A SMAART Question......it's about phase

J

Jdunk54nl

Noob
A SMAART Question......it's about time (really phase but play on words)

I posted this in my build thread but figured it might be good to have a dedicated thread to this so others can easily find the answer too. I think this is an important question that can lead others to get much better tunes.


I finished my tune the other day with these moving mic measurement results: The really good looking pink lines are the house curves.
View attachment 16734


I also have access to Smaart v8 to measure phase in real time and this is where my questions stem from:

Copied and pasted from my build thread:

Ok question for some of you that may know ( @dumdum or @Justin Zazzi)

This was my tweeter phase response after getting it the best I knew how.

Purple is my left tweeter and dark green is my right tweeter.
I circled in green where my questions stems from.

Here are my questions:

1) Should I even worry about this area since my crossover for tweeters is at 3500hz acoustically and it gets the weird wrapping phase at about 3000hz and that is -12db down?

2) If I were to worry about this, how would I fix it? As in, if this was somewhere at like 8000hz in the middle of the response. I tried some all pass filters on this and nothing would fix it. They would shift the weird wrap either higher or lower, but nothing would get rid of it.
View attachment 16735




Here are my mids, anything you see I should fix here? This one has a break in the phase at about 1290hz in the left mid (green). The right mid is the dark red. I tried an all pass filter but it didn't really fix it. I tired 2nd order and different q's for it. Do you worry about those?

View attachment 16736

Here are mids to tweeters: Left mid is green, left tweeter is orange, right mid is pink, right tweeter is blue.
View attachment 16738




Here are mids to woofers: Left mid green, left woofer bright green, right mid ink, right woofer brown

View attachment 16739



Here are my woofers and subwoofers, I don't see much to fix here. Subwoofer dark green, right woofer red, left woofer blue: Note* I forgot to lower the sub by 6db (since both woofers playing will raise the total response by 6db and I want my subs 10db louder than the woofers, in the below graph they are 16db louder than an individual woofer) in this below graph before grabbing the screenshot. I realized this and double checked the phase by lowering it 6db and it didn't change anything (as it shouldn't) phase wise but forgot to grab a new screenshot of that.
View attachment 16737
 
Final question about all of the above:


Once I was done doing that work, I got into the truck and was listening. My sound stage seemed to be shifted a few inches to the left of center. I tried changing left side overall levels by lowering them but the stage didn't shift at all. Should I mess with the timing? That seems like it would then mess up the phase responses. I tried that and it did shift the stage if I increased the delay on the right side as a whole by a couple clicks, but I know this definitely impacts the phase response.
 
Re: A SMAART Question......it's about time (really phase but play on words)

It seems we really need a tutorial about what delay and all-pass filters do to phase and how to properly "fix" phase problems.
 
Jdunk54nl said:
Final question about all of the above:


Once I was done doing that work, I got into the truck and was listening. My sound stage seemed to be shifted a few inches to the left of center. I tried changing left side overall levels by lowering them but the stage didn't shift at all. Should I mess with the timing? That seems like it would then mess up the phase responses. I tried that and it did shift the stage if I increased the delay on the right side as a whole by a couple clicks, but I know this definitely impacts the phase response.
Click to expand...
You really do need to forget about the position of the speakers, you say the stage was a few inches left of centre… centre of what? The car or between the perceived location of the hard left and right of stage, the latter is what you should be listening for…

most cars with pillars or dash mounted drivers have a reflection on the drivers side window

think how we hear, we get the location of a sound source from all sources of sound, so if you have strong reflections from the side window and above the driver then we will ‘hear’ the source of the sound as being outside the speaker towards the side window…

and also the passenger side we hear the sum of all the sound sources… it’s not uncommon to hear the passenger side several inches in from the pillar or speaker location…

so we can have two sources of sound that are over to the left… what will that do to your centre if it is timed correctly and between left and right as our brain decides there location? Just what you experience

people get hung up on the centre of the car being the centre, work out where your stage boundary’s are and then work out where the centre should be located… it’s likely not the centre of the car!

as for the phase, don’t get hung up on worrying about phase wraps above midranges and somewhere around 2khz, tbh I only tend to use all pass below 1khz max, installation location of drivers to get clean phase is far more important in the first instance

clean phase between two drivers gives depth of stage, along with correct eq work and freq response
 
@Justin Zazzi's response in my build thread: https://www.caraudiojunkies.com/showthread.php?3928-2014-F150-Limited-Build
Copied and pasted

I would not worry about phase alignment above 1kHz 'ish. If your tweeters are Goofy at 8khz, I wouldn't bother.

