Excess group delay and mismatched phase response

[pollock0424]

Hello,
I have moved my system to the longwall + 15inch Subwoofer and found out that major issues in the bass region have been solved. To get some perspective here is my old thread. In this new setup, I am trying to figure out measurements such as excess group delay (GD) and minimum phase ([1],[2]). Here are some plots for my current setup:

As you can see phase response of the left speaker is approximately constant from 600hz to 20khz as opposed to the Right speaker. What is going on ? Should I be concerned that phase responses are not matching ? My guess is that I could experience cancellations ? Also, in the left speaker 600hz to 20khz, waves seem to be having same delay, how will this influence the perceived sound ?

Also, I am trying to see the excess group delay so that I can set the delay on my sub. Here are the plots, why is the "Excess group delay" option greyed out ?

Also, you might have noticed a massive dip at 100hz, I think I can correct for that. It was probably because my work bench was put up as I have an amp on the bench, currently waiting to be fixed. I will re do the measurements again.

Any explanation of what's going on in the phase response will be greatly appreciated.
I am also looking for suggestions on setting the delay for the sub.
I feel like the treble response is not as good as before, is this my expectation bias or do you see something in the plots ?

 

[ernestcarl]

As you can see phase response of the left speaker is approximately constant from 600hz to 20khz as opposed to the Right speaker. What is going on ? Should I be concerned that phase responses are not matching ? My guess is that I could experience cancellations ? Also, in the left speaker 600hz to 20khz, waves seem to be having same delay, how will this influence the perceived sound ?

Whether or not you should be concerned depends on how you measured.

The right speaker's (blue trace) inclining phase to the right of the graph means that its upper-half of the FR is playing earlier than whatever is your set timing reference. If this is accurate, it could lead to a perceived image shift towards that side as well as cause some additional unnecessary comb filtering when both speakers are playing simultaneously.

Also, I am trying to see the excess group delay so that I can set the delay on my sub. Here are the plots, why is the "Excess group delay" option greyed out ?

You need to let REW calculate/generate those curves:

Also, you might have noticed a massive dip at 100hz, I think I can correct for that.

Likely caused by boundary interference reflection which you can either only partially compensate with EQ or move slightly via repositioning.

I am also looking for suggestions on setting the delay for the sub.

Try searching/skimming through previous threads or posts that discuss subwoofer and mains speaker alignment or the REW alignment tool.

 

[pollock0424]

@ernestcarl
Thanks for your response. I was able to generate minimum phase and look at the excess phase, I varied the phase knob on my subwoofer to achieve the best possible excess GD and RT60 decay. Soon, I will take a measurement of the subwoofer and speaker and get the time alignment value from REW and further fine tune the system. I was wondering if I should align my sub to left speaker or to the right or align it to L+R?

Regarding the phase mismatch, when I swapped the acoustic reference to the right side then the phase plot was also swapped, now the right speaker phase plot looks constant from 500hz to 20khz just like the left speaker when the reference was the left speaker.

 

[ernestcarl]

I was wondering if I should align my sub to left speaker or to the right or align it to L+R?

It is always going to be time adjusted/aligned relative to the speaker channel that you’ve chosen to be the time reference.

BTW, I sometimes do L+R just for convenience.

For example, speaker A is 1 meter away from the mic MLP and speaker B is 1.2, and speaker C is 2 m away. Let’s also say the sub is in one of the far corners of the room. If you cannot reasonably get all channels to be symmetrical in distance through positioning, you should use the channel that is the farthest as the reference — applying delays to rest of the other speakers so that they all have synchronized arrival times to the mic — more important for the HF.

If the difference in distance is relatively tiny (couple of mm), then it does not matter all that much.

When phases significantly do not match around the crossover, you might need to apply additional filters, for example, an all pass to rotate the phase a bit in either of the speakers.

 

[pollock0424]

Agreed, it is all relative to a reference. Dirac on minidsp used to introduce delay/gain sometimes depending on the placement. I think that might impose restrictions on seating.. critical for me is time aligning sub bass frequencies with L+R. I might do a sweep on L+R and use that as a reference. This is all new to me and I'm still learning.
Nowadays, I use Integra drx 8.4 and I tried DLBC on it and I was less than impressed with the alignment of the subwoofer. Dirac full range however does an awesome job for a 7.1.4 system. I'm planning to redo the DLBC and if it doesn't work out, I will ditch it and put my minidsp shd back into the signal chain for handling the xover and time alignment between main L+R and sub..this way I will have a good 2.1 setup and that will also serve my HT. I think that DLBC also optimizes between satellites and subwoofer..but I'm not interested in that.

 

[ppataki]

I tried DLBC on it and I was less than impressed with the alignment of the subwoofer.

I came to the same conclusion with Dirac and also with DLBC - the sub alignment was really poor
What I did to overcome that is that I manually mixed the sub into the signal (=setting the delay and the crossover in Jriver and using VST plugins) and fed Dirac with it - so Dirac did not 'see' the sub
The sound got waaaaay better (both objectively and subjectively)

 

[pollock0424]

@ppataki
I am moving a few things out of my office to clear some space. I will wait until then to redo all of it. I have a mode at 50hz and I will use PEQ to mitigate that, although it is not audibly disturbing with subwoofer crossed at 55Hz. I plan to set the system to 7.0.4 setting in my AVR and send full band signal to minidsp SHD and I will handling the crossover to the sub there. All this is to hide my SUB from Dirac lol... this is my continegent plan if DLBC fails utterly again!

 

[ppataki]

@ppataki
I am moving a few things out of my office to clear some space. I will wait until then to redo all of it. I have a mode at 50hz and I will use PEQ to mitigate that, although it is not audibly disturbing with subwoofer crossed at 55Hz. I plan to set the system to 7.0.4 setting in my AVR and send full band signal to minidsp SHD and I will handling the crossover to the sub there. All this is to hide my SUB from Dirac lol... this is my continegent plan if DLBC fails utterly again!

So essentially you are doing the same as I did

 

[RenPa]

I tried DLBC on it and I was less than impressed with the alignment of the subwoofer.

Hi,
I know not tha much about group delay. I read John help in REW many times. I have DLBC with an Arcam receiver. I now know that DLBC can't be trusted about group delay and phase alignment. I have 2 subs. The delays set by Dirac varies between 0 and 2,33ms in my 5.1 setup.
Which measurements with REW to do to get the results of DLBC filter and find the less worst?
Do I measure first each speakerto large individually and eanch sub full range without xo and DLBC off?
Do I measure each speaker individually and the 2 subs a each with DLBC filter on so bass management is in play?
What are the measurements method you follow to see what is good and bad in DLBC?
Or what to read about that?

 

[ppataki]

Hi,
I know not tha much about group delay. I read John help in REW many times. I have DLBC with an Arcam receiver. I now know that DLBC can't be trusted about group delay and phase alignment. I have 2 subs. The delays set by Dirac varies between 0 and 2,33ms in my 5.1 setup.
Which measurements with REW to do to get the results of DLBC filter and find the less worst?
Do I measure first each speakerto large individually and eanch sub full range without xo and DLBC off?
Do I measure each speaker individually and the 2 subs a each with DLBC filter on so bass management is in play?
What are the measurements method you follow to see what is good and bad in DLBC?
Or what to read about that?

What microphone are you using?
After many failed attempts it turned out in my case that the UMIK-1 mic shall not be used for precision delay measurements so I have upgraded to UMIK-2 that has its own clock source and is capable of really precise and consistent delay measurements. That might help in your case too to begin with

 

[RenPa]

...umik-1...for me. If needed I will get it. But in the mean time what is the right way to measure if searching for the less bad DLBC filter?

 

[ppataki]

...umik-1...for me. If needed I will get it. But in the mean time what is the right way to measure if searching for the less bad DLBC filter?

I did it with my PC and not with an AVR so I am not sure if this will help you
What I did is that I assigned my two subs to the sub channels in the Dirac Live Processor VST plugin and did a 9-point measurement right away
DLBC has created the crossover, set the volume and the delay per sub accordingly
Then I listened to it and it sounded totally horrible. I performed subsequent measurements too, all of them sounded horrible
I suspect that the culprit was the UMIK-1 since all my non-DLBC measurements also resulted in non-realistic and inconsistent delay values
That is no longer the case since I replaced it with a UMIK-2

 

[RenPa]

Sorry I expressed myself poorly. I would like to know what settings and how to measure with rew to see good or bwd group delay. Maybe it is not the right place to ask

 

[ppataki]

If the question is how to measure group delay then here is how I do it:

- I place the microphone in the main listening position, 90 degrees (pointing to the ceiling)
- Do a sweep measurement, from 0Hz to 20kHz, 512k FFT length
- When measurement is done, I set the smoothing to 1/12 (you can use Var smoothing too if you need higher resolution in the bass)
- I check the GD tab to see the Group Delay

I do that once my sub is already integrated so I see the full picture
I do this before and after Dirac as well to see the difference

 

[RenPa]

If the question is how to measure group delay then here is how I do it:

- I place the microphone in the main listening position, 90 degrees (pointing to the ceiling)
- Do a sweep measurement, from 0Hz to 20kHz, 512k FFT length
- When measurement is done, I set the smoothing to 1/12 (you can use Var smoothing too if you need higher resolution in the bass)
- I check the GD tab to see the Group Delay

I do that once my sub is already integrated so I see the full picture
I do this before and after Dirac as well to see the difference

That's it. Thanks. So you measure speakers set to small with bass management or with DLBC filter on and then in REW we generate minium phase and compare excess group delay for each channel.

Here group delay with DLBC filter for my fronts which sound way better than the other filter

Here the second filter whicch sound not very good in compariaon for the one above.

What information those graphs can tell me to explain the difference in the SS&I quality?

 

[pollock0424]

That's it. Thanks. So you measure speakers set to small with bass management or with DLBC filter on and then in REW we generate minium phase and compare excess group delay for each channel.

Here group delay with DLBC filter for my fronts which sound way better than the other filter

View attachment 406257

Here the second filter whicch sound not very good in compariaon for the one above.


View attachment 406258

What information those graphs can tell me to explain the difference in the SS&I quality?

Can you please describe what's the difference in sound ? Is it boomy ?

 

[ppataki]

That's it. Thanks. So you measure speakers set to small with bass management or with DLBC filter on and then in REW we generate minium phase and compare excess group delay for each channel.

Here group delay with DLBC filter for my fronts which sound way better than the other filter

View attachment 406257

Here the second filter whicch sound not very good in compariaon for the one above.


View attachment 406258

What information those graphs can tell me to explain the difference in the SS&I quality?

Can you please post the mdat file?

 

[RenPa]

Can you please describe what's the difference in sound ? Is it boomy ?

No. The first with blue and green is the best: I hear the low mid, mid, upmid and presence way better. The bass are more integrated, the dialogue clearer etc.

 

[RenPa]

Can you please post the mdat file?

Here it is.
3 days

GD xcl 7595.mdat

1 file sent via WeTransfer, the simplest way to send your files around the world

we.tl

 

[ppataki]

Here it is.
3 days

GD xcl 7595.mdat

1 file sent via WeTransfer, the simplest way to send your files around the world

we.tl

Thank you
So there are 9 measurements there, can you please explain them?
Which are the two that you would want to compare?