Originally Posted by WLC
I believe I noticed a significant improvement in my calibration after level matching the subs at the exact volume at which the calibration was to be performed. The technical support, Anthem and SVS, both responded to a different question than what I asked. However, an electrical engineer told me that my suspicion is correct and when you change the volume, the phase also changes. I don't completely understand this technical answer, but I have included a REW graph showing my 4 subs at two different volume levels.
As you can see, by changing the subs only 5 db , there is a variation in the lines. At 41.1hz, there is a 5.3db difference, at 60.2hz a 4.9db difference and at 71.1hz a 4.1db difference. Wouldn't this 1.2db variation mean that there would be two different sets of corrections? If you level matched at one level and then changed the volume by 5 db before calibrating, wouldn't the calibration system be correcting the wrong information?
I believe this issue would exist with any calibration system and multiple subs.
What you are saying makes sense to me. There is more than one purpose involved in the level-matching process that occurs in an auto-calibration routine. First, of course, we simply want to have all of the channels playing the same volume at the MLP, so that the various speaker and subwoofer levels will sound balanced with respect to each other. But second, starting with all of the channels playing at the same volume level is probably the best way for the automated software to attempt to EQ for a flat frequency response. (This is addressed in some detail in Section II of the Guide.)
As far as I know, all of the systems of automated room EQ operate more-or-less in this way. First, they should typically attempt to level-match all of the channels to the same volume level (which in today's AVR's is typically 75db) and only after the channels are level-matched are any correction filters applied to those channels. (Does ARC not attempt to level-match the subs to the target SPL of ~75db?) Once the channels are all level-matched, correction filters are applied to minimize upward and downward deviations, at any frequencies, from the target SPL of ~75db.
After level-matching, it stands to reason that we wouldn't want to change the volume level of any channels, including the subwoofers, between the level-matching process and the auto-calibration process, although after
calibration is complete, we can change any channel levels we want without changing the EQ filters that were set. We will obviously change the acoustic balance of our HT system, by increasing the volume of our subwoofers or our center channel, for instance, in accordance with our own listening preferences. But, increasing the volume of the center channel as a whole, or increasing the volume of our subwoofers as a whole, won't change the filters that try to promote an overall fairly flat frequency response.
Where things may seem to get a little more complicated at first is when we deliberately introduce PEQ changes at specific frequencies after
an auto-calibration routine. That might be done by introducing a house curve from a specific frequency downward, or something of that nature. But, that is frequently done in accordance with measurements of our frequency response, with REW or something similar, so that we can see as well as hear the effects of the PEQ tweaks on our overall frequency response. And, where we are doing it based solely on our own listening preferences, we are doing it because it actually does sound better to us, and we are still leaving the rest of the EQ filters intact.
Frankly, I'm not sure why someone would want to deviate from having all of the channels level-matched before
running auto-calibration. Although I'm not sure that I fully understand the phase cancellation that's occurring in the graph, I can't imagine that the auto-calibration routine could perform its job as effectively if all of the channels weren't playing at approximately the same volume, as measured at the MLP.
Edit: Just to be clear, the issue being described in the post that follows this one, seems to be limited to ARC Genesis. It didn't seem to be a function of previous versions of ARC, and I know that is not the way that any version of Audyssey, or Dirac, or YPAO operates. The two-step process that WLC is describing is certainly something that ARC Genesis users should be aware of. Since so much of Genesis has changed/updated in just the last few months, it's hard to know if the need to level-match again, after being told to turn down the subwoofers, is a software glitch or not. If so, perhaps it will be addressed in future software updates. But, if the final result of level-matching to 75db again, after turning down the sub gains works, then that's all that really matters.