If you wanted to though, I think all pass filters are a good start. They can only bend the phase trace one direction so if you're not getting what you want, try all pass filter on the other channel instead.

It is curious to see the knee in phase response though. Usually I see a sharp transition like that when I'm measuring a 2 way or a coaxial set because the acoustic center of the midrange and tweeter are different distances from the mic. So in that case the phase plot would be two straight lines that intersect somewhere, and you get a shape that looks much like your tweeter phase does.

I would double check that the frequency response is what you think it is. Frequency and phase response are linked so adjusting one can help the other match.

Oh, also reflections.


Second Response from Justin Copied and Pasted:

A couple clicks of time alignment is insignificant phase change in the lower frequencies where phase becomes more meaningful. If you can improve your stage with a small time adjustment, go for it?

I would also focus on what frequencies in the stage are not centered. Have you tried playing bandwidth limited pink noise tracks? I would try that with only one pair of channels engaged at a time, like woofers only, then mids only, etc. Somewhere around here we have some Tuning Stones that work great for this :)
 
dumdum said:
You really do need to forget about the position of the speakers, you say the stage was a few inches left of centre… centre of what? The car or between the perceived location of the hard left and right of stage, the latter is what you should be listening for…

most cars with pillars or dash mounted drivers have a reflection on the drivers side window

think how we hear, we get the location of a sound source from all sources of sound, so if you have strong reflections from the side window and above the driver then we will ‘hear’ the source of the sound as being outside the speaker towards the side window…

and also the passenger side we hear the sum of all the sound sources… it’s not uncommon to hear the passenger side several inches in from the pillar or speaker location…

so we can have two sources of sound that are over to the left… what will that do to your centre if it is timed correctly and between left and right as our brain decides there location? Just what you experience

people get hung up on the centre of the car being the centre, work out where your stage boundary’s are and then work out where the centre should be located… it’s likely not the centre of the car!

as for the phase, don’t get hung up on worrying about phase wraps above midranges and somewhere around 2khz, tbh I only tend to use all pass below 1khz max, installation location of drivers to get clean phase is far more important in the first instance

clean phase between two drivers gives depth of stage, along with correct eq work and freq response
Click to expand...


It was left of center of dash. You are right, I need to rethink what center should be.

I did find out one of the reasons, apparently I sit shifted to the left, if I sit in the middle of my seat like you are supposed to, it is centered. I moved my head about the same distance the center image was off and this fixed most of the issue. I guess if I want to continue to sit like that, I need to make sure my microphone is actually in that position that my head would be in, and not in the middle of the headrest like I had it when measuring in the above. It is a difference of a few inches.


Also, thanks for the info on phase, it corresponds with Justin's advice too. This is why I asked. I figured it was something that the data is showing me but my ears can't really interpret. I know at those frequencies it is usually more volume dependent than it is phase dependent for positioning. Sometimes we need to learn to ignore some of the information our measurement devices are giving us. I have been better at this with REW, but not with smaart.


Unfortunately, I am not skilled enough nor do I want to get into all of the work it takes to proper position drivers beyond what they are. My tweeters and mids are currently in Valicar pods and the tweeters (swivel mount option) are angled to the best I could get with their swivel mounts and within their location limitations to mount them without doing way more work or cutting things I am unwilling to cut.
 
Also, I really love that we can have these deeper conversations on this forum, as in, there are people that can answer and I trust their answers.
 
It's a little off but closely related to the topic but... Justin Zazzi mentioned the "Tunning Stones" method which always makes me smile :) I have been playing around with using the NPR Tiny Desk Concert Series to help with visualizing the staging. I stole it from Matt Schaeffer and Gary Bell as they talk about it on their podcast, its something they use when verifying staging in their work and when demoing vehicles for the customer. I wonder what your experience would be with it??? I have been fortunate enough to hear your latest tune which sounds fantastic and when i was in the driver's position everything appeared to play very transparent, sounds came from where they should, i could mentally see what i was hearing and the overall experience left me rethinking my approach and as you set a high bar for me to try to reach!!!
 
