System:
X3700H into two Neurochrome686 Power Amps (20db gain) into two KEF R700
(Speaker Measurements:
KEF R700 loudspeaker Measurements | Stereophile.com )
Room: TV area, untreated (very undampend)
5.6x4.2x2.6m (L,W,H), MLP Symmetric at 62% from front wall.
Speakers 0.85m from sidewall, 1.35m front wall (60deg and ca.2.6m away from MLP).
I read about the 686 and noticed that the gain is adjustable. Just curious about why you adjusted it to 20 dB. I would think 26 dB, or at least 23 dB should be better for the Denon AVR even if you use preamp mode only. The decrease in noise at the minimum gain is not likely going to be audible.
Questions:
to step 1): Am I right to assume one uses the Audyssey On FR response to base the ratbuddyssey corrections on?
to step2): Given the above L FR example (Audyssey On, green line), there is e.g. a peak at 31.4Hz, 75.74dB. Is that the one to pick for example?
top step3to5): Which target SPL to pick? 73dB maybe? If yes would the correction than be at 31.4Hz, 73-75.74 = -2.74db to entered into ratbuddyseey? Here I am really unsure.
First of all, 1/12 smoothing is fine to show the results but for determining the cuts/boosts for use with the Rat/App, I would use no smoothing for the 10 to 300 Hz range and may be 1/24, even 1/48 for the higher frequencies.
In my opinion, it is best to use the Audyssey reference curve to base the Rat
"corrections on". I put that in quote because I want to be clear that Rat does not correct, it is just an UI to make it easier to make manual adjustments than to draw the so called custom curve by fingers of stylus pens. Audyssey is always responsible for calculating and creating the filters.
I think by entering the cuts using Rat (or use fingers with the App without Rat), I would basically be providing feedback to Audyssey that would enable to it to make corrections, sort of like how amplifier's negative feedback scheme work. The difference is, amps use negative feedback to lower distortions in the output voltage whereas Audysssey might use it this way to lower FR distortions. I have had some lengthy discussions with Audyssey recently, and the person who I discussed with did not agree or disagree with my reasoning, but did not/obviously could not as I showed the results either; and he/she suggested I do certain tests for him/her but I have yet to time to do it.
So if I am right about the "my reasoning" of why/how doing it with the manual adjustments to the reference curve would work/or improve (depending on the goal), the customized curve is simply a way for Audyssey to improve on the effectiveness of filter set based on the feedback in the form of the manual adjustments, i.e. cuts and boosts entered via the app.
I feel that there is no point to go back to the REW curve with Audyssey off but start with the Audyssey corrected curve, that is, the reference curve. That is, for the first trial, or any further subsequent trials,
I would always go back to the original reference curve. For example, if after applying a cut to 100 Hz by 2 dB and the resulting REW curve shows you should apply another 1 dB at the same or nearby frequency, say 105 Hz, then I would not try to manually apply 1 dB at 105 Hz but instead of keep chasing a moving target, I would simply apply another 1 dB cut to 100 Hz. That is, use Rat to change the 2 dB cut I made in the first trial to 3 dB cut instead. To keep chasing, that is, use the new REW graph to base the new manual adjustments via the App/Rat will be extremely time consuming and hard to keep track. I am not sure if it would do a better job, but I am sure doing it my way is easier, quicker and has done a good enough job for me. Now, if after the first trial you noticed a new peak/dip appeared at frequency points a little far away from any of the anchor points used, then yes I would apply cuts/boosts to those new frequency points, but for the magnitude of the cuts/boosts, I would still reference back to the original reference curve. For example, if you did not see the need to apply 3 dB cut to 300 Hz on the original reference curve, but after the first trial you see the need at 300 Hz, then I would entered - dB at 300 Hz and see what happens first.
Generally: Do you try to optimize the individual L, R channel or the L&R joint FR?
Yes I have tried L, R, L+R, L+S, R+S and L+R+S. They all would work but obviously L+R+S (I found 80 Hz XO was always best for my setup) is the easiest and quickest way to do. If the result is good, and it was for me, then I would not have bothered doing anything else. Unfortunately, I did not know that until I had tried them all. So I ended up spending hours and hours on trying all combinations.
My suggestion is, try and focus on optimizing L+R+S first. If you can get good results such as +/- 3 dB peak to peak 20 to 200 Hz, then just leave it alone and enjoy the music. I am quite confident that if your REW curve for L+R+S is that good, then L, R, L+R (set to large) will also look good, may not be best but will still be very good.
Some thoughts: As one can see the joint L&R response is worse than the individual response. For example, I am assuming when looking at the pre-Audyessy data, that Audyessy reduced one channel at around 53Hz in order to flatten the response of that one (right channel), however as the left channel has a node here one could try by boosting the right channel back up to compensate for the node of the left channel at that frequency. Especially in the range where one can not locate the sound (below ca. 100Hz). Of course this would unflatten the right channel but the joint L&R resonse would be flatter? Or am I overthinking this, as we are usually to stereo and not mono?
It would be easy to say yes you are overthinking, but I don't think it is because I only realized I might have been overthinking after the fact, that I have spent so much time and plotted numerous curves, to the point I thought I was going to need to buy another Umik-1 mic soon before it fails. As I mentioned earlier, I would not keep chasing the new "anchor points" for the manually entered cuts/boosts, just stick to those identified in the original reference curve. If you have time and are as crazy as I am, then go ahead and try it once or twice to see if it would improve anything, but you have to know when to stop chasing eventually.. Remember, we are humans, a couple dB peaks and dips here and there will likely have no audible effects especially if the peaks and dips are narrow.
Sorry for the lengthy first time post, but I thought I post as much background as I have momentarily. Again, I would be very happy to follow any advise / input you might have.
Any time, I do have one suggestion, as soon as you get the result that meet you target, try plotting curves away from the mlp, and or aiming the mic at different angles to the speakers, and do it again for Audyssey off vs ref.
I have tried up to 9 positions, mlp, +/- 10 to 12 inches to the left, right, above and below so that I can compare off to ref for each of the positions, and for the average curves (also off vs ref).[/QUOTE]
