Neumann MA1 vs. Dirac Live Bass Control (DLBC) - Dual KH750 subs

[DJBonoBobo]

Hello,

i have a dual sub Neumann system (2x KH310, 2x KH750 DSP). I use the MA1 microphone and software for room EQ. I also own a Dirac Live license (PC standalone) that i have not used for a long time. Because Dirac released DLBC, i decided to get the trial version and compare the results of DLBC and MA1.
Dirac offers a discount for DLBC until March 22nd, so maybe this is of interest for some. Unfortunately i had no time making listening tests and my trial license is already expired today.

Disclaimer: I am a home user with no special technical expertise in doing this comparison. I am only sharing what I noticed in my room and make no claim that the results can be applied to others.

Room:
This is how the setup looks like at the moment, listening distance is about 1.80 meters. Best positions for the subs are the corners. The room is about 3.0x5.0x2.5 meters. As you can see i have some treatment (mostly basotect, also 2 DE-MITTER from greenacoustics and 2x PSI AVAA at the back of the room. There is also a bed and a big closet filled with clothes) in the back of the room.

Abbrevations:
MA1: Measurements with MA1-correction active.
DLBC: Measurements with DLBC correction active.
no prefix: Measurements of speakers without DSP.
L: Left KH310
R: Right KH310
LSub: Left KH750
RSub: Right KH750
2Subs: Both KH750 together

Mic, crossover, volume and drivers:
In both cases i used the same mic with calibration file and the same positions (as far as possible). The crossover frequency was 80Hz in both cases. Due to the different hard- and software requirements i made the measurements not at the same level (Dirac calibration was quieter) and tried to align them afterwards for comparison. Particularly for the subwoofers i don´t really know how to do that correctly, so i can not interpret the subwoofer graphs 100% correctly. For Dirac i made Java-measurements with RWE, for MA1 i used ASIO. I don´t know i some of the differences may be influenced by this. I always used the same channel (left main) for acoustical timing reference.

Target:.
MA1: I used the automatic target curve with no customization.
DLBC: I reduced the bass from +6db to +3.5dB. As you can see it is still more bassy than the MA1 target. The default target has way too much bass for me.
In my comparison i focused on the most problematic crossover area from 70-120Hz.

Frequence response graphs:
Without correction:

With DLBC:

With MA1:

Left channel comparison:

Right channel comparison:

RMS Average comparison:

Left subwoofer comparison:

Right subwoofer comparison:

Both subwoofers (you can see different approaches of the programs - DLBC seems trying to fit both subs tightly, much more variance between MA1-subs):

Phase response graphs:

Group delay:

Step response (disclaimer: i don´t know what that means/how to interpret this - i just noticed the big difference and wondered if it´s important somehow)

Additional DLBC stuff (could not do this with the MA1)
Left KH310 only

Right KH310 only

Preliminary conclusion, what I think I understand:
- MA1 leaves >3khz alone in my case, DLBC fiddles a bit with this.
- MA1 boosts the sub around 85Hz, DLBC boosts the mains around 120Hz
- Below 30Hz is much louder with DLBC. I don´t know why MA1 is so shy there.
- DLBC looks a bit better overall because of the 120Hz-area.
- MA1 has a much better automatic target (for me in my nearfield-situation at least).
- In terms of phase and group delay the MA1 looks better, i think? Not sure though.
- The results are pretty close, but some different "choices" lead to a few differences.
- I have not yet compared latency - i don´t know how to do this.
- If you already have a Neumann system, I think the MA1 has much better value (around 250 EUR for mic+software) than DLBC (729 EUR without mic for stereo), but DLBC works with other subs and has a multichannel-option.

What do you think – is there a winner?

