Pulse Peak of linearized XOs not at sample position 32768

Post Reply
PeterB
Posts: 8
Joined: 17 Nov 2019 19:52

Pulse Peak of linearized XOs not at sample position 32768

Post by PeterB » 20 Feb 2021 09:31

Hi,

I am trying to linearize and time align active 3-way speakers plus subwoover following the steps in Mitch Barnetts book.
During the time aligning process I noticed that the pulse peaks of the linearized XO2L48.dbl, XO2R48.dbl, XO1L48.dbl and XO1R48.dbl (woovers and the subwoover) are not at sample position 32768.

Is this the expected behavior or did I make a mistake somewhere in the linearization process?

I repeated the linearization and took some screenshots

Best Regards,
Peter
Attachments
1.png
1.png (72.85 KiB) Viewed 1643 times
2.png
2.png (56.85 KiB) Viewed 1643 times
User avatar
UliBru
Posts: 445
Joined: 19 Oct 2019 13:58

Re: Pulse Peak of linearized XOs not at sample position 32768

Post by UliBru » 20 Feb 2021 10:57

The reason is simple:
the linearphase XO has its peak in the center = 32768 (of 65536 taps).
Now you convolve it with a minimumphase linearization and thus the peak of the mixed phase result gets shifted. This is normal.
You can still follow the steps of Mitch's book.
Stay well tuned
Uli

Moderator
Acourate system: JRiver/Roon -> AcourateConvolver -> miniDSP U-DIO8 -> TacT M/S2150 amps -> DIY horn speakers
PeterB
Posts: 8
Joined: 17 Nov 2019 19:52

Re: Pulse Peak of linearized XOs not at sample position 32768

Post by PeterB » 20 Feb 2021 11:58

Hi Uli,

thanks a lot, so my linearized XOs are ok.

So now follows the time alignment.
I rotated the individual linearized XOs for the delay measurement, made a 4-Way XO and took a Logsweep-measurement.

Now comes the comparison of the Logsweep with the linearized XOs.
For the Left Woover, the (for the comparison rotated) linearized XO2L48.dbl has its peak at 4014.
The pulse peak for the left woover in the measurement is at 4037.

Now I want to rotate the linearized XO2L48.dbl, which has its peak at 32782, for the final measurement.
Do I have to rotate it by -37 samples or -23 (37-14) samples?

Best Regards,
Peter
User avatar
UliBru
Posts: 445
Joined: 19 Oct 2019 13:58

Re: Pulse Peak of linearized XOs not at sample position 32768

Post by UliBru » 22 Feb 2021 12:59

Peter,

the underlying basic idea is the following:
- it is assumed that the pulse peaks are perfectly aligned
- now e.g. XO1 is rotated by -2000 samples
- the measurement will show the tweeter peak at 6000 and the bass peak is at 4000 (as it is rotated by -2000)
- now if the bass peak is at position 4037 (non-perfect world), it is 37 samples too late and thus we have to rotate XO1 at the end by -37 samples from original version without -2000 samples test rotation

The real world of course introduced mixed phase XOs plus driver behaviour. Thus the rotation method is helpful but can be mind-boggling.
Thus in addition I have proposed a further more accurate method. You find it described by www.audiovero.de/freedownload/Time_Alig ... lution.pdf
Stay well tuned
Uli

Moderator
Acourate system: JRiver/Roon -> AcourateConvolver -> miniDSP U-DIO8 -> TacT M/S2150 amps -> DIY horn speakers
Post Reply