REW - Time Align 3-Way Active Speaker Step or Impulse? - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 7Likes
  • 2 Post By 3ll3d00d
  • 2 Post By 3ll3d00d
  • 1 Post By 3ll3d00d
  • 1 Post By 3ll3d00d
  • 1 Post By 3ll3d00d
 
Thread Tools
post #1 of 19 Old 10-13-2019, 05:27 PM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
REW - Time Align 3-Way Active Speaker Step or Impulse?

Hi All,

Just having a bit of trouble understanding time alignment with Active speakers using REW.

I have read much information and even a bunch of threads on the overall matter. It seems the general wisdom is to align the impulse. But with a 3way active speaker, the impulse is obviously very, very different, the tweeter, sharp, the midrange less so, and the LF is just a long wave.

So I have a large modular 3-way speaker, in which I placed mic at listening position for the time alignment.

Now in my messing around, it seems that the REW estimate for the delay is not quite on point and totally reliable to use, its close, but if you follow the estimate and align only based on that, the impulse and step wont match properly.

So, something I tried to do, keep the same crossover slopes in the final design, but change the LF and Mid speakers to fall off at 6k each so they produce a similar/better Impulse and step, and then align those... Is that the right thing to do?

Also, if I do that, I can never get impulse and step both perfectly lined up at the same time even when talking about a 6k rolloff for the task.

Yesterday in my latest efforts I settled on aligning the step response so the peaks all lined up in the initial step, the impulse was relatively close but not quite matching.

Which one should take priority??

Step



Impulse



Project file.

https://www.dropbox.com/s/jei860h1qw...ment.mdat?dl=0

p.s. the mid and lf woofer impulses in the project are inverted due to inverse measurements, its the tweeter that will be inverted in the speaker, not the mid and LF... I just did that so the step pointed up...

Any feedback, help welcome.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
Sponsored Links
Advertisement
 
post #2 of 19 Old 10-13-2019, 11:54 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Broadly speaking you would align the rise of the mid and woofer impulse with the tweeter. The combined step response is then the best way to look at how coherent the alignment is though the expected shape depends on the crossovers used/resulting phase alignment (linear phase will look different to minimum phase).

In your case the mid and woofer are early (as you are aligning the peaks).
3ll3d00d is online now  
post #3 of 19 Old 10-14-2019, 03:19 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by 3ll3d00d View Post
Broadly speaking you would align the rise of the mid and woofer impulse with the tweeter. The combined step response is then the best way to look at how coherent the alignment is though the expected shape depends on the crossovers used/resulting phase alignment (linear phase will look different to minimum phase).

In your case the mid and woofer are early (as you are aligning the peaks).

Ok thanks so I was going about it the wrong way.

This is what happens when you align the drivers to the exact delays that REW quotes when using an acoustic timing reference on the other speaker, so the delays are all relative to that and matched. This is also the normal XO points, so the impluses look way different. 200/1200 is the XO points...



Looks like the woofers are late yeah? Should I pull them backwards to try and match the rises, the uphill parts slightly? should be very easy to get the mid aligned since its not too dissimilar, but the LF is going to be more tough.



Is there any merit to having each speaker play full range when doing sweeps to get a good impulse to match and then aligning the rise then?

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
Sponsored Links
Advertisement
 
post #4 of 19 Old 10-14-2019, 03:34 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
In this one I made the XO for all speakers 6k roll off to get a similar impulse and I aligned the rise on all speakers.

Is this looking better? Or should I not be touching the XO points to align the rises?



System Step looks like this then with all drivers at once and proper XO restored



Impulse.


JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #5 of 19 Old 10-14-2019, 03:49 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Is that step with a crossover in place or the sum of the raw drivers?

Another option is to compare to the crossover itself with some delay added per way. For example you space each way apart by 40ms and load each filter into rew as a wav then apply the same delays to each way, with the filter in place, and measure. Now overlay the filter against actual and adjust alignment til it matches. Finally remove the excess 40/80ms delay and measure the sum. The point being to eliminate the difficulty involved in aligning peaks of different shapes.
Javs and Skinfax1 like this.
3ll3d00d is online now  
post #6 of 19 Old 10-14-2019, 03:50 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
It is also a good exercise to sum the 3 filters manually and then examine the step response. This gives you an idea of what the theoretical ideal response is for your chosen crossover shapes. Remember to low and high pass the tails based on the driver characteristics for best results.
Javs and Skinfax1 like this.
3ll3d00d is online now  
post #7 of 19 Old 10-14-2019, 03:55 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by 3ll3d00d View Post
Is that step with a crossover in place or the sum of the raw drivers?

