Originally Posted by
RythmikAudio
Well I don't know how robust the EQ algorithm is. I deal with software that affects millions of dollars of design works every day. In that case, there is always a verification step that uses an independent type of algorithm to verify the design work. That can catch a lot of problems such as software not able to handle certain combination of data, errors in the software... etc. Unfortunately, in EQ, you don't get that. Therefore common sense is very important. If you xover at 80hz and that is a wavelength of 12ft. So any program puts a 6 ft extra is literally inverting the phase on the sub. What I would do is I will wire the subwoofer output of phase (if there is a switch on the plate amp) and now see the results from the program. If the algorithm performs consistently, it should now remove that 6 ft distance. If not, then the problem it tries to resolve is not phase problem.
EDIT: another issue is digital latency. This applies to all digital process not symmetrically applied to all channels. My Denon 4802 has a very bad latency (defined as the delay between signal in and signal out) on subwoofer channcel as compared to other channels when I use analog mode. The reason is in their design, there is no analog domain filtering on sub. It is all done in digital domain. So it needs to convert analog to digital, then digital filtering, then digital to analog, not to mention a couple of cycles to do digital FIFO bufferring to avoid overrun errors. SO it is possible that the distance is for this. BTW, the sampling rate for subwoofer is not same as the one for front channels.
Another point is, the plots I give you are all based on close-mic measurement with the least amount of interference. When you place mic at listening position, the room mode will blur phase information and making it more difficult to extract the exact phase relation. In my view, a close-mic measurements should be very important part of EQ calibration process. With that as reference, the program would know how much is contributed to subwoofers itself, and how much is to the room. These two should be handled differently.
On a different subject, I would like to say EQ and servo IS complimentary. I one time asked the designer of Audyssey about how his program impacts the basic quality of the speaker system itself. I didn't get an answer. We all know the answer. If we think of audio components such as pre-amp, power-amp, speaker as individual components in a chain and EQ is part of it, these components cannot correct the mistake/error/distortion made by other components as there is no close loop to feedback the error back to EQ in real time. One example is, if we have a ringing in the woofer at 1000hz, one may say we just EQ it out. However, most people forget that the distortion in speaker is generated at the very last stage of this chain. EQ can control the signal passing through it, but it cannot control the distortion happen at the end of the chain. For those who reads distortion plots all the time would know, if we have a 1000hz ringing, we would get a bump in 2nd order distortion number at 500hz (because 500x2=1000), and a bump in 3rd order distortion number at 333hz (333x3=1000). The problem is still there. EQ solves part of the problem. I also think making the frequency response as perfect as possible (such as getting a very flat close-mic response from speakers and good speaker location) can make EQ algorithm easier to find an optimal solution.