more experimenting, I've discovered the cause, my subwoofer amps have a phase adjustment control that seems to be a straight delay circuit, set at 180 degrees there is 10ms less delay than having it set at 0 degrees. So I am totally blaming the subwoofer amps for the additional delay. Maybe someone else with an amp with a phase control can test their amp and see if this is the case for all of this type phase control.

I hope this little investigation will help someone else.

Quote:

---

Originally Posted by Justin Zazzi

I think the middle cycle where the two speakers overlap will be twice as tall because they should sum together.
At least, this is how it might look in a textbook before real-life makes it messy.

OMG you HAVE to try this!


Attachment 10584

---

I've actually tried this just this morning, sort of, while doing more investigation, which I just posted about. My audio OCD wouldn't allow me to leave it that way. But, yes, if the waves are lined up in phase and at the same amplitude, the result is twice as tall or double the voltage or the textbook 6 db increase. The math was pointless in this case, the subwoofer amps' circuitry was adding the delay, most likely related to the phase adjustment. The group delay added by the xovers is consistent between the low pass and high pass slopes as long as they are matching slopes, so I think an all pass filter will bring things nicely in line at frequencies away from the xover slope now that I've discovered and corrected the extra delay.
I do absolutely appreciate the suggestions though.

Quote:

---

Originally Posted by Justin Zazzi

I think the middle cycle where the two speakers overlap will be twice as tall because they should sum together.
At least, this is how it might look in a textbook before real-life makes it messy.

OMG you HAVE to try this!


Attachment 10584

---

I'll try some of this in the next couple of days, but I think for an 80 hz xover example (one cycle is 12.5 ms) the added delay would just delay every frequency by 12.5 ms, making things much worse. Enter the all-pass filter, for the win, maybe, if real life doesn't screw it up.

I almost forgot the screenshots of the different delay from the phase adjustment on the amplifier. First is the 0 degree setting, second is the 180 degree setting, horizontal scale is 50 ms per major division with hash marks at 10 ms, vertical scale is 100 mV per major division with hash marks at 10 mV. Excuse me, 20 mV hashes.
Attachment 10626
Attachment 10627

Invalid links :(

Quote:

---

Originally Posted by ckirocz28

Ok, after some modeling, I've learned that the xovers only add group delay (same group delay for LP and HP) and the processing time is under 1ms in most cases, so totally ignore the above mentioned math.
After some more experimenting, I've discovered the cause, my subwoofer amps have a phase adjustment control that seems to be a straight delay circuit, set at 180 degrees there is 10ms less delay than having it set at 0 degrees. So I am totally blaming the subwoofer amps for the additional delay. Maybe someone else with an amp with a phase control can test their amp and see if this is the case for all of this type phase control.

I hope this little investigation will help someone else.

---

If it is truly a phase adjustment knob, then it will be a variable time delay that is frequency dependent. A 10ms delay at 80 doesn't mean you'll get a 10ms delay at 70hz too.

If it is a polarity switch and mislabeled as a phase knob (which is totally possible), then it will just make the ups and downs on the scope go flippy-floppy.

Quote:

---

Originally Posted by Justin Zazzi

I think the middle cycle where the two speakers overlap will be twice as tall because they should sum together.
At least, this is how it might look in a textbook before real-life makes it messy.

OMG you HAVE to try this!


Attachment 10584

---

I wish I was set up to try this - I have ZERO equipment with a delay set up right now. I have an O-scope, so technically I could perfectly align them...

If someone can do this, here's my prediction:
You wouldn't notice. Especially at higher frequencies. At least not with a single frequency.
The problem with this also is that it's ONLY going to align at a single frequency. So in actuality, with full-range music - you'd definitely notice.

The reason is - if you aligned it at, say, 100hz - then at 100hz, you probably wouldn't notice. The waves are perfectly aligned -zero loss, really. You'd never notice "one wave's worth" on one side or the other.
But what you've really done is create a delay with a time factor equal to one wave (whatever time that calculates out to at 100hz, using the speed of sound and 100 cycles per second).
So - progressively - as you move away from 100hz in the lower direction, your waves will fall out of alignment and gradually cancel increasingly...if I'm thinking about this right (one octave = half or double the frequency), 50hz would cancel. Maybe I'm thinking of time wrong, or not thinking of it all in my head - but there will be some point where it shifts enough to cancel entirely.

And - as you move away from 100hz in the upper direction, your waves will fall out of alignment and gradually cancel increasingly also...
BUT - we have many more octaves in this direction. So I think by the time you reach 200 cycles per second (unless, again, my mental math is not factoring in time), you'll completely cancel - but then by 400hz you'll be back in alignment. 800 hz, more cancellation... 1600 hz back in alignment. Etc.

So you create that "comb pattern" that is bad news:
https://audiouniversityonline.com/co...ing-explained/
The leading image there shows an actually aligned response on top, and a slightly offset response on the bottom (as measured from a listening position).

Not to complicate things further - but we inherently suffer this in car audio. Just from sitting off-center, in a car - you get this delay, inherently. And while you can fix it for one side OR the other by introducing delay, you can't fix it for both sides - if you add delay to the left speaker to fix it for the driver's side, you inherently compound the problem for the passenger's side. It's why people try to put their stereo speakers at the furthest distance possible for any given interior, as that equalizes the two pathlength distances - therefore minimizing this delay effect.

Quote:

---

Originally Posted by Justin Zazzi

If it is truly a phase adjustment knob, then it will be a variable time delay that is frequency dependent. A 10ms delay at 80 doesn't mean you'll get a 10ms delay at 70hz too.

If it is a polarity switch and mislabeled as a phase knob (which is totally possible), then it will just make the ups and downs on the scope go flippy-floppy.

---

I watched the o-scope as I turned the phase adjustment, it literally just added 10 ms delay at 0 degrees and none at 180 degrees (other than the base 14 ms delay), with no change between the two.

https://www.crutchfield.com/S-i3PJ6R...-NOX4BDSP.html