Another option is to compare to the crossover itself with some delay added per way. For example you space each way apart by 40ms and load each filter into rew as a wav then apply the same delays to each way, with the filter in place, and measure. Now overlay the filter against actual and adjust alignment til it matches. Finally remove the excess 40/80ms delay and measure the sum. The point being to eliminate the difficulty involved in aligning peaks of different shapes.
That's the final xo crossover in play with all speakers after I aligned the rise on the step. It's a 3 way, does it not look good?

Sent from my SM-N950F using Tapatalk

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #8 of 19 Old 10-14-2019, 03:58 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
I feel like the 2nd part to your posts got quite complicated I am not sure I am following that.

Sent from my SM-N950F using Tapatalk

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #9 of 19 Old 10-14-2019, 04:15 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Quote:
Originally Posted by Javs View Post
I feel like the 2nd part to your posts got quite complicated I am not sure I am following that.
Do you mean the delayed crossover?
3ll3d00d is online now  
post #10 of 19 Old 10-14-2019, 05:04 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Quote:
Originally Posted by Javs View Post
That's the final xo crossover in play with all speakers after I aligned the rise on the step. It's a 3 way, does it not look good?
There are multiple quite distinct steps which suggests the alignment is not right, you should see a smoother handoff from one to the next if they were aligned.

Fig 12 in https://www.stereophile.com/content/...art-two-page-3 is an example but use your actual filters for a real world comparison

What filter and orders are you actually using?
3ll3d00d is online now  
post #11 of 19 Old 10-14-2019, 05:12 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by 3ll3d00d View Post
There are multiple quite distinct steps which suggests the alignment is not right, you should see a smoother handoff from one to the next if they were aligned.

Fig 12 in https://www.stereophile.com/content/...art-two-page-3 is an example but use your actual filters for a real world comparison

What filter and orders are you actually using?
Yeah it's never going to look like your example. I've been playing with this for a long time and tried just about every combination of delay possible to try get the step to look like that, even flipping polarity, it just won't do it without seeming to add a lot of what looks like pre ringing in the step. In actual fact, the step and impulse I have now is arguably the best it's looked in a long time.

The crossover slopes are steep. All Butterworth

Lf 200hz 24db
Mid 200/24 and 1200/48
Tweeter 1200/48

If I change everything to 6db first order I can get the step to look much better, but that doesnt sound good.

Hence all the head scratching. When I google 3 way crossover step I do see even some stuff from linkwitz which seems to look a lot like my step response. Your example looks like a 2way to me based on my research.

I know linear phase can have a perfect step response thoygg. Even with 3 way.




Sent from my SM-N950F using Tapatalk

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves

Last edited by Javs; 10-14-2019 at 05:29 AM.
Javs is offline  
post #12 of 19 Old 10-14-2019, 05:21 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Here it is scroll down to where he talks about group delay. Those step responses look a lot more like mine than the one you showed me.

https://www.linkwitzlab.com/frontiers.htm

Sent from my SM-N950F using Tapatalk

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #13 of 19 Old 10-14-2019, 05:29 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Yes a perfect step is a straight vertical line up, a perfect loudspeaker (i.e. a bandpass device) is a basically a triangle. High order minimum phase crossovers will not look like this due to the significant phase rotation around the crossover. I can post an example later.
Javs likes this.
3ll3d00d is online now  
post #14 of 19 Old 10-14-2019, 05:35 AM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by 3ll3d00d View Post
Yes a perfect step is a straight vertical line up, a perfect loudspeaker (i.e. a bandpass device) is a basically a triangle. High order minimum phase crossovers will not look like this due to the significant phase rotation around the crossover. I can post an example later.
Thank you

Sent from my SM-N950F using Tapatalk

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #15 of 19 Old 10-14-2019, 01:29 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
to simulate using freely available tools, you can

- download RePhase
- set taps to 65536, optimisation to moderate, rate to 48000, format to 32bit LPCM mono
- for each way
- add a minimum phase filter as per your spec
- name after the way (W, M, T)
- click generate

- open REW
- for each wav
- import as an impulse
- go to IR tab & remove delays

This should leave you with a minimum phase set of filters, sum them using trace arithmetic to confirm - W+M, T+M then (T+M) + W - you should see the characteristic hump at each crossover point

