3-way active correction with Summation Issue

[jjazdk]

Hi All,

I have made a 3-way correction with minimum delay for streaming purposes, but... Measuring the summation of the Front Left & Front Right speakers, there is obviously something wrong with the summation in the 100-250Hz area.

So what is going on here..?! How can the summation get that bad..?

My measurements are made with REW, through JRiver with ASIO Line-In (I am using an extra sound card for REW to feed analog signals to the Line-In).
JRiver outputs 7.1 channels to the AL Convolver (8 input channels, 12 output channels).
3-way is setup as:
Tweeter >250Hz
Midbass 100-250Hz
Subwoofer <100Hz

Front Left, Front Right and the summation of the two.

Only the Midbass still with correction through the convolver (I turned off the other power amplifiers).
The summation is almost non-existing.

The raw response of the Midbass drivers, with no correction/filter.
Here the summation is definitely a lot better.

AL Speaker setup.

AL Channel Routing.

AL XO Configuration.

 

[juicehifi]

You need good acoustic symmetry to obtain good left +right summation without using TTD correction.

 

[hulkss]

I had to use minimum phase crossovers that are three octaves wide to get good results without TTD. Check the step responses. If they are aligned in time like this you may get good results. Small changes in crossover frequency can make a big difference.

 

[jjazdk]

You need good acoustic symmetry to obtain good left +right summation without using TTD correction.

You can see the non-corrected summation on my last plot, it sums pretty well over the entire frequency range and also in the 100-250Hz area.

After correction the summation is a lot worse.

 

[jjazdk]

I had to use minimum phase crossovers that are three octaves wide to get good results without TTD. Check the step responses. If they are aligned in time like this you may get good results. Small changes in crossover frequency can make a big difference.

The thing is that only the midbass drivers exhibit poor left+right summation after the correction. If anything I would expect poor summation in the tweeter/midrange, but they actually sum fairly well in most of the range.

 

[hulkss]

Your "raw" response shows big dips right where you are putting the crossovers. That is probably causing the problem.
Dips like that usually have associated phase shifts.

 

[jjazdk]

Your "raw" response shows big dips right where you are putting the crossovers. That is probably causing the problem.
Dips like that usually have associated phase shifts.

Hmm... Good input. I can try and move the crossover points, though I do not have much wiggle room there.

I still find it odd that the uncorrected response has better summation, than the corrected, also at the dips.

It is time to experiment, with some stuff..

 

[hulkss]

I believe the software works to match phase at the XO points so the acoustic sum of the two drivers at the XO point is +6 dB.

With no TTD correction there can be issues at other frequencies.

 

[juicehifi]

Pretty impressive results there, hulkss, considering it is made for minimum delay.

 

[mollie]

Pretty impressive results there, hulkss, considering it is made for minimum delay.

Hi,
It looks like that one of the two mid-bas drivers is out of fase by 180 degrees!!!

 

[jjazdk]

Hi,
It looks like that one of the two mid-bas drivers is out of fase by 180 degrees!!!

Yes, more or less. Which is odd, since they are in phase when I measure their summation without the correction filter.

 

[mollie]

Hi again,
In that case, than probably both mid-bass drivers are out of fase!!!
Check out and correct, if neccesary.
Succes
Mollie

 

[jjazdk]

Hi again,
In that case, than probably both mid-bass drivers are out of fase!!!
Check out and correct, if neccesary.
Succes
Mollie

What are you talking about. They are not out of phase. And, even if they where Audiolense should flip their phase.

I appreciate you trying to help, but what you are suggesting does not make sense.

 

[juicehifi]

I believe the software works to match phase at the XO points so the acoustic sum of the two drivers at the XO point is +6 dB.

With no TTD correction there can be issues at other frequencies.

Yes, they are matched at the xo point. So if the two drivers are in opposite phase, it may get nasty outside the xo point .... which will be in opposite phase when they are matched at the xo center.

 

[jjazdk]

The drivers themselves are definitely not out of phase, when there is no correction applied.

But, perhaps the correction makes them out of phase in the passband, because different minimum phase filters are applied to each channel.

Anyway, it really does not work in my setup :-(

Yes, they are matched at the xo point. So if the two drivers are in opposite phase, it may get nasty outside the xo point .... which will be in opposite phase when they are matched at the xo center.

 

[juicehifi]

That could happen if they are somewhat out of phase before the crossovers are applied. Furthermore, drivers plus room can be out of phase on cancelling frequencies ... even if they are generally in phase.