Holy crap Audyssey!

I've been doing some more measuring and tweaking over the last couple weeks and was trying to get Audyssey to integrate my sub with my mains the way I wanted (I hate to admit it but I've never been able to get the crossover quite "right"). After a bunch of trials I was able to get it to set the crossover to 90hz which wasn't really where I wanted it but it was the closest I'd gotten so I wanted to give it a listen (in the previous 7 runs I was always getting either Full range or 150hz )

I threw on some midbass heavy music and gave a listen... unimpressive as I expected (the IB isn't great for a 90hz crossover, the TD15M's in the mains are much more suited for midbass). Then I threw "Bass, I Love You" at it and turned it up to what should have been a moderate level, like -20 (I've been able to run this song at +10dB before without issue). Song starts out with the higher bass stuff then the 7hz tone kicks in and the EP2500 clips HARD into the IB subs that are throwing themselves around lol.

I decided to measure and see what Audyssey had done to my subwoofer. Now note, in my effort to get Audyssey to set the XO to the correct frequency I had inadvertently given it a bit of a mess to start out with but nothing like this!

Light blue is where I got the hard clipping. I EQ'd that down to the gold trace, messed with the high pass on the woofer in the mains and ran again (this time got a 70hz XO = much better!). Green is the result of that Audyssey run. Dark blue is the final results. With the green trace I redid the Bass, I Love You test and although it was pretty safe I felt it was necessary to throw a HPF on there just to be sure - especially after the hard clipping incident!

Thanks for the input Mark! I suspect the same



That said you can see it wasn't just me messing with things though. The 2nd pass of Audyssey took the very decent starting point of the gold trace (that was a 60hz crossover) and made it the green one throwing away 10dB of headroom.


Trying to get Audyssey to cross where I wanted I was putting different high pass filters on my mains (passive TM and active W using DCX2496). It got a bit more complicated though as I was doing EQ in phases - I ran audyssey then measured the preouts to see what Audyssey had done and tried to mimic that with filters in the DCX (where the filters were very close for both L&R speakers). I did the same thing with the subwoofer but it was a very "odd" filter that Audyssey had used. It had made a large wide cut centered at 40hz then increased the overall level by like 15dB (something like 40hz, -15dB, wide Q, I don't remember at all what it was). When I got the first result above (light blue trace with 10hz at like +30dB) I had that filter still set in the DCX. It looks like it basically did the same thing again - made cuts, etc then increased the overall level by 15dB resulting in the +30dB.


Darn! I should have measured the sub preout to see the actual signal it was sending the sub! Really it was supposed to be a quick re-run of Audyssey after getting the speakers "closer" but it turned into something very amusing and I just had to share

Regarding changing the crossover after Audyssey has run. For some reason I thought Audyssey could/would help get the phase properly aligned through the crossover it selects but might not necessarily have corrected it for a different crossover frequency. I could certainly be way off base though as that was just an assumption I made.

One good thing did come out of all this, I had a major "Duh!" moment. I'd never been able to get the crossover right when using the built-in XO on my AVR but Audyssey seemed to have no issues. To get Audyssey to set it right I had put a 70hz 12dB/octave LR HPF on the woofers in the mains. After Audyssey did its thing and I measured to find the integration pretty good I tried removing that HPF to see if I could get that hole at 90hz to fill in some. I found a peak and null at the crossover frequency now - similar to what I've always ended up with when using the build-in XO. I didn't think much of it last ngiht but this morning it hit me. The XO in the AVR is usually a 4th order LPF on the sub and 2nd order HPF on the mains. The 2nd order HPF is supposed to sum with the natural 2nd order HPF that the mains are supposed to have to provide the matching 4th order of the sub XO. Because my mains remain flat lower than where I was trying to cross using the XO in the AVR without an additional HPF on them was resulting in asymmetrical XO slopes giving me the nastiness I always found at the XO.

This is what I've read in the past as well.

Not before the light blue trace but for the second run Gold was pre-audyssey, green is post-audyssey and dark blue is after I EQ'd the green trace.

I measured the preouts with Audyssey off and then on this evening that I'll post later. I did it while on the phone with a friend though and the dB scale is totally messed up? I didn't have time to mess with it though and I believe although the labels are wrong the results are accurate.

Edit, nevermind, the laptop rebooted and I lost the measurements anyway. I'll have to do it again but basically the Audyssey off trace was mostly flat as expected. Between 6hz and 20hz the Audyssey signal was +6dB as compared to the audyssey off signal. They cross over around 20hz and above that the audyssey signal was -4dB or so as compared to audyssey off. But then around 80hz there was a big bump with the Audyssey signal going back +5dB or so.

What I suspect is going on is Audyssey isn't EQ'ing below 20hz (I believe I've read that it's not supposed to) and isn't doing any boosting (again I believe I've read that it's not supposed to). In order to make it measure flat it needed +10dB at 80hz so it cut everything else down to 20hz then raised the subwoofer signal by 10dB total resulting in an effective +10dB boost for everything below 20hz. I had similar suspicions in the past that it was doing this but didn't have all the measurement gear to check it out (or more accurately didn't think of it lol).

I suppose I should include that. Onkyo PR-SC885 is the pre, not sure what version of Audyssey it includes but it is quite dated (no DynamicEQ). I still need to rerun the measurement of the subwoofer preout to show what Audyssey did to the sub signal too, just haven't gotten around to it yet. It was quite interesting

Interesting, looks like I'm not the only one who's measured this phenomenon recently. From the subwoofer section thread on the Submersive:

http://www.avsforum.com/t/759877/seaton-sound-submersive1/6840#post_22100727

Got a chance tonight to do some measurement of the sub preout with Audyssey On and off:



Magenta is Audyssey on, Green is Audyssey off, Red is Audyssey off with the MV up 6dB

This is an older version so no dyn EQ, Audyssey MultEQ XT.