The No.1 Website for Pro Audio
Lesson learned using Wavelets & Group Delay for sub alignment (REW)
Old 14th December 2018
  #1
Gear Addict
 

Lesson learned using Wavelets & Group Delay for sub alignment (REW)

A quick thread to share some things I've learned whilst using the Group Delay (GD) and Wavelet tools in REW for aligning a subwoofer with main speakers.

Here's a shot of what my speakers look like with the subwoofer engaged:

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_gd_lp.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_wavelet_lp.jpg

Easy to notice the big spike in the Excess Group Delay plot and the 'dislocated' appearance of the Wavelet with a null at 100Hz, which also happens to be the crossover frequency between my sub and mains. I've seen these exact same patterns in the measurements of other rooms... and in most cases it's assumed that the subwoofer and mains aren't suitably time-aligned (hence the 'step change' in delay above/below 100Hz).

I thought exactly the same myself. So I borrowed a delay line unit and set about trying to correct what appeared to be a ~10ms delay between sub/mains. But the results were a disaster! The speakers behaved as though no delay was needed. So what's happening?

To investigate, I took some measurements up close to remove the effects of the room and see the direct speaker response more clearly. Here's how it looks:

Without sub engaged:
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left_gd_close.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left_wavelet_close.jpg

With sub engaged:
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_gd_close.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_wavelet_close.jpg

So...?

The spike in Group Delay is gone (obviously caused by reflections) and instead the GD just rises steadily in the low frequencies, which I assume is quite normal - ? Difference between with/without sub at 100Hz is around 1ms; so negligible and no real problem with time-alignment.

Aside from the big 'splurge' of the sub around 30Hz, the Wavelets look almost identical. Again no significant delay at the crossover frequency of 100Hz.

Concluding thoughts:
  • Be wary of using Wavelets and Group Delay to time-align speakers when you have nulls interfering with the REW measurements, as the results can be misleading.
  • The 'hole' in the Wavelet / spike in Group Delay at 100Hz features in many measurements posted on Gearslutz. My guess is that it's possibly caused by front wall SBIR. Why? Because around 100hz would suggest a distance between speaker drivers and wall of around 40cm, which seems to typify most nearfield monitors when pushed up against the wall: e.g. in my case 30cm cabinet depth, plus taking account of them being angled, plus space needed for cables at the rear = ~40cm. EDIT: see correction from ReDRuMx below.

Probably not the most enlightening thread on this forum, but perhaps useful for anyone who sees the same patterns in their REW results when running a sub and mains.

PS: big thanks to Akebrake for getting me interested in Wavelets
Attached Thumbnails
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left_gd_close.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left_wavelet_close.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_gd_close.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_gd_lp.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_wavelet_close.jpg  

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-left-sub_wavelet_lp.jpg  
Old 14th December 2018
  #2
Gear Addict
 
ReDRuMx's Avatar
Quote:
Originally Posted by Skol303 View Post
[*]The 'hole' in the Wavelet / spike in Group Delay at 100Hz features in many measurements posted on Gearslutz. My guess is that it's possibly caused by front wall SBIR. Why? Because around 100hz would suggest a distance between speaker drivers and wall of around 40cm, which seems to typify most nearfield monitors when pushed up against the wall: e.g. in my case 30cm cabinet depth, plus taking account of them being angled, plus space needed for cables at the rear = ~40cm.
Nice analysis

But -> 40 cm distance driver to front wall would result in a > 200 Hz SBIR null (total extra flight path is 80 cm, which is 1/2 lambda for approx. 215 Hz)
Old 14th December 2018
  #3
Gear Addict
 

Quote:
Originally Posted by ReDRuMx View Post
Nice analysis

But -> 40 cm distance driver to front wall would result in a > 200 Hz SBIR null (total extra flight path is 80 cm, which is 1/2 lambda for approx. 215 Hz)
Ah! Of course… I forgot about the 'return journey'. Thanks for correcting

Thinking about it, wouldn't a distance of 40cm from wall to driver cause cancellations at around 200hz (quarter wavelength) and 100Hz (half wavelength)? I have dips at both of these frequencies... but I'm guessing here.
Old 14th December 2018
  #4
Gear Addict
 