Go to the overlays screen and show the steps, zoom into to about -2ms to 10ms and show all 3 individual traces. It should look like



show the sum instead and it should look like



you can do the same with the impulse if you like

this is the perfect response for your chosen filters
Attached Thumbnails
Click image for larger version

Name:	wmt_sr.png
Views:	43
Size:	41.6 KB
ID:	2627758   Click image for larger version

Name:	sum_sr.png
Views:	50
Size:	24.8 KB
ID:	2627760  
Javs likes this.
3ll3d00d is online now  
post #16 of 19 Old 10-14-2019, 01:38 PM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
for the "comparing like to like" view on time alignment

- go to the IR tab
- select the W
- set an offset of -20ms & click apply
- select the M
- set an offset of -10ms & click apply

go back to the overlay
select the IR tab
expand the x range to -2ms to +40ms

it should look like



Now go to your DSP and add those delays (or some other delays depending on what delay capability you have) to each way and measure

The alignment should be against the tweeter so that will be naturally aligned at 0, zoom in on M and then W and compare against the location of the peak in the M or W filter respectively. The room (or wherever you are measuring) may make matching the W tricky (e.g. if in a room you will most likely get a better view by squelching any modal peaks) but the M should be relatively easy to match up.

Make sense?
Attached Thumbnails
Click image for larger version

Name:	delayed_ir.png
Views:	46
Size:	31.9 KB
ID:	2627764  
Javs likes this.
3ll3d00d is online now  
post #17 of 19 Old 10-14-2019, 04:16 PM
AVS Forum Special Member
 
jamin's Avatar
 
Join Date: May 2001
Location: moot point
Posts: 1,167
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 61 Post(s)
Liked: 77
Perhaps only useful as a late night soporific, here is an additional take on the matter.

https://community.klipsch.com/index....tween-drivers/

plug in to play
Acoustic Mafia - Hear No Evil
jamin is offline  
post #18 of 19 Old 10-15-2019, 05:29 PM - Thread Starter
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 498 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6642
Quote:
Originally Posted by 3ll3d00d View Post
to simulate using freely available tools, you can

- download RePhase
- set taps to 65536, optimisation to moderate, rate to 48000, format to 32bit LPCM mono
- for each way
- add a minimum phase filter as per your spec
- name after the way (W, M, T)
- click generate

- open REW
- for each wav
- import as an impulse
- go to IR tab & remove delays

This should leave you with a minimum phase set of filters, sum them using trace arithmetic to confirm - W+M, T+M then (T+M) + W - you should see the characteristic hump at each crossover point

Go to the overlays screen and show the steps, zoom into to about -2ms to 10ms and show all 3 individual traces. It should look like

show the sum instead and it should look like

you can do the same with the impulse if you like

this is the perfect response for your chosen filters
Thanks, this post is really awesome!

My step is not far off that actually, using the REW delays its actually pretty close I think I might leave it there (which is where I had it).

One thing thats interesting though is my woofer impulse travels down a little bit before going up, maybe thats because its ported? Would a sealed woofer have a clean rise? Or perhaps its the 48db XO at 25hz which I put in place to protect the woofers. Or maybe also because I have two LF Woofers though they are not wired out of polarity.

Thats probably the only thing which visually looks different to the ideal step you posted.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #19 of 19 Old 10-16-2019, 01:19 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 4,233
Mentioned: 273 Post(s)
Tagged: 0 Thread(s)
Quoted: 2548 Post(s)
Liked: 2835
Quote:
Originally Posted by Javs View Post
One thing thats interesting though is my woofer impulse travels down a little bit before going up, maybe thats because its ported? Would a sealed woofer have a clean rise? Or perhaps its the 48db XO at 25hz which I put in place to protect the woofers. Or maybe also because I have two LF Woofers though they are not wired out of polarity.

Thats probably the only thing which visually looks different to the ideal step you posted.
the sim I showed was for a sealed sub as I put a 2nd order rolloff on it at 20Hz. The faster rolloff of ported and the 8th order at 25Hz will certainly make a difference though I would think that would manifest in the tail of the step. Discontinuities early on are pretty hard to track down as early on means something v early in the signal, this could be diffraction (at higher frequencies) or perhaps SBIR at lower frequencies? The 2nd driver is certainly one candidate though.
3ll3d00d is online now  
Sponsored Links
Advertisement
 
Reply Audio Theory, Setup, and Chat

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off