DiyMobileAudio.com Car Stereo Forum banner
41 - 51 of 51 Posts

·
Registered
Joined
·
4,057 Posts
One question (well, actually it's more than one...). Isn't this phase shift of the cars acoustics done to all drivers at the same time? F.e. in the area where you pass over the duty from midrange to tweeter?
If yes, the relative phase would stay the same even due to the induced phase shift by the car, right?
In this case, an allpass filter as created by either PhaC or the Helix phase control would still work as intended.
Only if both drivers were in the exact same place in the car literally, even 5” centre to centre will change the reflections which will then potentially influence phase

this is why driver positioning and crossover freq have to be chosen very carefully

Playing a midrange below beaming for example doesn’t mean you can mount it anywhere as some people would have you believe, an inch difference… (aiming will change off axis response and also move this centre to surface distance in smaller degrees so does need consideration also) up/down/back/forwards can make a decent sized difference to the response in both phase and also magnitude response 👍🏼
 

·
Registered
Joined
·
512 Posts
Only if both drivers were in the exact same place in the car literally, even 5” centre to centre will change the reflections which will then potentially influence phase

this is why driver positioning and crossover freq have to be chosen very carefully
Yes, but these reflections will be the very same at all times, right? No matter if you set a crossover or driver specific filters, the reflections will be always the same and therefor relative phase will stay stable. It is not like the distance changes between drivers while listening.
Therefor this cannot be the answer. When i set the time alignment without crossovers (in theory that is, of course you need crossovers) the direct sound will all arrive in time, right?
When i set crossovers and driver specific filters arrival times will be changed, especially the lowpassed drivers will lag behind the highpassed drivers, no matter the always existing reflections. When i set an appropriate allpass filter on the highpassed speaker i can bring the arrival time back in time to the lowpassed speaker as the reflections will be stable.

Or is this way of thinking too simple?

Just trying to understand.
 

·
Registered
Joined
·
4,057 Posts
Yes, but these reflections will be the very same at all times, right? No matter if you set a crossover or driver specific filters, the reflections will be always the same and therefor relative phase will stay stable. It is not like the distance changes between drivers while listening.
Therefor this cannot be the answer. When i set the time alignment without crossovers (in theory that is, of course you need crossovers) the direct sound will all arrive in time, right?
When i set crossovers and driver specific filters arrival times will be changed, especially the lowpassed drivers will lag behind the highpassed drivers, no matter the always existing reflections. When i set an appropriate allpass filter on the highpassed speaker i can bring the arrival time back in time to the lowpassed speaker as the reflections will be stable.

Or is this way of thinking too simple?

Just trying to understand.
Each driver will have its own reflections and phase issues to deal with, they will remain constant yes. The phase shifts and rolls round because of them… but each driver will be unique to its own phase shifts, not every driver will have identical shifts because they aren’t in identical places

move a mid up a pillar and it’s interaction with the car will shift in various ways

it’s why you should always put crossovers where both drivers have no big dips etc

in theory if phase was as it left the speaker then yes phac could work… however the environment skews phase very often and it’s not possible to put crossovers in a perfect textbook place

So we sometimes have to make small adjustments for the phase being out to bring it back in

it’s the environmental phase shift, no matter how constant… that phac can’t possibly predict or correct for, how can software with no idea of if mids are in upper doors or a pillars correct for phase being off or slightly out at the listening position? It also can’t predict the drivers own phase shift, be that big or small

Too many variables to predict with not enough information to predict them

Phase leads or lags with crossovers… it doesn’t change times of arrival or impulse response would change… I had this very discussion with the jl guys because it looks like timing will change, when in fact the phase rotates… it’s a hard concept to get your head around at times
 

·
Registered
Joined
·
5,929 Posts
If you talk to Frank Miketta about the theory it totally makes sense to use a tool like PhaC or the Helix phase control for exactly this purpose.
From the workflow point of view with PhaC you are correcting phase differences that are caused by crossovers through the different slopes, especially with odd order crossovers and driver specific filters that all have an effect on phase after you have done the time alignment, and you do the PhaC on both side drivers, not only on one driver, and you do the PhaC on the crossover freqency, not an arbritary frequency that you see fit.
As allpass filters cause a frequency dependant delay they actually can work under these circumstances as you f.e. usually need to delay the highpassed speakers due to the phase shift caused by crossover slopes.
But again, it only works correctly if you're doing the time alignment as the very first step before setting driver specific filters including crossovers.

Problem is, you always have to measure if you really have an actual problem in this area after the car messed around with the acoustical phase due to all the reflections and stuff like that.

I personally listened to cars where the PhaC worked like expected and intended, but i also heared my own car where it didn't really work as expected and intended.