ReDRuMx's Avatar
Quote:
Originally Posted by Skol303 View Post
Ah! Of course… I forgot about the 'return journey'. Thanks for correcting

Thinking about it, wouldn't a distance of 40cm from wall to driver cause cancellations at around 200hz (quarter wavelength) and 100Hz (half wavelength)? I have dips at both of these frequencies... but I'm guessing here.
40 cm from wall causes a 1/2 wavelength cancellation at 215 Hz, and at all odd multiples of the value (x3, x5...).

It doesn't work backwards, i.e., you wouldn't get cancellation at 215 Hz / 2.
Old 14th December 2018
  #5
Gear Addict
 

Quote:
Originally Posted by ReDRuMx View Post
It doesn't work backwards, i.e., you wouldn't get cancellation at 215 Hz / 2.
Got it. Thanks again for clarifying
Old 14th December 2018
  #6
Lives for gear
 

fancy sharing a pic of your front wall, including speakers and especially the sub? still not on the floor, still firing downwards?

good response starts with proper physical setup/positioning, appropriate crossover settings and eq alignment - only then comes phase alignment.

also worth remembering: one cannot really time align a sub, one can only phase align a sub, for one frequency and for one distance (as you noticed when playing with the delay time). what makes things worse is that the distance is often given (listening distance) as well as x-overs (built into speakers) so you simply cannot get perfect alignment with active monitors...

also, the sub in small room is actually getting used to smoothen the room response as well as to extend lf response - two related yet slightly different topics...
Old 14th December 2018
  #7
Gear Addict
 

Quote:
Originally Posted by deedeeyeah View Post
fancy sharing a pic of your front wall, including speakers and especially the sub?
Hi D! Some photos below along with Sketchup graphics which perhaps better illustrate my setup. The sub is positioned just off centre and up against the front wall (within ~10mm) and directly beneath the mains.

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-img_7918.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-img_7919.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-listening-position.jpg
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-plan-view.jpg

NB: the speakers are angled to a point about 14" behind where I sit, rather than directly at my listening position as they appear in the graphic above. And off topic, but the foam slabs really help with desk reflections... something I discovered by accident!

Quote:
Originally Posted by deedeeyeah View Post
Still not on the floor, still firing downwards?... good response starts with proper physical setup/positioning, appropriate crossover settings and eq alignment - only then comes phase alignment.
The sub is a sealed cabinet and designed to be down-firing. I’m cautious about placing it on its side for fear of causing excursions that it might not be designed for, but maybe I'm being too careful.

To be honest the sub seems well situated where it is: in fact I followed your advice shared here and on email in setting it up So that in its current location I’m getting maximum output (freq. response amplitude) when sub and mains are measured together in REW. And as you can see the Wavelets I posted above (with sub on/off) show that the alignment appears good at the crossover frequency of 100Hz. Here's a shot of the sub on its own, measured from the floor beneath the desk, for comparison:

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-sub.jpg

Quote:
Originally Posted by deedeeyeah View Post
also worth remembering: one cannot really time align a sub, one can only phase align a sub, for one frequency and for one distance (as you noticed when playing with the delay time). what makes things worse is that the distance is often given (listening distance) as well as x-overs (built into speakers) so you simply cannot get perfect alignment with active monitors...
Absolutely and a good point. I totally understand how delay times vary at low frequencies, making any attempts at 'uniform' time-alignment futile, other than at one specific frequency as you mention. I'm working on the assumption that it's best to aim for alignment at the crossover frequency - ? - which is what I've done here, within about 1ms between sub/mains at 100Hz.

As you say not perfect, but I think I'll struggle to get it much better in my small room

PS: thanks again for your help on this topic by email. Really appreciated!
Attached Thumbnails
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-img_7918.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-img_7919.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-listening-position.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-plan-view.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-sub.jpg  

Old 15th December 2018
  #8
Lives for gear
 

Quote:
Originally Posted by Skol303 View Post
...I'm working on the assumption that it's best to aim for alignment at the crossover frequency?...

