New Neumann MA 1 - Automatic Monitor Alignment

[ernestcarl]

Sorry, I did not look carefully! Duh. It's already included in the mdat. Thanks!

 

[ernestcarl]

Here are some plot views for the corrected response:

SWEPT SINE (FDW 10 cycles)

WAVELET & GROUP DELAY

Excess GD is kept as low as possible.


COMPARED TO MY S8 (PORT BLOCKED)

L+R VECTOR AVERAGED up to 200ms

ZOOMED-IN WAVELET (time-domain) COMPARISON

My EQ'd Sceptre S8 measured at 2m

*The Sceptre tweeter+mid woofer step looks 'quicker' than the KH310 minus the phase uncorrected sealed sub -- dunno how this will translate with actual listening. Some people criticize 'overly fast' drivers... I think with the ported KH120 will still be slower even after time-domain correction.

All-in-all, results are astonishingly good! I might get a KH750 to pair with my existing KH120s in the future.

 

[thewas]

*The Sceptre tweeter+mid woofer step looks 'quicker' than the KH310 minus the phase uncorrected sealed sub -- dunno how this will translate with actual listening. Some people criticize 'overly fast' drivers...

It is problematic using LP measurements from 2 different rooms to compare those, better would be nearfield measurements at the same exact position and room, even better in a large anechoic room, also comparing a step response of 3-way and 2-way loudspeaker and even more making optical analogies to "fast drivers" but we have discussed this in another place before.

 

[ernestcarl]

It is problematic using LP measurements from 2 different rooms to compare those, better would be nearfield measurements at the same exact position and room, even better in a large anechoic room, also comparing a step response of 3-way and 2-way loudspeaker and even more making optical analogies to "fast drivers" but we have discussed this in another place before.

The plots posted for comparison are not really meant to show absolute differences replicating stuff published by S&R etc. -- I think most people will already understand that -- hopefully. Yes, I said the tweeter-mid driver step looked 'quicker', but that is an objective observation clearly visible in the graphs. If you look at the peak energy time (wavelet plots), though, they look almost exactly the same. I have measured the step as close as 15 cm for other two-ways like the KH120 and the difference between this at 15cm, 1 meter and 2 meters is minimal. How that translates in audible differences between a three-way and two-way coax which are very different designs placed in the same room, I have no idea -- and so I left no subjective opinion of my own.

 

[DJBonoBobo]

@ernestcarl

Thank you for working with my measurements! I don't understand exactly what it all means, but I'm happy to see it looking good!

Just to be on the safe side and for clarification, I'll explain again what I did:

Corrected system:
measurements are called "L+sub network" and "R+sub network": RME ADI-2 DAC -> KH750 DSP network mode (filters set by MA 1) -> KH310

Additional measurements:
"L local", "R local": RME ADI-2 DAC -> KH310 (no sub, no filtering)

"sub network": RME ADI-2 DAC -> KH750 DSP network mode (filters set by MA 1) -> 1 KH310 for acoustical timing reference, the other one switched off, so the measurements shows sub only

"sub local": RME ADI-2 DAC -> KH750 DSP local mode (no filtering) -> 1 KH310 for acoustical timing reference, the other one switched off, so the measurements shows sub only

"L+sub local", "R+sub local": RME ADI-2 DAC -> KH750 DSP local mode (no filtering; delay and phase not optimized) -> KH310

Edit: UMIK-1 untouched at same position the whole time.

 

[ernestcarl]

@ernestcarl

Thank you for working with my measurements! I don't understand exactly what it all means, but I'm happy to see it looking good!

Just to be on the safe side and for clarification, I'll explain again what I did:

Corrected system:
measurements are called "L+sub network" and "R+sub network": RME ADI-2 DAC -> KH750 DSP network mode (filters set by MA 1) -> KH310

Additional measurements:
"L local", "R local": RME ADI-2 DAC -> KH310 (no sub, no filtering)

"sub network": RME ADI-2 DAC -> KH750 DSP network mode (filters set by MA 1) -> 1 KH310 for acoustical timing reference, the other one switched off, so the measurements shows sub only

"sub local": RME ADI-2 DAC -> KH750 DSP local mode (no filtering) -> 1 KH310 for acoustical timing reference, the other one switched off, so the measurements shows sub only

"L+sub local", "R+sub local": RME ADI-2 DAC -> KH750 DSP local mode (no filtering; delay and phase not optimized) -> KH310

Edit: UMIK-1 untouched at same position the whole time.

Does the software show the total time delay for sub+speaker phase linearization? I'm interested in how much delay it adds for viewing streamed web video content.

 

[DJBonoBobo]

Does the software show the total time delay for sub+speaker phase linearization? I'm interested in how much delay it adds for viewing streamed web video content.

No, i don´t think so. I don´t know how much delay is added. From my experience with trying different delays manually in this room to try aligning the sub and speakers i would guess something between 10 and 20ms, but i really don´t know.

Edit: By the way, physical distance LP to sub is quite close to LP to mains, less than 10cm difference, i think.

Edit2: Could it be possible to derive the delay from the acoustic timing reference signals? If you compare "L local" with "L+sub+network"? I don´t know.

 

[ernestcarl]

No, i don´t think so. I don´t know how much delay is added. From my experience with trying different delays manually in this room to try aligning the sub and speakers i would guess something between 10 and 20ms, but i really don´t know.

Edit: By the way, physical distance LP to sub is quite close to LP to mains, less than 10cm difference, i think.

10-20 was indeed cited for the main speakers by Neumann and Hedd (youtube videos I saw). The sub should be in the order of hundreds of milliseconds -- but varies a huge deal depending on the correction required and extension of the bass response -- still would be informative if the MA-1 showed this. My understanding is that the KH80's own internal phase correction primarily for the xo uses only 2ms which is fast. The Sceptre is 2.5ms -- but any kind of phase corrections extending to the rest of the response + room = something much longer.