Again , I’m a mosconi dealer , I tune a LOT of Aeros

it’s sorta works, it’s about the time that you spend about two hours fiddling with it and you get it to sound good, and then you leave the car and come back and turn it on again to a complete disaster because your ears weren’t ready for the mess that you caused…

so yeah like I said before, it makes the passband great, stop the band is a catastrophe (using phac calculator as frank designed)

I’m not a fan of really good coherence that’s narrow banded , then a mess outside of that area

I would rather sum the crossovers properly to have good allpass behavior (as a crossover network sums to be an all pass) and that’s the only time I want to move the phase in a minim phase system

as far as everything else as long as your magnitude is right that phase will fix itself. so there’s no point in trying to fix something that’s not broken when all you need to do is a better job at tuning…..


and as far as the dips again like I said the energy is still there! just because your microphone shows a dip in one spot don’t mean it’s not there , perhaps on the other side of the car…..

If you think about it like this plus take for example a 400 Hz all pass filter, with a Q around .5 or 1

The lagging side of that would be about 4 to 6 ms behind the leading side, that as a group delay why in the world would you want to add group delay 8-12 times!!! that’s an insane premise!

I can see sometimes using one way way way out of band on a driver that’s just not cooperating, and you have nothing left to do. But that’s A single driver!!! I would never ever ever add that to every driver in the networks on a multi way!

i’m surprised it doesn’t start reverberating with all of that wrapping going on!! Jesus Christ, it’s like , “ hey guys let’s intentionally add 8 more phase wraps , because it’s cool” 🤤🤤

Why would you make the compromise and trade should have one tiny part a little bit better in exchange for a very big part to be a complete disaster? Nicht verstehe!!!
 

·
Registered
Joined
·
5,929 Posts
Yes, but these reflections will be the very same at all times, right? No matter if you set a crossover or driver specific filters, the reflections will be always the same and therefor relative phase will stay stable. It is not like the distance changes between drivers while listening.
Therefor this cannot be the answer. When i set the time alignment without crossovers (in theory that is, of course you need crossovers) the direct sound will all arrive in time, right?
When i set crossovers and driver specific filters arrival times will be changed, especially the lowpassed drivers will lag behind the highpassed drivers, no matter the always existing reflections. When i set an appropriate allpass filter on the highpassed speaker i can bring the arrival time back in time to the lowpassed speaker as the reflections will be stable.

Or is this way of thinking too simple?

Just trying to understand.
Reflections can be a variety of diffraction (the worst) refraction , and a few other terms which I can’t think of atm (I’m drinking rn so this is off the cuf)

so it depends , most of the time as long as it’s not a compound reflection (a reverbant reflection ) then even simple eq can change the timing between frequencies that can alter the reflection to have different characteristics, which can be beneficial or detrimental….

You just kinda have to remember there almost is no such thing as direct sound in a car…. Most of the frequencies don’t exactly fit in a car…..

I would recommend using The wavelet, you will be able to see the time domain much clearer…

All you really need to do is use an acoustic timing reference and a single point measurement in the center of your ears and do a left and right and you can really see what’s going on… (of course zoom in because a car will never be above 200 ms as far as window size)

pretty much everything I’ve ever tuned was around 190 ms window size, to have a late reflection above 200 ms in a car is pretty much unheard of…

that being said it should be very very easy to see in the wavelet what’s happening…

I would recommend using a 15 cycle window below the Schroeder, and then a 4 to 8 cycle window above that, and then compare and you will see your time domain very clearly and it will become perfectly obvious where your reflections are.

of course it’s always easier to see this with the crossover‘s off and doing single driver measurements, The crossover will alter the time domain and you want that speaker flat baby flat!

You could also take impedance measurements, which is basically the density times the wave speed, z=Pv
 

·
Registered
Joined
·
4,057 Posts
Again , I’m a mosconi dealer , I tune a LOT of Aeros

it’s sorta works, it’s about the time that you spend about two hours fiddling with it and you get it to sound good, and then you leave the car and come back and turn it on again to a complete disaster because your ears weren’t ready for the mess that you caused…

so yeah like I said before, it makes the passband great, stop the band is a catastrophe (using phac calculator as frank designed)

I’m not a fan of really good coherence that’s narrow banded , then a mess outside of that area

I would rather sum the crossovers properly to have good allpass behavior (as a crossover network sums to be an all pass) and that’s the only time I want to move the phase in a minim phase system

as far as everything else as long as your magnitude is right that phase will fix itself. so there’s no point in trying to fix something that’s not broken when all you need to do is a better job at tuning…..


and as far as the dips again like I said the energy is still there! just because your microphone shows a dip in one spot don’t mean it’s not there , perhaps on the other side of the car…..