...As you say not perfect, but I think I'll struggle to get it much better in my small room...
hey
thx for posting some pics! you did quite some work in there... a couple of things which strike me, so here's my view on some functional aspects:

- the spacing between your mains seem to be narrow: i can see there is not much space left 'cause of the two large bass traps, but do you really like the image you get?

- same topic, but opposite finding: your small speakers are positioned far wider; i imagine the image you get between your different speakers then is radically different? - on purpose?

i'm using a small broadband speaker myself, but just as a single mono center speaker, something i much prefer for listening in mono rather than relying on the phantom center between any two speakers...

- i wouldn't hesitate moving the table closer to the frontwall and have just small stands for the mains: this would move your listening position in a spot which i imagine could sound a bit better.

- same for your sub: cannot see how putting the sub on the side should affect it in a bad way. anyway, having the cone close to two boundaries would get you still a bit more output...


...which brings me to propose a radically different design: take away your bass absorbers from the front wall, move them to the rear wall, keep treatment of first reflections from mains, user wider spacing for your mains, put your sub in a front corner and load it! let mains run full range, try two settings for hi cut of the sub - of course all hell will break loose! keep cool and use dsp for eq alignment of mains only at your listening position, don't care for any other position!

roughly along these lines and described much better than i can (but practiced for dozens of years with great success):

The Elephant In The Control Room |

then measure and report back a) which 'sound' you like better and b) which setup measures better (meaning you need less tweaking to get reasonable response): must not necessarily be the same thing!




p.s. no cloud? could save you from damping to the sides of the desk. and there is not much on the rear wall (due to a door)?!
more related to your first post: besides delay, does your dsp have allpass filters?
and to answer one of your questions: when using symmetrical crossovers, alignment mostly aims at crossover frequency - now what do you do with asymmetrical crossovers/huge overlap? - here come the allpass filters into play... :-)

Last edited by deedeeyeah; 15th December 2018 at 01:16 AM.. Reason: p.s. added and edited
Old 15th December 2018
  #9
Gear Addict
 

Thanks for the input D!

Quote:
Originally Posted by deedeeyeah View Post
- the spacing between your mains seem to be narrow: i can see there is not much space left 'cause of the two large bass traps, but do you really like the image you get?
Difficult question. I’ve got so accustomed to the speakers being set up how they are that it’s hard to know whether I “really like it” or not… I guess I must do! It’s a narrow spacing for sure, but the soundstage doesn’t sound too small as a result. At least not to my ears. I think it might even help me to put a little more width into my mixes. I mixed for years on headphones and my material around that time sounds quite ‘mono’ by comparison.

Quote:
Originally Posted by deedeeyeah View Post
same topic, but opposite finding: your small speakers are positioned far wider; i imagine the image you get between your different speakers then is radically different? - on purpose?
Yeah, the MixCubes are positioned wider to give the opposite effect - a very wide stage. By switching between the two sets of monitors I get a more balanced picture of what’s happening.

I have tried moving the mains to the rear edges of the desk, forming the classic ‘equilateral triangle’, but this causes a big dip in the frequency response due to the speakers being further away from the front wall (SBIR null).

Quote:
Originally Posted by deedeeyeah View Post
i wouldn't hesitate moving the table closer to the frontwall and have just small stands for the mains: this would move your listening position in a spot which i imagine could sound a bit better.
Unfortunately the table can’t move any further forwards - the front corner bass traps prevent this - but there’s also a room mode null in front of my listening position that I’m keen to avoid. Here’s an illustration kindly produced by Akebrake that shows the location:

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-skol-test1.jpg

At some point in future, if/when my mixer gives up and dies, I might get rid of the big desk and simplify my set up: smaller desk and a rack mounted summing mixer instead. But that’s for another day

Quote:
Originally Posted by deedeeyeah View Post
same for your sub: cannot see how putting the sub on the side should affect it in a bad way. anyway, having the cone close to two boundaries would get you still a bit more output...
Good tip. Thing is I’m not sure I need any more output from the sub. It already gives me plenty to help fill in the low end (I need to cut the lows quite a lot to get it back under control), but if the opposite were happening I’d experiment with this for sure.

