Anything wrong with linear phase through the bass roll off?

[Tim Link]

I saw this post:

New EX MACHINA ARCTURUS 3-WAY ACTIVE STUDIO MONITORS

Price is $25K

audiosciencereview.com

The speaker has a symmetrical step response. I hear people call it pre-ringing, but it's what has to happen if there's a low frequency cut off and the phase is flat down to the cut off frequency. I call it a lack of group delay.
Do we really hear a pink noise ramp up effect from the bass response falling while phase is maintained flat? Is this the result of a steep cut-off in frequency response, or just maintaining a flat phase through a gentle roll off?

I've been confused about why most everybody desires the "textbook" minimum phase step response that has group delay instead of a linear phase symmetrical step response. What's so awesome about group delay?

 

[dasdoing]

I saw this post:

New EX MACHINA ARCTURUS 3-WAY ACTIVE STUDIO MONITORS

Price is $25K

audiosciencereview.com

The speaker has a symmetrical step response. I hear people call it pre-ringing, but it's what has to happen if there's a low frequency cut off and the phase is flat down to the cut off frequency. I call it a lack of group delay.
Do we really hear a pink noise ramp up effect from the bass response falling while phase is maintained flat? Is this the result of a steep cut-off in frequency response, or just maintaining a flat phase through a gentle roll off?

I've been confused about why most everybody desires the "textbook" minimum phase step response that has group delay instead of a linear phase symmetrical step response. What's so awesome about group delay?

I say it all the time, minimum phase might be the goal in analog, but maintaining a minimum phase through the roll-off if you can make it linear makes no logical sense. if there is no negative group delay, there is no pre-ringing

 

[Tim Link]

Here's an interesting experiment that talks specifically about the issue - no complementary driver on the woofer's lower cutoff, leaving the possibility of pre noise. Some can be heard according to this article, but in this case it was extremely faint. So it seems it is possible to get a good sounding woofer with linear phase through the bass roll off. Group delay is not required to avoid audible artifacts.

https://www.bodziosoftware.com.au/LP_MP_Subwoofer_Tests.pdf

One possible concern about using a digital filter to correct phase is that it can cause clipping. The square wave with the phase distorted has a lot louder peaks than the undistorted square wave, so a phase manipulation can add significant SPL. How much extra headroom do we need to have?

 

[ppataki]

I also prefer LP filters for both crossovers (LPF & HPF) but also for simple HPF filtering when I want to remove frequencies below 20Hz or similar
Pre-ringing (or rather pre-echo actually) is audible though if the filter is not implemented correctly - however in terms of the transients (less post-ringing, lower GD) the whole system actually benefits; at least in my ears' opinion

Based on my experience, to avoid audible pre-echo:
- one shall not go steeper than 24dB/octave
- one shall use the highest available resolution of the filter

Check this out for more details about how I use it:

Sphere shaped speakers with CHN110 drivers

Inspiring.:D Also in this thread, speakers were built with Ikea Blanda Matt bamboo salad bowls. I think you got inspiration from that thread: https://audiosciencereview.com/forum/index.php?threads/sphere-shaped-speakers.49807/ You can also build a classic two-way speaker with a sphere shape...

audiosciencereview.com

 

[Tim Link]

I also prefer LP filters for both crossovers (LPF & HPF) but also for simple HPF filtering when I want to remove frequencies below 20Hz or similar
Pre-ringing (or rather pre-echo actually) is audible though if the filter is not implemented correctly - however in terms of the transients (less post-ringing, lower GD) the whole system actually benefits; at least in my ears' opinion

So you're not necessarily linearizing the total phase response, you're just using a linear phase filter to change the roll off rate without affecting the phase?

 

[Tim Link]

I say it all the time, minimum phase might be the goal in analog, but maintaining a minimum phase through the roll-off if you can make it linear makes no logical sense. if there is no negative group delay, there is no pre-ringing

This was my thought too, but I see a lot of DSP experts aiming for minimum phase and I'm trying to understand the thinking on that better.