 

[diavolo]

Thank you very much for your valuable work! As an MA1 user, I would be delighted to see more related tests that support its quality, despite the software's many issues. My long-standing subjective experience and objective measurements all demonstrate its value

 

[Kervel]

Nice work!! I think the key difference is that DLBC targets the summed response of the subwoofers; which MA1 does not seem to do. In the third plot, MA1 with 2 subs shows a strong 12db dip at 85 hertz, which reduces to about a 6db dip with DLBC (second plot).

That said, the overall differences seem quite small.

 

[custom]

Where does the null and the bad group delay at 85Hz on the left channel come from? If the room and the placement are symmetrical, the problem should be visible on both channels. I had a similar problem at same frequency with my KH80+KH750 (just one) setup and it was a SBIR reflection over the back wall.

 

[DJBonoBobo]

Where does the null and the bad group delay at 85Hz on the left channel come from? If the room and the placement are symmetrical, the problem should be visible on both channels. I had a similar problem at same frequency with my KH80+KH750 (just one) setup and it was a SBIR reflection over the back wall.

I don't know, but probably the door. Everything except door and window is symmetrical.

 

[Kervel]

Where does the null and the bad group delay at 85Hz on the left channel come from? If the room and the placement are symmetrical, the problem should be visible on both channels. I had a similar problem at same frequency with my KH80+KH750 (just one) setup and it was a SBIR reflection over the back wall.

This frequency is right at the crossover, so the two subs and the main play about equally loud. It seems that 2subs+L gives a dip while 2 subs+R do not. There must be some reflection from L that destructively interferes with one of the subs.

 

[custom]

I don't know, but probably the door. Everything except door and window is symmetrical.

That could explain it. You can start the REW SPL meter and then play a 85Hz tone with the generator on the left channel. If you hold an absorber (min 4 inch deep) in front of the door, the SPL readings should increase if the door is the problem. I assume that the lower part of the door is the problem.

Did you use the latest MA 1 version 1.6.2 for the alignment?

 

[DJBonoBobo]

Did you use the latest MA 1 version 1.6.2 for the alignment?

Yes, 1.6.2. Also Dirac Live 3.4.4 and Processor 1.7.5.

 

[palm]

Very interesting, thanks for sharing.

MA1 “knows” exactly the speakers and crossovers and can correct the phase theoretically (independently of measurements) and it shows I think. Doesn’t mean this is super audible (my own tests on this with rePhase with O300 and single O800 were in larger and non treated room, your nice setup has much more chance to reveal anything)

Looks like there is a dip at 180Hz with DLBC, this would probably make me check how annoying this is?

Your setup also look like an interesting case study for the room mode simulation, I don’t have any experience with this but with a rectangular room it has more chance to correlate with reality. Maybe you already posted on this, sorry if that’s the case.

 

[Muddywaters]

Thank you very much for sharing this. KH 420’s are on my short list although I’m doubtful I’ll incorporate Neumann subs and MA-1 but am very interested in multiple subs and DLBC.

 

[ernestcarl]

- MA1 has a much better automatic target (for me in my nearfield-situation at least).

What do you think – is there a winner?

Yeah, the MA-1's time domain performance appears better in your graphs here and has clearly done a better job at linearizing the speakers' phase -- kind of expected given the correction software is tailor-made for their own speakers. I do not particularly like seeing the ~30ms 0 dBFS peak (and frequency-dependent delayed bass energy) in those DLBC (step view) envelope time curves. As to which is the winner, I can't be so sure myself just by looking at the graphs, and would much prefer an in-person critical listening audition.

 

[Davide]

It would be interesting to measure the delay introduced by each correction system. Much of Dirac's research is aimed at finding sparing correction solutions in terms of delay and computing power, while achieving a satisfactory sonic target. There's a reason they sacrifice some linearity in the time-time domain.
It must be said that I don't know how MA works, so I don't know if the processing is done by the PC or by the internal DSP. In this last case, if it is not completely bypassable (I doubt), it will introduce an additional delay with Dirac, so the comparison is not even.

 

[DJBonoBobo]

Your setup also look like an interesting case study for the room mode simulation, I don’t have any experience with this but with a rectangular room it has more chance to correlate with reality. Maybe you already posted on this, sorry if that’s the case.