I wish I could chime in helpfully here... but I can't.
I'm going to be playing with my first AP filters in my own DSP install, first I've ever personally owned, and first I've taken further than an initial tune - which still isn't complete.

I definitely want to get my hands on SMAART because so SO SO much is hidden invisibly and potentially frustratingly, after you've done your EQ-ing, aligning, and even acoustical interior treatments. I've never had it that tool OR a way to mitigate those issues, so being able to actually MEASURE those phase shifts, apply AP filters, then observe your impact in the phase realm - just awesome.


I suspect your phase anomolies are actually DUE to your crossovers - as the sound rolls off below the high-pass frequency, there is a phase shift, no matter whether you picked bessel or butterworth or linkwitz-riley or anything - in fact that's the reason system designers have historically tried to use as few drivers as possible. Ideally a single point-source driver would most accurately reproduce the sound because it is perfectly phase-correct. Even if you had a mid and tweeter that were theoretically perfectly flat in dB level from 20hz up to 20khz and beyond, you need that crossover point - which even if perfectly aligned so there's no dip in frequency response, it does mess up the phase on both side of the Xover point... for the tweeter, below it, and for the mid, above it.
The steeper the slope, the faster they roll off, but also the faster the phase shift happens.
Traditionally it's been impossible to really correct for this - but it seems like with the right tools and DSP/filters - I mean wow, this really is one of the old-school fundamentals being tackled here.

If - *if* you could actually flatten the frequency response across a crossover point, AND flatten the phase response across the crossover point, you absolutely could dramatically improve your imaging.
I almost typed "to the degree where you emulate that theoretical-ideal point-source driver", but unfortunately the complex reflections in our car interiors would prevent us from QUITE getting to that perfect ideal, since the two drivers do need to be in two different locations in space physically - but maybe it would open the door for systems more like my current install (sub... midbass... midrange... wideband... tweeter... read "way too many Xover points for phase coherence) to be as good as my old-school ideal systems for imaging (sub... 6.5 midbass/midrange... tweeter that can reach from below 2khz to over 20khz). The old school ones WERE difficult to nail imaging particularly right around the Xover point - which had everything to do with this phase shifting, and that's a fascinating aspect of Psychoacoustics...
Consciously, you can't really "listen" for phase.... other than being aware that the image isn't correct. And it's entirely your subconscious recognizing that the sound isn't exactly correct. It's completely what led me to be fascinated with the very concept of imaging. :loser:
Still - what a tool, so effing cool... At least to a geek like me.
 
It really is an amazing tool. Especially to use polarity flips, all pass filters, other ways to impact phase. Once you are done, there is no second guessing your time/phase alignment. Much like when you are done with REW, there is no second guessing your frequency magnitude response (unless you did something wrong). I can't go back to not using SMAART just like I can't stop using REW (well if smaart would ever implement an EQ window like REW has I would stop using REW)

If you don't have smaart though, head on over to https://opensoundmeter.com/en/
It is a free version of smaart that does nearly everything smaart does.
 
geolemon said:
I wish I could chime in helpfully here... but I can't.
I'm going to be playing with my first AP filters in my own DSP install, first I've ever personally owned, and first I've taken further than an initial tune - which still isn't complete.

I definitely want to get my hands on SMAART because so SO SO much is hidden invisibly and potentially frustratingly, after you've done your EQ-ing, aligning, and even acoustical interior treatments. I've never had it that tool OR a way to mitigate those issues, so being able to actually MEASURE those phase shifts, apply AP filters, then observe your impact in the phase realm - just awesome.