 

[ppataki]

So you're not necessarily linearizing the total phase response, you're just using a linear phase filter to change the roll off rate without affecting the phase?

The phase is impacted too, see this example here:

In case of the LP crossover filter you can see that it became much more linear vs when using min-phase crossover filters

Similarly, GD also became lower:

There is some pre-echo though but totally inaudible:

 

[Tim Link]

The phase is impacted too, see this example here:

What's the phase look like without a filter? Are you actually flattening the driver's native phase response by correcting it, or just not adding more phase shift to it by using a linear crossover?

 

[ppataki]

What's the phase look like without a filter? Are you actually flattening the driver's native phase response by correcting it, or just not adding more phase shift to it by using a linear crossover?

It is the latter
If I used MP crossover filters the phase would get shifted way more vs using LP crossover filters
btw. the same happens if you just apply a simple HPF for example on a subwoofer (the above graphs will show the same kind of differences)

 

[gnarly]

I've studied, set subs and speakers up, measured, then listened .. ....to this issue of system high-pass ...probably more than any other single FIR phenom.
A few quick observations about system high pass....
1. Measurements can look a hell of a lot worse than they sound....both group delay for min phase system high pass, and downward diving step for lin phase system high pass.
2. it can be damn hard to hear differences, even outdoors.
3. Pre-ring, pre-echo, pre-whatever, is hard to hear if FIR isn't bungled.....even when including a mild lin phase system hp.


4. Like folks say, complementary acoustic linear phase xovers don't cause pre-ring.
5. Providing good acoustic coupling. Use a lin phase xover between sub and main only if the two are colocated. Otherwise, lack of complementary acoustic coupling, may cause ring out of either or both.
Same issue with mains, if drivers have wide spacing, think a little harder about using lin phase xovers. I don't know this for sure though, because I have not fooled with such designs since switching to FIR. (and such acoustic designs are compromised to begin with, imho)

Ok, anyway here's some electrical sims from a new measurement program I'm learning, that also has a FIR generator built in.
They are all 5-ways with 96 dB/oct linear phase crossovers at 100, 280, 750, and 6300Hz. (a design I'm currently running)

Top panel is the Step response of the entire summation (like a full speaker with sub)
Second panel shows magnitude and phase of each section, and mag summation. Phase is flat for all sections and sum, as is group delay.
Third panel is the impulse response of sub alone, with it's lin phase low-pass via xover.
And bottom panel is subs mag and phase.

damn perfect in my book, other than I'm probably gonna ruin my vented subs with no system high pass, once I get good and drunk.

Here's the same but with a 20 Hz 4th order linear phase hpf.
Note the big dive in step response prior to rise in top panel, but that group delay in second panel remains at zero.

Now a 20 Hz IIR Butterworth 4th order hpf.
Step cleans up, at the cost of almost 20ms group delay.

Hope this all made sense / and helped.

edit: realized mistake on IIR high-pass...had it in for sub only, not system. Corrected

 

[Tim Link]

I'm coming to the conclusion that pre-ringing is generally a non-issue issue with linear filters, so long as nothing too out of bounds is done with them. We just have use our ears and decide if we hear something we don't like. The good thing about targeting a clean step response is primarily I suspec that it makes it easy to verify that two speakers in a stereo setup are going to be very similar for better stereophonics. Some people may be bothered by a little group delay or pre-ringing. I suspect neither of them are normally an issue for me.

 

[ppataki]

pre-ringing is generally a non-issue issue with linear filters, so long as nothing too out of bounds is done with them.

Yes, fully agreed

We just have use our ears and decide if we hear something we don't like.