Quote:
Originally Posted by deedeeyeah View Post
...which brings me to propose a radically different design: take away your bass absorbers from the front wall, move them to the rear wall, keep treatment of first reflections from mains, user wider spacing for your mains, put your sub in a front corner and load it! let mains run full range, try two settings for hi cut of the sub - of course all hell will break loose! keep cool and use dsp for eq alignment of mains only at your listening position, don't care for any other position!
There’s no more room on my back wall for treatment - not unless I want to block up the door! - it’s maxed Here’s a Sketchup for reference (the panels left and right do actually extend right down to the floor - i.e.no gap):

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-back-wall.jpg

In the past, I have tried temporarily removing my front wall treatment and positioning the main speakers wider / more into the corners of the room. But again this caused problems with frequency response and I took the decision that it wasn’t worth sacrificing freq response for any gains I was getting in terms of the sound stage. So I decided to leave the mains where they are.

Interesting idea about running the mains full range. I'll try some experimenting with that...

Quote:
Originally Posted by deedeeyeah View Post
p.s. no cloud? could save you from damping to the sides of the desk. and there is not much on the rear wall (due to a door)?!
Yeah I do have a cloud but removed it from the Sketchup pics as it would have obscured the view. My rear wall has a mix of velocity-based traps, some home-made and some from GIK (all 200mm thick or more), plus a series of pressure-based traps on the back of the door (dealing with a room mode at 70Hz). I also have a moveable bass trap on a stand that I position in front of the door when mixing… which also functions as a ‘kid-proof barricade’ to keep my young boys out of the room when I’m busy. Good parenting tip right there

Quote:
Originally Posted by deedeeyeah View Post
more related to your first post: besides delay, does your dsp have allpass filters?
I must admit I know very little about filters… I must do some reading… but I do know that my DSP (Dirac Live) uses mixed-phase filtering - a combination of IIR and FIR filters.

In summary what I’m trying to say is… I think my room is finished! Which is perhaps a weird thing to say in an acoustics forum where being an obsessive perfectionist seems to come with the territory (I know I'm one!). There’s always something more that can be done, of course. But I think I’ve reached the point at which I can’t make any worthwhile improvements - well, not without tearing everything down and re-building the room brick-by-brick. And I don’t think I need to, because the acoustics at my listening position are pretty good for a small room - some measurements here.

The final tweak I was seeking to make concerned better aligning my sub and mains. The purpose of this thread has been explaining why - in my situation - that’s not necessary. And specifically how I’d misinterpreted the REW results to believe there was a delay problem, when in fact it was an artefact of a null in the low end. I've seen this same problem in other threads on here - measurements being skewed by nulls and misinterpreted as sub/mains misalignment. So hopefully this might help others to tread more carefully around the topic.

TL/DR the lesson learned for me has been: measurements don’t always show the true picture of what’s happening. They can lie (that is if you don't understand them well enough, like me...).

Thanks again for your input D and also to ReDRuMx. Really appreciate it guys
Attached Thumbnails
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-skol-test1.jpg   Lesson learned using Wavelets & Group Delay for sub alignment (REW)-back-wall.jpg  
Old 15th December 2018
  #10
Lives for gear
 

nothing wrong with having a room that's finished, sounds/measures/feels and works well - congratulations!

...but don't call a desk such as yours ever 'big' again :-)
Old 15th December 2018
  #11
Gear Addict
 

Quote:
Originally Posted by deedeeyeah View Post
...but don't call a desk such as yours ever 'big' again :-)
Haha! It must be the photo... bad lighting... here it is from another angle

Lesson learned using Wavelets & Group Delay for sub alignment (REW)-9544d64a-6475-4580-b660-1d905bb5dceb.jpeg
Attached Thumbnails
Lesson learned using Wavelets & Group Delay for sub alignment (REW)-9544d64a-6475-4580-b660-1d905bb5dceb.jpeg  
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump