Four ways active speakers alignment

[kmajjo]

The values follow the cursor position - currently positioned at 16.38

Ok , thanks.

I use time acoustic reference for the measurements from the tweeter.
In the measurement of the bass medium speaker i can read the delay of -2.4221ms and -1.6913 ms for the medium.
My first conclusion is that the bass medium speaker is delayed compare to the medium.
The time reference being the same , the delay between bass medium ans medium is 0.7308 ms.
If i compare the "ref" in the impulse windows ,for bass medium ref = -0.505 ms and for medium "ref" is -1.50ms ,so the delay between the two speakers is 1,005ms.
If i compare the IR delay , for bass medium IR is -2.9087ms and for medium -1.6913ms , so the difference is 1.2174ms.
Now , i choose the alignment tool in all SPL for bass medium and medium , align phase at cursor (350 hz) and the delay on the medium is -14,37ms.
Could you help me to explain ?
Br
Jacky

 

[jtalden]

I use time acoustic reference for the measurements from the tweeter.
In the measurement of the bass medium speaker i can read the delay of -2.4221ms and -1.6913 ms for the medium.

Yes, but in the AT process we have been using these values are not used. We are looking for good SPL support and close phase tracking through the XO range of the direct sound.

My first conclusion is that the bass medium speaker is delayed compare to the medium.
The time reference being the same , the delay between bass medium ans medium is 0.7308 ms.
If i compare the "ref" in the impulse windows ,for bass medium ref = -0.505 ms and for medium "ref" is -1.50ms ,so the delay between the two speakers is 1,005ms.
If i compare the IR delay , for bass medium IR is -2.9087ms and for medium -1.6913ms , so the difference is 1.2174ms.

The initial rise is also not a reliable indicator of strong SPL and close phase tracking. That works only for ideal LR-24 conditions and was shown as a example in the learning exercise to demonstrate that aligning the peaks of the impulses is not a reliable indication of favorable SPL or phase tracking. Both of these indicators do place the timing in the general range of the correct timing however so we should find a good solution using the AT without straying too far away from this range.

Now , i choose the alignment tool in all SPL for bass medium and medium , align phase at cursor (350 hz) and the delay on the medium is -14,37ms.
Could you help me to explain ?

The SPL support of the -14.37 ms medium setting looks okay. The phase is aligned at the cursor, but it does not result in close phase tracking throughout the XO range of the direct sound. The phase chaos due to the room effects can confuse the AT result sometimes. That is why it is recommended to apply some level of FDW to clean up the phase trace to better indicate the direct sound before the 'Align phase at cursor' button is applied.

FDW = 6 cycles


Then we can remove the FDW to confirm if the SPL is still favorable. If not, we can use the slider in the AT to see if another nearby setting is more favorable. This setting looked okay to be without manual adjustment so I did not do any further investigation.

No FDW

 

[kmajjo]

Thank you very mutch for this explanation.

With IR Window option , i can obtain the same result on the SPL and phase curves than the "Add frequency dependent window" when i reduce the "Left window (ms)" and "Right Window(ms)" values.
Could you give me some explanations or a link to understand the purpose of the "Add frequency dependent window" option.
Br
Jacky

 

[kmajjo]

Some more questions :

- what is AT process (#27) => Alignment Tool?

- in the Alignment Tool , when i have a negative delay proposal , what is the meaning ?
For exemple , in the #27 post, your calculation give a delay of 3.21 ms on the 1-ms-bm-110-375 speaker (Low medium), so i can setup this delay in my DSP but if the delay is negative ?
What is the signification and the solution to apply the negative delay ?

Br
Jacky

 

[jtalden]

Could you give me some explanations or a link to understand the purpose of the "Add frequency dependent window" option.

The FDW helps us to see the phase rotation of the direct sound. A FDW of 6 cycles limits the data analyzed to the first 6 cycles at each frequency rather then having the entire window setting being applied to all frequencies. This removes much of the impact of the later arriving sound at each frequency.

 

[jtalden]

- what is AT process (#27) => Alignment Tool?

Yes

- in the Alignment Tool , when i have a negative delay proposal , what is the meaning ?
For exemple , in the #27 post, your calculation give a delay of 3.21 ms on the 1-ms-bm-110-375 speaker (Low medium), so i can setup this delay in my DSP but if the delay is negative ?
What is the signification and the solution to apply the negative delay ?

This is a relative delay, so negative delay of the 'med' the same as positive delay on the 'bm'.

 

[kmajjo]

Thank you very much for your help.

 

[kmajjo]

This is a relative delay, so negative delay of the 'med' the same as positive delay on the 'bm'.

If i do that, SPL , Impulse, GD curves are the same but phase curve is different

 

[jtalden]

See Post-12.

 

[kmajjo]

Hello,

If the usage of REW is more familiar for me now , thanks to you , but your experience is needed to solve the following problem :

- delay between sub and bass is 2.46ms , i must delay bass loudspeaker of this value.
- delay between bass and medium is 1.84 ms , i must also delay bass loudspeaker of this value.
- delay values are different but must be set to the same loudspeaker and i can set only one delay value for the bass loudspeaker to my dsp.
What is the best solution for you ?

 

[jtalden]

These are relative driver to driver delays. A speaker management box is the normal way to control XOs, delays and PEQ for all outputs. Does your DSP not have a delay capability for each output?

If not ,and you must chose, then the bass medium is probably more likely to be the better choice. You can look at the SPL tradeoff and chose the one that that you think is more favorable in you particular situation.

 

[kmajjo]

True , so absolute delays are :

 

[jtalden]

Good - be sure to confirm no mistakes were made. Set those delays in the DSP and measure each driver alone. The AT analysis of each set of adjacent drivers will now match the chosen timing relationship without any delay adjustments.

 

[kmajjo]

Hello,
In the Alignment tool , i can open a "Show phase trace window" (see attached picture).
Which are the phase curves displayed ? ( i suppose that are loudspeaker phase , minimum phase , and excess phase) : green curve , black curve , purple curve ?
Br
jacky

 

[jtalden]

You have selected measurements 15 and 10 so those show up in the phase chart. You should also select 15 and 10 in the legend so that they also show up in the SPL chart. Usually we want all other measurements deselected so the charts are not too busy to read. The black traces are the AT calculated result with the current settings in the AT.

 

[kmajjo]

Hello,
Could you examine my alignment problem please ?
Br
jacky

 

[jtalden]

Given the current data choices the better SPL and time alignment appears to be your AT analysis #7.

I also used your 4 driver measurements to determine the timing I would choose using this data. Your Tw was the first arriving sound so I used that as the reference time for convenience. The table below then shows the the timing changes to the other 3 drivers.


Comments:
It appears the M and Tw impulses have been FIR EQ filtered. The resulting direct sound phase rotation is greatly disrupted. Normally the room response creates a very chaotic phase response at the LP, but using a REW FDW filter the direct sound shows the expected smoothly changing phase response of the driver. In this case both the LP room phase and the FDW direct sound phase are very chaotic. This is not the phase response most of us would strive for. That said, we are not particularly sensitive to phase differences unless there are gross timing errors. The SPL response of your analysis #7 is identical to my suggested timing in the table so I would not necessarily expect there to be sound quality difference.

The REW AT was not very helpful in this case as the direct sound phase response was so disrupted. I first relied on the group delay and impulse charts to find the initial timing estimate and then adjusted timing as needed to cross the phase at the XO point to achieve reasonable SPL support in the XO ranges. The driver polarity was chosen to maximize the SPL and minimize the timing shift from the initial setting.

Attached is mdat file that only provides the 4 driver measurements and the resulting summed response. The 4 driver impulses have been shifted and inverted as needed to represent the situation shown in the above table. You can use this file to compare this analysis to your previous files.

 

[kmajjo]

Hello,
Thank you very much for your detailed and very informative response for me.

It appears the M and Tw impulses have been FIR EQ filtered. The resulting direct sound phase rotation is greatly disrupted. Normally the room response creates a very chaotic phase response at the LP, but using a REW FDW filter the direct sound shows the expected smoothly changing phase response of the driver. In this case both the LP room phase and the FDW direct sound phase are very chaotic. This is not the phase response most of us would strive for. That said, we are not particularly sensitive to phase differences unless there are gross timing errors. The SPL response of your analysis #7 is identical to my suggested timing in the table so I would not necessarily expect there to be sound quality difference.

Yes, REW measurements have been made with FIR EQ ON.
I didn't try to make REW alignment with FIR EQ OFF.
So , you will find attached file with FIR EQ off , the sweep measurement is only limited by REW frequencies limit to protect Medium and Tweeter loudspeakers.
Br
Jacky

 

[kmajjo]

Given the current data choices the better SPL and time alignment appears to be your AT analysis #7.

I also used your 4 driver measurements to determine the timing I would choose using this data. Your Tw was the first arriving sound so I used that as the reference time for convenience. The table below then shows the the timing changes to the other 3 drivers.
View attachment 38055

So , if i understand well , the setup delays in my DSP must be :

Br
Jacky

 

[kmajjo]

So , if i understand well , the setup delays in my DSP must be :
View attachment 38060
Br
Jacky

Something is wrong in my conclusion, i am not able to reproduce your SPL

 

[jtalden]

So , you will find attached file with FIR EQ off , the sweep measurement is only limited by REW frequencies limit to protect Medium and Tweeter loudspeakers.

There is no need to change the analysis. The best possible analysis was done given the FIR Filter usage. There is no needed to time the direct sound phase without the FIR filter as it will be disrupted again when the FIR is again turned on. My point was the FIR filters are making unwanted changes to the direct sound and thus further disrupting the room phase responses of the system. It would be advisable to replaced these with FIR filters with ones that do not do that. If new FIR filters with good phase behavior are created then new timing analysis should be done. This optional effort may not impact sound quality unless the SPL is also improved.

 

[jtalden]

So , if i understand well , the setup delays in my DSP must be :
View attachment 38060

No, this isn't correct. Note that these are changes from the current delay settings used for the measurements. If you change all current settings per my table by 8.5 ms then the SB would need to be changed by 17 ms to obtain the target timing.

But...
I found an error when double checking the table values. The SB value in the table should have been -6.8 ms. Below I corrected the table and also added an example showing how the changes would be made if the measurement delay settings were as shown in column D. I hope you haven't forgotten the actual delay settings used for the measurements. They are needed for Column D.

 

[kmajjo]

But...
I found an error when double checking the table values. The SB value in the table should have been -6.8 ms. Below I corrected the table and also added an example showing how the changes would be made if the measurement delay settings were as shown in column D. I hope you haven't forgotten the actual delay settings used for the measurements. They are needed for Column D.
View attachment 38082

Sorry, but I can't follow your reasoning, I'm lost.

I hope you haven't forgotten the actual delay settings used for the measurements. They are needed for Column D.

What do you mean by this ?

Ultimately , I think i will keep my current delays even if the phase curve is not as good as yours because i am not able to reproduce your reasoning.
Sorry, thank you again for your help.

 

[jtalden]

Why don't you send me the actual delay setting you used for each of the 4 drivers during the measurements of Post-43. From that I can use the table to tell you the actual 4 delay settings needed to create the optional response.

 

[kmajjo]

Sorry , but post-43 measurements are without delay and no xo ( delays setting in the DSP are all to 0 ms) .
In post-41 it's the same same measurements ( delay setup = 0 ms) but xo are on.