 

[ernestcarl]

There is something quite curious about the correction in the Neumann sub+mains phase-linear system in that it does not try re-create a more "perfect" looking step between the sub & mains like we see in others e.g. Audiolense & Dirac. I dunno if this is just a one-off thing where its AI decided not to do it in this particular situation for "best results".

 

[thewas]

There is something quite curious about the correction in the Neumann sub+mains phase-linear system in that it does not try re-create a more "perfect" looking step between the sub & mains like we see in others e.g. Audiolense & Dirac. I dunno if this is just a one-off thing where its AI decided not to do it in this particular situation for "best results".

Maybe also not to increase the latency too much?

 

[DJBonoBobo]

I don´t know what you mean with curious, but it should be noted that L and R are quite different in the crossover region (FR and phase). So maybe aligning to both channels had to be a compromise.

 

[ernestcarl]

Maybe also not to increase the latency too much?

That's definitely a possible reason. Their target audience (working pros) is sensitive to that kind of thing. Whereas I suspect Dirac caters more to people with more traditional dedicated stereo listening environments where delay is not necessarily a priority.

At least Hedd provides an estimate in their software's selection view:

 

[ernestcarl]

I don´t know what you mean with curious, but it should be noted that L and R are quite different in the crossover region (FR and phase). So maybe aligning to both channels had to be a compromise.

The step in your left and right channels look almost identical in REW. This is just a suspicion, but I think the guys at Neumann might see that achieving a "good looking" overall sub+mains step response as low priority. Mind you, the tweeter+mids+bass driver corrected step already looks perfectly aligned anyway. In the sub bass, this must not matter all too much.

 

[ernestcarl]

I think I see clearer why the phase-linearized step of the S8 looks "quicker" than that of the KH310 -- it could just be an illusion.

Extreme windowing:

*much more evident with the extreme nearfield measurement I took at 15cm.

The high resolution wavelet is a much better default view of the energy peak time response performance between different speakers.

I think with the ported KH120 will still be slower even after time-domain correction.

In the mids and highs where linear phase correction is more relevant (not really sure), the result probably would be the very similar after all.

@DjBonoBobo

Whenever you can find the time, I was wondering if you can take a quick listen to the track Foc by Rodrigo y Gabriela with your KH310's phase-linearization turned on vs off. The bass is unimportant. I'm just interested if you can perceive any "brightening" colouration effect in the mid and high transients from the guitars etc. or any kind of improved clarity.

Maybe you won't be able to hear any difference at all, which would be very informative, nevertheless.

 

[DJBonoBobo]

I think I see clearer why the phase-linearized step of the S8 looks "quicker" than that of the KH310 -- it could just be an illusion.

Extreme windowing:
View attachment 105788
*much more evident with the extreme nearfield measurement I took at 15cm.

The high resolution wavelet is a much better default view of the energy peak time response performance between different speakers.

View attachment 105791



In the mids and highs where linear phase correction is more relevant, the result probably would be the very similar after all.

@DjBonoBobo

Whenever you can find the time, I was wondering if you can take a quick listen to the track Foc by Rodrigo y Gabriela with your KH310's phase-linearization turned on vs off. The bass is unimportant. I'm just interested if you can perceive any "brightening" colouration effect in the mid and high transients from the guitars or any kind of improved clarity.

Maybe you won't be able to hear any difference at all, which would be very informative, nevertheless.

Unfortunately i am not able to do this because in my current setup i cannot turn linear phase on/off without touching the EQ for the frequency response. This would only be possible if i would reset the EQ filters and reactivate the iPad-app, but i will not do this, because i would have to make all the measurements again. (Also it would require to go to the subwoofer and press the switch for "local", so no "on the fly" comparison possible, i think)

 

[ernestcarl]

Unfortunately i am not able to do this because in my current setup i cannot turn linear phase on/off without touching the EQ for the frequency response. This would only be possible if i would reset the EQ filters and reactivate the iPad-app, but i will not do this, because i would have to make all the measurements again. (Also it would require to go to the subwoofer and press the switch for "local", so no "on the fly" comparison possible, i think)

No probs!

 

[NoTBaTMaN]

I have attached my measurements and what the MA-1 software is showing.
Both KH310A are now soft-flush-mounted between my basstraps to reduce SBIR problems, the sub is at 33% room width from the right side.
The measurements are:
- KH310A full-range without sub (L,R,L/R)
- KH310A with sub and bad manual settings (L/R)
- KH310A with sub after alignment process but alignment inactive (L,R,L/R)
- KH310A with sub after alignment process with MA-1 alignment (L,R,L/R)

Everything is measured with the uncalibrated MA-1 microphone.
What do you think?

EDIT: Download link: https://www.file-upload.net/download-14439092/KH310A-KH750DSPMA-1_TEST.mdat.html
EDIT2: Room dimensions: 489x357x240 cm (LxWxH)

 

[ernestcarl]

EDIT: Download link: https://www.file-upload.net/download-14439092/KH310A-KH750DSPMA-1_TEST.mdat.html
EDIT2: Room dimensions: 489x357x240 cm (LxWxH)

While I was able to download the mdat (scanned fine), upload site tried to redirect me to grab another *.exe file so... pretty sketchy so I would avoid that file host site.

 

[ernestcarl]

I think your REW measurements (if correct) tell a different story... seems like your sub may be positioned badly as you aren't getting much volume from it even when "aligned".

 

[ernestcarl]

Less important, the positioning of your monitors/mic seems a bit off... but the phase seems okay if I time offset each channel before averaging.

*Sub SPL is hardly there... as it's fighting with the room.