Home Theater Forum and Systems banner

1 - 20 of 28 Posts

·
Registered
Joined
·
41 Posts
Discussion Starter #1
I've only just discovered the benefits of applying filtering to IR measurements.
Setting the filter point as close as possible to the preferred crossover point you can very quickly nudge the waveforms into alignment.
In the phase plots this filtered IR alignment at the crossover point will also result in the closest phase tracking.
Cheers
G
 

·
Premium Member
Joined
·
1,562 Posts
Interesting! I hadn't noticed that before. Well Done. :T

It does work reasonably well. It adds another way to look at and adjust the XO timing.

It seems to add an extra 360° phase rotation at the XO, but that does not impact its utility in aligning the phase tracking. We just need to keep in mind that the actual phase rotation is distorted by the filter, but that tracking between the 2 drivers is still representative.

Thanks for sharing!

If you have the time an example may be helpful to members to understand the process.
 

·
Registered
Joined
·
98 Posts
Please explain how this is done. I'm planning to tune my new DSP in my car later this week, and would like to incorporate this if possible.
 

·
Registered
Joined
·
41 Posts
Discussion Starter #5 (Edited)
Hi there. Hope some of the information below can help.
I have performed multiple sweeps from a distance of 1meter, in increments of 0.1ms.
We're looking at 12" bass and 4.5" bass midrange here crossed over at 400 hz with LR8 filters. My processor is the Behringer DXC.
The overlays below include impulse response, wrapped, unwrapped phase and SPL.

impulse response - unfiltered.jpg

As you can see, the impulse response of the bass driver exhibits a very broad wave form making it somewhat confusing as to where exactly the mid should line up.
The two phase plots give a better picture and show that I'm getting close to alignment.

wrapped phase - 400 hz lr8 - 1meter.jpg
unwrapped phase - 400hz lr8 - 1 meter.jpg

In order to render a filtered IR view go to the Filtered IR tab and apply filtering to each impulse to value of your crossover point.
Now go to your overlays view and examine how the IR waveforms have changed. As you can see from the graph the filtered Impulses now align more symmetrically.

filtered ir in overlays view.jpg


From this point adjust your time delay settings so as to achieve the closest possible alignment.

filtered ir alignment - 400hz.jpg

This alignment will result in the closest possible phase tracking at the cross over point as is demonstrated in the phase plots.

wrapped phase alignment 400 hz.jpg
unwrapped phase alignment 400 hz.jpg

Now remove the filtering on the IR and go back to the normal IR view. You will note that the mids properly align with the bass waveform at the very beginning of the impulse, not at the peak.

unfiltered ir alignment.jpg

I've attached an spl overlay spl response.jpg

Take note of the following points.

1- Any adjustment to your crossover point will require you to re-measure and realign. Changing your crossover point changes the delay on the driver. The same can also be said for slope types which add or reduce delay as adjusted.
2- Always choose a crossover point with good spl support either side of the band pass.
 

·
Premium Member
Joined
·
1,562 Posts
Thanks geopango! Nicely done. :sn:

I will just add some additional comments that may be helpful in some circumstances.

The phase tracking data can become much more difficult to read than this example if the IR offset delay is greater, or the XO freq is higher, or the mic is at the LP rather than near field. The overlay of the IRs will still be reasonably similar to read however so the phase chart is not really required.

If we do need to read the phase more clearly in those conditions it is helpful to:
> Manually shift the IRs back to near 0ms using the same offset value for all IRs. It's important to shift both driver IRs the same amount as this keeps the relative phase relationship of the 2 drivers.
> Use no smoothing or 1/48 smoothing for the best readability of the phase.

Below is a difficult example using my FL speaker MW/TW 1700Hz XO measured at my 13 ft LP. IRs delays are significant at roughly 94.15ms.

1 irs.png

Below, when the IRs are filtered with 1600Hz, 1/3 octave, the close IR overlay is apparent. This to enough to confirm the close phase tracking through the XO.

3 irs filtered.png

Below, the phase chart is unreadable however.

2 phase unshifed.png

Below, the IRs are now both shifted exactly 94.15ms to place them near 0ms.

4 irs shifted.png

Below, the close phase tracking can now be seen reasonably well when no smoothing is applied. That is, it's barely readable if we are experienced with reading phase charts with lots of reflections that tend to hide/smear the direct sound phase.

5 phase tracking.png

Below, the IRs are windowed as shown in order to eliminate as much of the late arriving reflections as possible while keeping enough of the early IR response as possible to still see some detail. The window settings were adjusted via trial and error. We can see the phase path is similar to that in the unwindowed phase chart above. Had the measurements been done with the mic at 1m, the chart would be very smooth.

6 phase tracking windowed.PNG

Just additional hints for those dealing with reading phase tracking charts in difficult circumstances.
 

·
Registered
Joined
·
98 Posts
This is brilliant! Thanks so much for the explanations. I look forward to delving into this on Thursday-Saturday with my new Helix DSP Pro. It has time alignment in 3.5mm steps and Phase Angle adjustment in 11.5 degree steps.

I do have a question about measuring mic location. You mention the mic at 1m yields cleaner phase plots due to reduced reflectional influence. But isn't it more important to measure at the LP to properly dial in the time alignment?


Sent from my iPhone using Tapatalk
 

·
Premium Member
Joined
·
1,562 Posts
In a car there may be no other good options. The LP is a safe choice.