If you think about it like this plus take for example a 400 Hz all pass filter, with a Q around .5 or 1

The lagging side of that would be about 4 to 6 ms behind the leading side, that as a group delay why in the world would you want to add group delay 8-12 times!!! that’s an insane premise!

I can see sometimes using one way way way out of band on a driver that’s just not cooperating, and you have nothing left to do. But that’s A single driver!!! I would never ever ever add that to every driver in the networks on a multi way!

i’m surprised it doesn’t start reverberating with all of that wrapping going on!! Jesus Christ, it’s like , “ hey guys let’s intentionally add 8 more phase wraps , because it’s cool” 🤤🤤

Why would you make the compromise and trade should have one tiny part a little bit better in exchange for a very big part to be a complete disaster? Nicht verstehe!!!
That reads like all pass adds delay? They don’t add any delay in the signal, the impulse response doesn’t change at all

nor does it change through crossovers either, the phase rotation doesn’t effect time at all above the crossover
 

·
Registered
Joined
·
5,929 Posts
That reads like all pass adds delay? They don’t add any delay in the signal, the impulse response doesn’t change at all

nor does it change through crossovers either, the phase rotation doesn’t effect time at all above the crossover
So , actually it does ,

but I think your just misunderstood

there’s a way you can see it , download JL TUN,

click on the crossovers icon , and it shows what you have going and it shows how many ms that LPF is set back ,

They sum to be an all pass , (that’s what it all passes a low pass in a high pass summed together) look at an all pass , the “low side” is one cycle out of time (for 2nd order) with the “high side” which isn’t out of time

that one cycle equates to a center frequency, depending how much phase (or radians) is in that circle , will dictate the delay of the low side

And it does affect the impulse , if it moves the timing it changes the impulse , whether it does it negatively or positively two different discussions
it’s obviously changing the timing somewhere or we wouldn’t be having this discussion lol

transfer function of an allpass here for example

Rectangle Slope Font Parallel Circle

Rectangle Slope Line Font Parallel
 

·
Registered
Joined
·
4,057 Posts
So , actually it does ,

but I think your just misunderstood

there’s a way you can see it , download JL TUN,

click on the crossovers icon , and it shows what you have going and it shows how many ms that LPF is set back ,

They sum to be an all pass , (that’s what it all passes a low pass in a high pass summed together) look at an all pass , the “low side” is one cycle out of time (for 2nd order) with the “high side” which isn’t out of time

that one cycle equates to a center frequency, depending how much phase (or radians) is in that circle , will dictate the delay of the low side

And it does affect the impulse , if it moves the timing it changes the impulse , whether it does it negatively or positively two different discussions
it’s obviously changing the timing somewhere or we wouldn’t be having this discussion lol

transfer function of an allpass here for example

View attachment 348338
View attachment 348339
I was referring to you saying the all pass adds msec to the signal… it doesn’t or the arrival time would move if you applied it to freqs below where the signal is

it does change the impulse shape but not the actual arrival time of the direct sound above the all pass was my point, it literally effects the phase at the freq it’s at ‘if’ there is output at the freqs… it’s why applying a second order all pass at the subwoofer crossover point has zero effect on a tweeter which I’m sure I’ve seen you suggest before
 

·
Registered
Joined
·
5,929 Posts
I was referring to you saying the all pass adds msec to the signal… it doesn’t or the arrival time would move if you applied it to freqs below where the signal is

it does change the impulse shape but not the actual arrival time of the direct sound above the all pass was my point, it literally effects the phase at the freq it’s at ‘if’ there is output at the freqs… it’s why applying a second order all pass at the subwoofer crossover point has zero effect on a tweeter which I’m sure I’ve seen you suggest before
I get ya now :)

okay that makes more sense to me sorry dum
 

·
Registered
Joined
·
5,929 Posts
That reads like all pass adds delay? They don’t add any delay in the signal, the impulse response doesn’t change at all

nor does it change through crossovers either, the phase rotation doesn’t effect time at all above the crossover
Exactly, above the crossover….
below it is a delay, same with an all pass…. The low side past quadrature is a GD and it’s constant.

so an out of band APF below the passband of a driver , okay so what , depends how out of band were talking , if your trying to move meaningful phase in the passband , it can’t be that far off (like you said about the tweeter)

if it’s not that far off, then most likely it’s in the stop band in the interaction area…..which puts out out of time with the driver below it in the lower stop band…. That’s audible.

I mean, if your only worried about the passband , then by all means. I’ve tried it on a number of cars and wasn’t impressed

there was a tundra that it worked just ok on the right door only…. But not all the speakers

it did make the doors work nice together, but still had artifacts that made me like …… think, “is this really worth it” …….and I removed it and worked around it. It was causing serious issues with dash midrange on the same side…..

Maybe , just not my cup of tea perhaps :)
 
41 - 51 of 51 Posts
Top