TOPIC:

I attempted to perform the gain structure optimization procedure, setting as suggested the level of -6 dB for the subwoofer and -12dB for the speakers.
However, when running the 9 tests, some of the patterns - almost the same - produce errors or clipping, in a pop-up window and with the graph shown red instead of light blue, causing the running test to fail. The pattern producing errors is always located in the low bass region, where it sounds actually very loud.
I tried in several ways to lower the volume generated at the speakers, while increasing the mic sensitivity in order to keep the recommended signal level in the ratios (-12 / -6dB) but obviously nothing changes in the ratio of the levels. Rearranging the levels, keeping the volume absolutely moderate, i.e. bailing out of the green coloured area at much less than -12dB, gives then no error, but we are out of the specifications recommended by the procedure.
To a further extent, lowering the volume even more, causes a complain from the software due to the too low level of sound, aborting the test as well.
Am I doing something wrong?

Do you have a lowpass filter on your sub?

Oh I think I see what's happening. The setting the sub at - 6dB is only when doing the gain structure calibration (page 42 of manual). It's not for when you are doing measurements. When doing the measurement, just set the channel volumes so that the measurements complete.

However do also disable the lowpass filter on your sub, or raise to the highest frequency if it can't be disabled.

I'm using a little B&W AS1. As far as I can see, it's got just a level knob and movie/music button switch.

I've rather lost the way on this subject. So a couple of questions:

1) After doing the "dirty trick" you said, at the measurement stage, do I need, at playing, to raise back the volume of my set, so as to still reflect the optimized gain structure?
2) if this true, when finally playing, what prevents an actual clipping in the regions of sound space that clipped at the measurements?

I never said "dirty trick."

First, sorry.
It was not from you saying, indeed I cited myself.
That just sounds "dirty" to me until I do understand thoroughly.

Hi again, I don't know if this will help, but:

1a. When doing measurements, follow the instructions for doing measurements. (Forget about gain optimization, that was before). You can adjust channel levels etc (page 25). Dirac Live is just trying to get a good signal to noise ratio in the measured signal. However the signal (before Dirac EQ) may be very "peaky" so there's a tradeoff.

1b. If you are asking if, after doing the Dirac Live EQ, you will have some overall "insertion loss", I think that would be expected because of the effect of the EQ.

1c. If you are asking if you change the volume setting on the subwoofer after doing the measurements, the answer would be no, definitely not.

2. The "clipping" that Dirac Live complains about when doing a measurement is on the input signal. So it is irrelevant on playback (output).

[Edit] One more thought: be sure to use the right speaker configuration (Sound System page).

Yes, actually the manual has a wealth of information to dig, not counting yours of course.
When getting a clipping on a measurement on a speaker channel, readjusting the relevant output volume allows to complete the measurement and that fixed my problem.
As you suggested, my mistake was to overlook outputs and input in their correct context . So, outputs when doing gain structure optimization and inputs for taking measurements. Thank you for your advice.
Rather, I ran sometimes into problems with DLCT producing the test signal, getting "play stimulus failed" error, with no signal output, mostly after just moving the output volume slider. Closing and restarting the DLCT, perhaps more times, usually fixes the problem.