That's exactly the point - you can instantly hear if there is an issue
I personally use this track to hear pre-echo:

Jazzanova - Boom Clicky Boom Klack

Listen to Boom Clicky Boom Klack on TIDAL

tidal.com

(right at the start)
You can switch back and forth between MP and LP to hear if there is any pre-echo with your LP filter

 

[levimax]

I've studied, set subs and speakers up, measured, then listened .. ....to this issue of system high-pass ...probably more than any other single FIR phenom.
A few quick observations about system high pass....
1. Measurements can look a hell of a lot worse than they sound....both group delay for min phase system high pass, and downward diving step for lin phase system high pass.
2. it can be damn hard to hear differences, even outdoors.
3. Pre-ring, pre-echo, pre-whatever, is hard to hear if FIR isn't bungled.....even when including a mild lin phase system hp.


4. Like folks say, complementary acoustic linear phase xovers don't cause pre-ring.
5. Providing good acoustic coupling. Use a lin phase xover between sub and main only if the two are colocated. Otherwise, lack of complementary acoustic coupling, may cause ring out of either or both.
Same issue with mains, if drivers have wide spacing, think a little harder about using lin phase xovers. I don't know this for sure though, because I have not fooled with such designs since switching to FIR. (and such acoustic designs are compromised to begin with, imho)

Ok, anyway here's some electrical sims from a new measurement program I'm learning, that also has a FIR generator built in.
They are all 5-ways with 96 dB/oct linear phase crossovers at 100, 280, 750, and 6300Hz. (a design I'm currently running)

Top panel is the Step response of the entire summation (like a full speaker with sub)
Second panel shows magnitude and phase of each section, and mag summation. Phase is flat for all sections and sum, as is group delay.
Third panel is the impulse response of sub alone, with it's lin phase low-pass via xover.
And bottom panel is subs mag and phase.

damn perfect in my book, other than I'm probably gonna ruin my vented subs with no system high pass, once I get good and drunk.


View attachment 365477


Here's the same but with a 20 Hz 4th order linear phase hpf.
Note the big dive in step response prior to rise in top panel, but that group delay in second panel remains at zero.

View attachment 365478

Now a 20 Hz IIR Butterworth 4th order hpf.
Step cleans up, at the cost of almost 20ms group delay.
View attachment 365558

Hope this all made sense / and helped.

edit: realized mistake on IIR high-pass...had it in for sub only, not system. Corrected

Thank you for the informative post. One thing I am struggling to wrap my head around is what is the difference between a "Linear Phase Crossover" and a "IIR Crossover Linearized with a FIR allpass filter"?

 

[gnarly]

Thank you for the informative post. One thing I am struggling to wrap my head around is what is the difference between a "Linear Phase Crossover" and a "IIR Crossover Linearized with a FIR allpass filter"?

You bet, glad it made sense.

In 1D space, pure electrical space, there shouldn't be any difference between a linear-phase xover, and the same IIR xover phase-linearized via FIR inverse allpass.

If someone builds a FIR file that does nothing more than linearize the IIR xover phase rotation, and use it globally across an existing two-way IIR crossed speaker, I'd think it should match a linear-phase xover.
Not 100% sure though, as I haven't bothered to try this exact experiment. I've found extrapolation from what works in 1D space, to 3D acoustical space, can have unexpected considerations.

Once we move to three ways and above, I think phase-only linearization of mutiple IIR xovers gets tricky fast..

First issue is, most folks use a combination of high-pass, low pass, PEQs or shelves, and perhaps some delay to get smooth acoustic summations thru the xover ranges.
Exactly which of those components beyond hpf and lpf, go into the pool for phase linearization??? I'm not sure.. Some are unique to one side of the xover, like EQs or delays.

And even if we had super drivers with symmetrical acoustic roll-offs and say we can make 4-way where we can simply use fully complementary LR xovers.
The xovers need to be cascaded like Linkwitz shows. https://www.linkwitzlab.com/frontiers_5.htm#V
A correct cascaded implementation has all-pass on it's own How to globally phase linearize all that?
And we are still in 1D space !

 

[levimax]

You bet, glad it made sense.

In 1D space, pure electrical space, there shouldn't be any difference between a linear-phase xover, and the same IIR xover phase-linearized via FIR inverse allpass.