I suspect your phase anomolies are actually DUE to your crossovers - as the sound rolls off below the high-pass frequency, there is a phase shift, no matter whether you picked bessel or butterworth or linkwitz-riley or anything - in fact that's the reason system designers have historically tried to use as few drivers as possible. Ideally a single point-source driver would most accurately reproduce the sound because it is perfectly phase-correct. Even if you had a mid and tweeter that were theoretically perfectly flat in dB level from 20hz up to 20khz and beyond, you need that crossover point - which even if perfectly aligned so there's no dip in frequency response, it does mess up the phase on both side of the Xover point... for the tweeter, below it, and for the mid, above it.
The steeper the slope, the faster they roll off, but also the faster the phase shift happens.
Traditionally it's been impossible to really correct for this - but it seems like with the right tools and DSP/filters - I mean wow, this really is one of the old-school fundamentals being tackled here.

If - *if* you could actually flatten the frequency response across a crossover point, AND flatten the phase response across the crossover point, you absolutely could dramatically improve your imaging.
I almost typed "to the degree where you emulate that theoretical-ideal point-source driver", but unfortunately the complex reflections in our car interiors would prevent us from QUITE getting to that perfect ideal, since the two drivers do need to be in two different locations in space physically - but maybe it would open the door for systems more like my current install (sub... midbass... midrange... wideband... tweeter... read "way too many Xover points for phase coherence) to be as good as my old-school ideal systems for imaging (sub... 6.5 midbass/midrange... tweeter that can reach from below 2khz to over 20khz). The old school ones WERE difficult to nail imaging particularly right around the Xover point - which had everything to do with this phase shifting, and that's a fascinating aspect of Psychoacoustics...
Consciously, you can't really "listen" for phase.... other than being aware that the image isn't correct. And it's entirely your subconscious recognizing that the sound isn't exactly correct. It's completely what led me to be fascinated with the very concept of imaging. :loser:
Still - what a tool, so effing cool... At least to a geek like me.
Click to expand...

you can listen for phase… do your eq and make sure both sides sound the same to your ears (not with a mic)

if they don’t the eq and rta is telling white lies… purely because we hear differently to how a mic reads that picks up everything…

anyway… you want to know how to sanity check your left right time alignment is correct (a mic tip doesn’t quite replicate a set of ears 5-6” apart with head related transfer function) so play music… any music, well recorded and mastered ideally, now feed left and right into both left and right channels, where should all the information come from? The centre, if the timing is correct all information should be pin point in one spot, you will find when you go to far one way or the other the centre will blur to the left or right and you will hear information coming from the left or right, high frequency information will always move first and snap into the centre last over the lower registers just because the high wave lengths are shorter and it will come into phase and out of phase faster
 
dumdum said:
you can listen for phase… do your eq and make sure both sides sound the same to your ears
Click to expand...
I think you were responding to this:
geolemon said:
...that's a fascinating aspect of Psychoacoustics...
Consciously, you can't really "listen" for phase.... other than being aware that the image isn't correct. And it's entirely your subconscious recognizing that the sound isn't exactly correct. It's what led me to be fascinated with the very concept of imaging. :loser:.
Click to expand...
My point is - you can't "hear" phase, and even the things that you hear, it's your subconscious doing for you.

If you installed a component set, and installed one side out of phase relative to the other (the worst case scenario in most cars), I'm not saying "you can't tell one side is out of phase". I'm saying it's psychoacoustics, your subconscious that's somehow able to compare the two sides of the signal and create that confused, diffused sense of "image" in your head. It's that resulting sense of image that you are responding to. You aren't consciously staring over the dashboard and hood and listening for whether the cymbal hit sounds clear - you are consciously staring over the dashboard sure - hoping for a thing to occur in your subconscious - that illusion of depth and space that makes you doubt your eyeballs. The cymbal hit should be... THERE. A location in space, not just a sound. The sound you can listen for with your ears - the location in space requires you to wait for your subconscious to give you something.

The REW-measurable stuff doesn't really change - you could have a nice, flat-ish EQ curve right across the board, even with one side out of phase, and not miss a note, instrument, word, etc that's contained in the song that you can consciously listen for - but "are you fooled"? That's all subconscious trickery - like a sense of balance.
So this SMAART is one step closer (and I'll have to check out that free one- that's a no-brainer), although it's still in the "reproduce what's in the recording" realm.
I wish there really were a way to directly measure for "imaging". Scientifically there must be - maybe a dummy head with two mics placed inside anatomically correct ears and ear canals measuring phase and response... I'm sure we have the measurement power these days. But can we simulate what your subconscious does?

