The surround channels most definitely need to be attenuated by 3 dB in order to provide proper balance. I explained why in a post in the other DTS thread, which I will copy/paste into here:
Theatrical sound systems are calibrated such that a -20 dBFS pink noise tone will play at 85 dBSPL from the front left, right, and center channels, while the rear channels measure at 82 dB. This is to maintain backwards compatibility with Dolby Stereo analog tracks, which have a monaural surround capability rather than stereo rears. In Dolby Stereo, the entire array of surround speakers is treated as one unit, which together measures at 85 dB. So when 5.1 came along, they essentially split the existing surround in half, with the left and right each playing the same signal at 82 dB instead. When heard together, the combined result of all the surround speakers is at 85 dB as it should be.
What this means is that because the surround channels in movie theaters and post-production stages are deliberately set to play back at a lower level than the front channels, the content of the surrounds will end up being 3 dB louder than it otherwise would be. The mixers will set the level of the surround effects to sound 'correct' to them on the system they are listening on. When this is played on home theater systems, which are calibrated with all channels set equally to 85 dB, the surrounds will end up being 3 dB too loud.
Because of this, a theatrical mix played on a home system must have the rear channels lowered by 3 dB in order to sound the way the mixer originally heard it. 5.1 encoders typically have a selectable option to do this automatically when creating the files for home use, but this was not always the case.
Note that calibrating the surround channels at 82 dB instead of 85 applies to all theatrical 5.1 formats, not just DTS. Playing the same signal out of each surround results in a doubling of the acoustic power, which increases the total level of a mono signal by 3 dB. It's the same thing that happens with the phantom center in a stereo mix. Double power results in a 3 dB boost, but this must be distinguished from doubling voltage, which is what occurs when summing electrically—this creates a level increase of 6 dB for phase coherent signals.
So the surround channel calibration is easily explained when looked at in this manner. When it comes to the LFE channel, however, things become considerably more complicated. I tried to figure it out a while back and couldn't quite arrive at a satisfactory explanation for everything in the DTS system manuals, because we have no idea exactly what the analog chain inside their hardware actually is. I am going to go back and open my Pro Tools session for the '97 SE DTS again, so I can take another shot at giving you guys an exact method to reproduce the LFE calibration and crossover.