Best article explaining the differences stated in calibration levels:
The problems I have found most explain the 'loudness' people are experiencing is either in the mix itself (clipping), which is unfortunately NOT rare, or people without linear playback capability at high levels. Digital clipping is HARSH. it is not how a speaker distorts slowly, or an analog signal chain is overdriven. It goes from clean to crap very quickly. Tron:Legacy is horrible in this regard, which is why it 'sounds' louder.
Then there are amp/speaker problems.
Take your garden-variety 87dB 1W/1m sensitive, baffle-step corrected speaker. You sit 13 feet away. That's 87dB with one watt; instantly subtract 12dB for sitting 13 feet away, giving you 77dB. To get to 105dB (maximum dB RMS allowed on a soundtrack for LCRS channels), you would need 640 Watts. To listen at -10dBRef, you need 64 Watts, well within an AVR's capabilities. That's with zero equalization. If you use Audyssey, it can add up to 9dB of boost, taking away more headroom.
Now, mount those 87dB speakers flush in the wall. You gain 6dB instantly after rewiring the crossover, as the high frequencies no longer have to be attenuated (proper baffle step compensation was used). With 93dB sensitive speakers, now you only need 160 Watts to get to 105dB, minus headroom robbed by Audyssey, as it tries to correct 1st reflection point suckouts and the like.
You can now see why High Sensitivity speakers (95+dB 1W/1m) are almost necessary for reference playback. For LOW DISTORTION Reference Playback, you need Ultra High Sensitivity Speakers (>100dB 1w/1m). Why? If you can FIND speaker distortion graphs, they are taken at 1Watt. Distortion and power compression increase quickly as power is added. Even the most robust 87dB freestanding speaker will probably not sound good with 640 Watts in, nor will it actually give 105dB at 13 feet away (I am not counting boundary reinforcement at the moment, as it introduces more than just boost in the LF region). Even the high sensitivity designs will start to distort when fed over 100W, albeit they will be reaching much higher playback levels before sounding real bad. MKTheater runs the highest sensitivity LCR I know of. If he reports a mix is too loud, it is likely clipped, as his LCRS amp/speaker combos have lots of headroom, or it is a processor fault, leading to exaggerated levels.
When I measure films, I use MeyerSound's criteria. A 10% modulated sinewave will read -23dB when measured with RMS metering. That means 105dB RMS peaks, and 108dB instantaneous peaks.
In the bass, when all channels +LFE are driven at 0dBFS, you will get instantaneous peaks of 126dB at the listening position for 5.1, and 128dB for 7.1 (subtract 3dB for RMS peaks respectively). That is really powerful stuff. Hardly anyone has that kind of low distortion playback capability in an HT, and definitely not any movie theater I have ever listened to.
I use 93dB sensitive LCRS speakers. I can only playback at -7dBRef before the speakers start to sound bad, even with a 100W AVR. I sit 9 feet away. That's 93dB and subtract 8dB off the bat for distance, down to 85dB. Theoretically, 100W should get me to 105dB. But I use Audyssey, and it robs me of 7dB of headroom, limiting my loudest listening to -7dBRef before they distort more than I would like...Auto-EQ algorithms do not understand how much headroom they are taking most of the time..... With my Audyssey MultEQ XT, it applies cuts wherever it sees fit (sounds good, no headroom loss, right?). But then it applies an overall boost so that there will be no perceived volume change when Audyssey is switched on or off (correctly so). That boost robs you of headroom everywhere Audyssey DIDN'T apply a cut. Audyssey can boost up to 9dB.
Clean Reference is hard to reach (it is a tall order!). People like the folks at DIYSoundGroup and others have made approaching reference level playback much easier to attain with AVR power.
JSSEdited by maxmercy - 7/9/13 at 7:00pm