That's a good question, but hypothetical. As it stands with this data, it looks to me like a non-rising DI is preferred. Doesn't that support my contention? A few dB rise is fine, but smoother and flatter is still the goal.
The mid-range diffraction in the first speaker above is very pronounced. I would suspect this to be an audible problem. The second one also has the lowest diffraction, which should also add to its preference.
I find it odd that you would use a cap in an active system.
Would a transformer give better damping?
As I have been following this thread with great interest (Matbat, truly great job BTW) and I was waiting for the DI conversation to happen.
I may have some ideas to share in relation with the Olive Predicted Preference Rating (PPR) aka the Score vs DI.
The same with me. What catched my attention this time is this quote from the Harman study (page 6):
"... Target slopes were determined separately for Test One and for our larger test sample (70 loudspeakers) used for the generalized model described in section 5. The target values are based on the mean slope values of speakers that fall into the top 90 percentile based on preference ratings. Target slopes are defined for each of the 7 frequency curves (see table 2). The ideal target slope for the on-axis and listening window curves (0 and –0.2) is identical for both test samples, which indicates that the on-axis curve should be flat, while the off-axis curves should tilt gently downwards. The degree of tilt varies among curves for Test One and the larger sample. [...] This suggests that the ideal target slope may depend on the loudspeaker’s directivity."
If I understand the units in Table 2 (see Eq. 5), for the PIR ("predicted in-room") slope this would be a downward tilt around 4 dB between 1 - 10 kHz (and around 2.5 dB for power response alone). Is that right?
Of course this may be due to the fact that no other loudspeakers (i.e with more flat, and maybe also higher, DI) were present in the test. Would loudspeakers with more flat and high DI be rated even higher in this test? That's what we still don't know, I think.
b. - 2.99*NBD_PIR (100Hz-12000Hz) same as NBD_ON, we want NBD_PIR=0
Please remember that this is still only an assumption of the model, not an outcome.
Yeah, you are most probably right. As the PIR is a weighted sum somewhere between LW and SP, it should be the other way.Just a point there:
It is believed that the table is wrong: probably a clerical error that swapped the SP line and the PIR line.
Sure, that's what the model says, that's the assumption. But is that really true - is that really the optimum?I am not quite sure about the meaning of your sentence.
According to the model and ignoring LFX, if one wants to maximize the score then the target is to simultaneously
- maximize SM_PIR, the limit being 1 i.e. PIR is a line with a slope
- minimize NBD_PIR, NBD_ON the limit being 0 i.e. the PIR and the ON are perfectly FLAT line
Thats all I wrote.
Curiously enough, the Genelec 8361A doesn't seem as nice as its smaller siblings. It's not obvious what's wrong but it seems that as you make an enclosure bigger, the same roundover is not good enough anymore (or as effective) as for a smaller one. At least that would be my guess.