Yes, the room simulator (pink) indeed shows some similarity. Some of the peaks and dips correlate with the actual measurement.

 

[DJBonoBobo]

Here is phase again with a FDW of 15 cycles (disclaimer: i don´t know what i am doing with this, but i remember doing this has been recommended in the past):

 

[DJBonoBobo]

It would be interesting to measure the delay introduced by each correction system. Much of Dirac's research is aimed at finding sparing correction solutions in terms of delay and computing power, while achieving a satisfactory sonic target. There's a reason they sacrifice some linearity in the time-time domain.
It must be said that I don't know how MA works, so I don't know if the processing is done by the PC or by the internal DSP. In this last case, if it is not completely bypassable (I doubt), it will introduce an additional delay with Dirac, so the comparison is not even.

I don´t know how to do that - i don´t know of the following information from REW helps? I don´t think so. I think i should have used one of the analog rear channels as a reference instead, pity...

L: Delay -0,0026 ms (-0,89 mm, -0,04 in) using estimated IR delay relative to Acoustic reference played from 1: Analog 1 (1) with no timing offset
DLBC L+2Subs: Delay -0,0056 ms (-1,9 mm, -0,08 in)
MA1 L+2Subs: Delay -0,0477 ms (-16 mm, -0,64 in)

R: Delay -0,0014 ms (-0,47 mm, -0,02 in)
DLBC R+2Subs: Delay 0,0039 ms (1,4 mm, 0,05 in)
MA1 R+2Subs: Delay -0,0492 ms (-17 mm, -0,67 in)

 

[palm]

I have no experience with MA1 but I understand it is correcting for the analog crossovers in the system, like one could do with rePhase, so you can get close to a perfect step response (beside maybe the subsonic filter of the sub to avoid a too long latency) - so in room measurements should only show how the room disturbs this ideal response.

I haven’t played with REW for quite some time but you can adjust the IR centering and make Dirac and MA1 following each other around 8k on the phase plot. I assume Dirac will heavily window at high freq to only correct for quasi anechoic there. But going down in frequency it will correct for both speaker and room. I guess MA1 corrects phase based on known system properties and FR based on measurements.

 

[Davide]

I don´t know how to do that - i don´t know of the following information from REW helps? I don´t think so. I think i should have used one of the analog rear channels as a reference instead, pity...

L: Delay -0,0026 ms (-0,89 mm, -0,04 in) using estimated IR delay relative to Acoustic reference played from 1: Analog 1 (1) with no timing offset
DLBC L+2Subs: Delay -0,0056 ms (-1,9 mm, -0,08 in)
MA1 L+2Subs: Delay -0,0477 ms (-16 mm, -0,64 in)

R: Delay -0,0014 ms (-0,47 mm, -0,02 in)
DLBC R+2Subs: Delay 0,0039 ms (1,4 mm, 0,05 in)
MA1 R+2Subs: Delay -0,0492 ms (-17 mm, -0,67 in)

You must first find out, with good certainty, how the MA processing takes place, ie whether in the internal DSP or on the PC/Mac.

 

[diavolo]

You must first find out, with good certainty, how the MA processing takes place, ie whether in the internal DSP or on the PC/Mac.

in the internal DSP absolutely

 

[Davide]

in the internal DSP absolutely

In this case, the next thing to figure out is whether you can bypass the DSP and leave all the handling to Dirac, so you can compare the process delay introduced by each system.
However, I doubt that the DSP can be bypassed because I believe it is a crossover to the monitor units.

 

[DJBonoBobo]

In this case, the next thing to figure out is whether you can bypass the DSP and leave all the handling to Dirac, so you can compare the process delay introduced by each system.
However, I doubt that the DSP can be bypassed because I believe it is a crossover to the monitor units.

I think i could have done that by utilizing the surround speakers, but it is too late now.