If someone builds a FIR file that does nothing more than linearize the IIR xover phase rotation, and use it globally across an existing two-way IIR crossed speaker, I'd think it should match a linear-phase xover.
Not 100% sure though, as I haven't bothered to try this exact experiment. I've found extrapolation from what works in 1D space, to 3D acoustical space, can have unexpected considerations.

Once we move to three ways and above, I think phase-only linearization of mutiple IIR xovers gets tricky fast..

First issue is, most folks use a combination of high-pass, low pass, PEQs or shelves, and perhaps some delay to get smooth acoustic summations thru the xover ranges.
Exactly which of those components beyond hpf and lpf, go into the pool for phase linearization??? I'm not sure.. Some are unique to one side of the xover, like EQs or delays.

And even if we had super drivers with symmetrical acoustic roll-offs and say we can make 4-way where we can simply use fully complementary LR xovers.
The xovers need to be cascaded like Linkwitz shows. https://www.linkwitzlab.com/frontiers_5.htm#V
A correct cascaded implementation has all-pass on it's own How to globally phase linearize all that?
And we are still in 1D space !

Thank you, that makes sense and you bring up some interesting questions. I am playing around with Defonica for my crossovers on a 3 way plus sub active system. For every IIR crossover filter I create (most drivers have multiple filters) I have the option to linearize it with a FIR allpass so it is not a global allpass. It seems to work Ok but now I am in the process of taking outdoor acoustic measurements of each driver and once I have those I planned on going back to more accurately re-create "acoustic" LR 4 crossovers in which case I probably have to use some PEQ's which don't seem to have a "linearization" option and neither do the "delay" settings. Your point about cascading filters is interesting, I have an option to Linearize any LR filters so maybe the program is smart enough to do it right. I tend to be careful with pure phase adjustments as the benefits are subtle but I have heard some very audible issues if I mess phase corrections up enough. I guess the best way is to keep measuring as you work so you can understand the difference for every change you make.

 

[mdsimon2]

I don't see how a FIR linear phase XO is any different than a IIR minimum phase XO with FIR reverse all pass. The way I see it, work practice for both is the same.

FIR linear phase XO
1) Flatten magnitude and phase response of each driver with minimum phase filters
2) Add delays as necessary based on driver acoustic offsets
3) Apply FIR linear phase XOs

IIR minimum phase XO with FIR reverse all pass
1) Flatten magnitude and phase response of each driver with minimum phase filters
2) Add delays as necessary based on driver acoustic offsets
3) Apply IIR minimum phase XOs
4) Apply FIR reverse all pass corresponding to your minimum phase XO

In the reverse all pass case, because you flattened frequency response prior to applying a XO, your electrically applied XO will match your acoustic XO and a reverse all pass corresponding to your electrical XO is valid. It does not matter that different EQs / delays are applied on each driver as the only purpose of these is to flatten magnitude / phase response prior to applying the XO.

I personally think the system high pass roll off should be minimum phase. Therefore, I apply a HPF matching the system high pass roll off to all drivers (other than the driver that is actually causing the system high pass). This is all that is needed to ensure perfect phase matching between all drivers with either linear phase XO approach. The need for cascaded crossovers completely goes away when using linear phase crossovers (whether pure linear phase XO or minimum phase XO + reverse all pass).

Michael

 

[gnarly]

I don't see how a FIR linear phase XO is any different than a IIR minimum phase XO with FIR reverse all pass. The way I see it, work practice for both is the same.

FIR linear phase XO
1) Flatten magnitude and phase response of each driver with minimum phase filters
2) Add delays as necessary based on driver acoustic offsets
3) Apply FIR linear phase XOs

IIR minimum phase XO with FIR reverse all pass
1) Flatten magnitude and phase response of each driver with minimum phase filters
2) Add delays as necessary based on driver acoustic offsets
3) Apply IIR minimum phase XOs
4) Apply FIR reverse all pass corresponding to your minimum phase XO

In the reverse all pass case, because you flattened frequency response prior to applying a XO, your electrically applied XO will match your acoustic XO and a reverse all pass corresponding to your electrical XO is valid. It does not matter that different EQs / delays are applied on each driver as the only purpose of these is to flatten magnitude / phase response prior to applying the XO.