That also brings me back to one of my fallback points for people who argue that SQ is as simple as "reproducing what's on the recording exactly like the engineer intended":
First off, headphones do that. I've even been playing with in-ear monitors a lot lately (there's lots of inexpensive great ones!), and I have some fancy over-the-ear noise-cancelling cans, and my second set (anniversary edition!) of Koss PortaPro on-ear classic legends...
They all have great detail. And yet none of them "image". There's more to it than just "reproducing the audio signal".
Second off, "like the engineer intended" has it's own can of worms assumptions but that's for another thread.

But I think I'm preaching to the choir, here. You know what I mean, I'm sure.
 
And that is my point, our subconscious does things we can’t simulate with a pure measurement sustem, so use your ears for the fine tuning, and I’ve explained a method that you can use, you don’t need to argue how or why we can’t do what we can do as we sure can do it, you just have to set the stereo up to do it, in stereo it’s incredibly hard to hear or find the focus point where it snaps into place, but given certain conditions it gets a lot easier :)
 
So was talking with Nick Ames (JL Audio) and we had a good conversation as to why phase will bend like this and wrap quickly. He made me have another epiphany moment (every time I talk to him this happens).
So, for all of those other people that want to have an epiphany moment here you go :)

Think about it this way, the thing that causes the phase to get to a steeper slope is time delay, as in, if I add delay the phase bends down.

So, if all of a sudden there is an area that the phase is bending down, like in the circled part below, that means there is a time delay of that audio.

The speaker didn't suddenly change it's position for that information, so it is not the cause of this. Instead this would be sound arriving later than what the speaker caused...as in..... a reflection! This also means that the reflection was the dominant source of energy if that causes the phase to wrap like that!

Note: This is assuming that you don't have all pass filters that could also cause something like this.

attachment.php
 
Have you tried unwrapping the phase in the options menu? It sometime makes it easier to see what's going on...at least for me.
If the reflection is dominant, and it's not something you can eliminate, should you tune the system to the reflection instead of the driver itself?
 
Yes, you should definltely eq the dominant source of energy, but this would be hard to diagnose without being able to view phase in real time that the dominant source of energy is not the speaker but the reflection. The ultimate solution though would be to go back and visit your install and fix the issue if you can.

Also to unwrapping the phase, wrapped vs unwrapped doesn't really matter, you still have that increase slope for a short amount of time due to that dominant energy arriving later from the reflection. I like to just use the "scroll" function on the phase by pressing the up/down arrows on the keyboard to get the zone I am most interested in all in the same screen.

Also remember for this example, it is on the tweeter and not worth really fixing (although I probably still would...because ocd....)

But I did just experience this same thing in a dodge ram 2500 truck that was being tuned on the midrange due to placement issues.
 
BigAl205 said:
Have you tried unwrapping the phase in the options menu? It sometime makes it easier to see what's going on...at least for me.
If the reflection is dominant, and it's not something you can eliminate, should you tune the system to the reflection instead of the driver itself?
Click to expand...
Unwrapping phase does not display correctly in smaart or in rew, where the phase wraps the following phase trace steps down, even though the next portion of energy is no later, the phase trace will not really be later, the sound at higher freqs won’t arrive a whole wavelength later… you can add an all pass to emulate this and see what that does to the phase trace… it also doesn’t make the freqs above the all pass arrive later despite a wrap making it look like it does ���� the software just interprets the data as it sees it

the late arriving phase data does in fact corrupt the signal to a degree as stated above, and the software displays it as it ‘sees’ it, but unwrapping is not really helpful that I’ve ever seen
 
Jdunk54nl said:
So was talking with Nick Ames (JL Audio) and we had a good conversation as to why phase will bend like this and wrap quickly. He made me have another epiphany moment (every time I talk to him this happens).
Click to expand...

Omg me too!
Every. Single. Time.


Interesting thought about the dominant energy being from a reflection.
I've seen this kinda thing in the phase plots before too, not very often, but it's always a puzzle.
 
You must log in or register to reply here.
Back
Top