Music in the Round #50 Page 3
I concluded my July 2011 column confused about my inability to get a satisfactory MultEQ Pro calibration with Integra's DHC-80.2 A/V controller ($2300). MultEQ XT32 room correction with the Integra's microphone was just fine, but the MultEQ Pro correction with the individually calibrated microphone just sounded wrong. I had focused on identifiable problems with the cumulative response of my two subwoofers, but I also found that, even with a bit of help from MultEQ Pro's curve editor, the upper midrange was glassy (ie, hard and too prominent), the treble dull. No matter what I tried, plain-vanilla MultEQ XT32 sounded much better.
I contacted Integra and Audyssey about this, and both swiftly responded to the draft version of my July column. (I suspect they talked to each other as well.) Shortly thereafter, Chris Kyriakakis of Audyssey told me that the current Integra firmware had an error in it. Apparently, someone at Integra "forgot" to set a flag to tell the DHC-80.2 to use the individually calibrated mike curve when in MultEQ Pro mode. As a result, the processor applied "the consumer mic correction curve to the Audyssey Pro mic!!!! No wonder you are hearing audible differences!" Although Audyssey had approved the original factory firmware, it was only after reading my column that they realized that the latest firmware (11/10/10) was faulty. Not only that, the same error occurred in the contemporaneous update for the Onkyo PR-SC5508 and, perhaps, similar models.
As soon as Integra had identified the problem, they promised a speedy fix, and announced a firmware update on May 27, 2011. The website listing this update says only that it "Improves functionality of presetting Internet Radio in WEB Setup Menu." Indeed, my Integra contact initially told me that "my problem" would be addressed soon. That really annoyed me, if indeed all that had been lacking was for someone to set a flag at code compilation time! As Chris Kyriakakis says, "Computers have officially taken over . . ."
However, I was later informed that the update released actually did invoke the proper calibration curve in MultEQ Pro, and that I could confirm this for myself. So on the final weekend before the deadline for the column you're now reading, and before the handoff of the KEF Q900 speakers (reviewed elsewhere in this issue) to John Atkinson for measurement, I began yet another MultEQ Pro calibration with the Integra. I needed to do this before the KEFs disappeared so that I could compare the new results with the old, spurious ones. I found subtle variations in levels and distances, but striking differences in the Audyssey "before" graphs. Equally striking were the similarities between the "after" graphs, even though my ears told me otherwise.
The "before" difference makes sense. If the frequency response (FR) of the system and room is a constant, any measurement differences would be a function of the FR of the microphones. With Audyssey's Pro mike and the proper calibration curve (fig.1), the measured FR should be pretty accurate, and I confirmed that with the Omnimic system (review underway). With the Pro mike and the mismatched calibration curve (fig.3), the measured FR should be incorrect, and it was. Note that the accurate curve (fig.1) reveals a wider swing in FR below 100Hz, a deeper trough just below 120Hz, and a slow, progressive rolloff above 1kHz. In particular, compare this with the earlier graph's steep high-frequency rise beginning at about 7500Hz (fig.3).
The fact that the "after" curves are nearly identical (figs.2 and 4) also makes sense. First, MultEQ Pro does attempt to achieve the same target FR, regardless of the original measurements. Second, the "after" curves aren't based on direct measurement, but on the application of the calculated corrections to the original measurements. If those corrections are accurate, fine (fig.2). If the measurement with the mismatched mike corrections (fig.3) is spurious, then the corrections based on it must be spurious and, consequently, the nice, smooth, flat "after" graph (fig.4) must be equally spurious. Although the graph trace in fig.4 is relatively flat, the actual response would have a peak at 50Hz, a dip at 120Hz, and impose excessive rolloff above 1kHz!
Similarly, the subwoofer "before" measurements varied with the earlier, incorrect graph, displaying much wider FR variations than the later, correct graph. That would result in overcorrection despite, again, similar "after" graphs. No wonder I couldn't fix the sound by fiddling with adjusting delays for the individual subs.
After all this, the results with the Integra DHC-80.2 using MultEQ Pro were audibly superior in several ways to those with the Integra's built-in MultEQ XT32: in HF smoothness; in deep, detailed bass from the range of listening positions within the measurement envelope; and in a very well-balanced overall response. They were as different from the earlier, incorrect calibration as from not using Audyssey at all, though of course in different ways. All of my commendations of the DHC-80.2 now extend further, to include Pro's more precise calibration. In addition, I strongly recommend that all users of the Integra DHC-80.2 and of Onkyo's PR-SC5508 install the latest firmware and rerun Audyssey MultEQ Pro. The results may greatly surprise you.
But two cautions: First, don't presume that every update will create an improvement for you and your system. Trust your ears and your intelligence: If it doesn't sound right to youand you know your system better than anyone else canit probably isn't. Second, take Audyssey's "after" graphs with a grain of salt. Spock-like, Audyssey has always strongly stated that their "after" graphs must be reliable predictions of in-room results because they're based on good measurements and the mathematics of acoustics. But what if the measurements aren't good? Again, trust your ears.