I personally think the system high pass roll off should be minimum phase. Therefore, I apply a HPF matching the system high pass roll off to all drivers (other than the driver that is actually causing the system high pass). This is all that is needed to ensure perfect phase matching between all drivers with either linear phase XO approach. The need for cascaded crossovers completely goes away when using linear phase crossovers (whether pure linear phase XO or minimum phase XO + reverse all pass).

Michael

Hi, I think that work practice has the best chance of making both methods give the same results.
As both key on the first all important step of flattening each driver's mag and phase response with minimum phase filters.
And I agree strongly that delays should be based on acoustic offsets.

I also agree if the correct inverse all-pass filters are added for the full set of IIR crossovers, there would be no need for cascading xovers.
Just made a quick electrical 1D sim using IIR xovers with inverse all-pass on each.
Uses the same xover freqs as my previous sim...100, 280, 750, & 3200Hz. But xovers are LR 24 dB/oct rather than previous 96 dB/oct.
So the sub driver (red) with a 100Hz lp gets a 100Hz inv all pass.
The low driver (dark green) gets 100Hz hp w/ 100Hz inv all pass, and 280hz lp w/280 Inv all-pass.
Ditto process for remaining drivers...each hp or low pass gets inv allpass aplied. (all 2nd order Q=0.707)

Here's the individual sections mag and phase.
And summed mag and phase...the black line across the top.
Each driver essentially has a linear phase xover for itself, just as if linear-phase xovers were being used directly.
Perfect as expected.

Now, all the xover hpf's and lpf's are turned off....and just the phase of inv allpass and their summation is left to see.
Nothing but inv allpass, so no magnitude involved.
Each of the sections inv all pass can be seen, along with the wavy black line which is their electrical summation....
The black line is what the FIR file doing the global IIR xover phase linearization will need to replicate.

Ok, so what's the point of all this.....
Here's my take...this example is as easy as it gets...simple LR24's that are easily, perfectly matched by a 2nd order inv allpass.

I've been playing around trying to use the technique on higher order xovers, and can't so far.
I can't get the inv allpass's to match up and flatten phase.

Michael, is that something you can do? With Camillia maybe? Directly put several higher order inv all-pass into one FIR file?

I don't fool with IIR much other than system high-pass as discussed.
And really because, it seems kinda like beating my head against the wall, to use IIR xovers and inv allpass, than simply using complementary linear phase xovers to begin with....
All to get the FIR into just one file ???
If I am already multiway active, I already have the channel processing and amps in place. I'm using FIR. Why make it more complicated by not using multi-channel FIR?
I don't get it Lol

 

[dfuller]

Pre-ringing is most audible in low frequencies. That's why most companies don't bother with correcting for GD down there - which isn't particularly audible unless pathological.

 

[gnarly]

Pre-ringing is most audible in low frequencies. That's why most companies don't bother with correcting for GD down there - which isn't particularly audible unless pathological.

I dunno.
I don't try much to ascertain why most companies do what they do. You may be right.
My guess is that it is more about latency than anything else, certainly so in the prosound world.
In the home audio world, my guess is it's more about audio-phoolia and anti-dsp for classic stereo, followed by latency for HT.
Latency is real, and the only cost to lin phase in general, imo. Bottom end is the discussion however, and different topic.

Imo, the ONLY way to effectively eliminate group delay is to have displacement capability to f-3 that requires very low order rolloff. Seldom, if ever encountered.
I will say this though, the closer I can get to that scenario, the less I believe group delay is inaudible.

 

[levimax]

Imo, the ONLY way to effectively eliminate group delay is to have displacement capability to f-3 that requires very low order rolloff. Seldom, if ever encountered.
I will say this though, the closer I can get to that scenario, the less I believe group delay is inaudible.

So are you saying Thiel and Vendersteen and their first order crossover are onto something?