For a small home "bookshelf" size speaker as I measured above, both drivers are closely stacked on the baffle in the traditional way. In that case we could have moved the mic closer while keeping it on the listening axis. At 1m or even a little closer the relative delay timing would still be accurate. We are just removing some of the excess distance/delay without changing the relative delay needed between the 2 drivers. The closer mic position reduces the magnitude of the reflections and results in a much cleaner phase chart. The closer the mic is placed though, the harder it is to be sure the mic is still on the listening axis.

Often SWs are placed away from the mains and thus the LP is the only option for the mic placement. That is why I added the hints about how to clean up phase charts when needed.

Just choose the mic position appropriately for the situation.
 

·
Registered
Joined
·
98 Posts
Take note of the following points.

1- Any adjustment to your crossover point will require you to re-measure and realign. Changing your crossover point changes the delay on the driver. The same can also be said for slope types which add or reduce delay as adjusted.
2- Always choose a crossover point with good spl support either side of the band pass.

Just to be clear, we need to pick a crossover point first and then measure as above to determine the delay?

Or should we get a ballpark delay, and then set crossover and fine tune the delay as above?



My situation is a little different because it's a car, with listening position in the driver seat. So I need some large delays on the left midbass and the subwoofer in particular. Furthest speaker is my right compression horn under the passenger side dash.
 

·
Registered
Joined
·
41 Posts
Discussion Starter #12
Just to be clear, we need to pick a crossover point first and then measure as above to determine the delay? Or should we get a ballpark delay, and then set crossover and fine tune the delay as above? My situation is a little different because it's a car, with listening position in the driver seat. So I need some large delays on the left midbass and the subwoofer in particular. Furthest speaker is my right compression horn under the passenger side dash.
The selection of the correct crossover point based on the most linear frequency response of the two drivers combined is paramount. Ensure that both drivers exhibit excellent linearity before and after the crossover point. This attention to spl will allow the best phase matching and well matched filtering after the pass band.
My bass and mid drivers are very flat between between 250 and 500 Hz. I've chosen 345hz as crossover point providing excellent linearity above and below the crossover point.
Cheers
 

·
Registered
Joined
·
98 Posts
When you say you took multiple measurements at .1ms increments, do you mean you measured... then adjusted the DSP delay for a driver by .1ms, and measured again, then repeat, etc..?
 

·
Registered
Joined
·
41 Posts
Discussion Starter #14
When you say you took multiple measurements at .1ms increments, do you mean you measured... then adjusted the DSP delay for a driver by .1ms, and measured again, then repeat, etc..?
You just keep measuring and changing the delay incrementally till the phase align at the crossover point. You can theoretically do this just once. You can make a copy of that measurement , nudge it into alignment in REW, then subtract the time difference between the two measurements. That delay amount can then be applied on your DSP. It works.
 

·
Registered
Joined
·
98 Posts
When we are trying to do this at a higher crossover point from the LP, and we get a phase chart that looks like this:





Is it OK to use the Unwrap Phase button to view it that way?
 

·
Registered
Joined
·
41 Posts
Discussion Starter #16
You really have to zoom right into the crossover point and examine how the wrapped and unwrapped phase aligns. You can then apply filtered IR for each measurement approximately at the crossover point. In overlays, view the alignment of IR and adjust delay accordingly till the wave forms sit perfectly on top of each other .
Good luck.
 

·
Registered
Joined
·
98 Posts
So I definitely understand that the crossover point must be set first because that will affect the timing.

What about EQ? If there will be a lot of channel-independent EQ applied, should the EQ be done before or after using this approach to adjusting time alignment/phase?
 

·
Premium Member
Joined
·
1,562 Posts
When we are trying to do this at a higher crossover point from the LP, and we get a phase chart that looks like this:

Is it OK to use the Unwrap Phase button to view it that way?
No, that will not normally help. When the IR is shifted very far away from 0ms there is just too much phase rotation to read the chart. Any reflections in the chart causing phase rotations also make the phase unreadable when unwrapped. The best action is follow the advice in Post 6 when the chart looks like this.
 

·
Premium Member
Joined
·
1,562 Posts
So I definitely understand that the crossover point must be set first because that will affect the timing.

What about EQ? If there will be a lot of channel-independent EQ applied, should the EQ be done before or after using this approach to adjusting time alignment/phase?
For a new setup I would prefer to:
> Set the XO freq and filters
> Set the delay via; phase tracking, or maximum SPL in the XO range, or this new filtered IR method.
> EQ to the house curve.

That said, having at least rough EQ in place to the house curve makes it easier to set the delays as the IR and phase charts are a little cleaner. I just leave the any old EQ in place if I decide to change XO settings and delays. Then revisit the EQ settings as needed. There is nothing wrong with adjusting back and forth as needed to assure that the best combination has been found.
 

·
Registered
Joined
·
98 Posts
First of all, thanks so much for all of the guidance. I am having a lot of fun learning about this subject, and have spent a fair amount of time testing things. Although I'm still not achieving good results using this method, I think my environment being a car is largely to blame. Unequal path lengths to the different speakers from the LP in the driver's seat definitely make time alignment and phasing tricky.

I do have more questions:

My subwoofer to Midbass Xover point is currently 90 Hz. I have tried using Filtered IR at both 80Hz and 100Hz. Is there a preference for which filter to use when the xover point is between 2 filters?

Same is true for my horns to midbass transition, which currently is at 900 Hz.
Do I use 800Hz, or 1000Hz filter?

I have noticed that using the filtered IR method for my time alignment has yielded some wildly different results than expected when I take rough measurements of the distance to each speaker with my measuring tape. I have also not yet achieved good results with getting the proper phasing using this method. While the phase charts look correct, my ears are certainly telling me otherwise. My results have been better setting things by ear with pink noise.
 
1 - 20 of 28 Posts
Top