Kevin, Depends on the material that is being processed (bass management, not surround processing). With discrete multi-channel sources, full range content in each channel exists in the recording. Where ever you set the crossover for each channel, the frequencies below that point should be moved to the subwoofer output. You can set your subwoofer's crossover as high as you want, but the only content to reach the sub should be the frequencies shaved off the other channels. So if you're crossing over your mains at 40Hz, then the subwoofer output should only be putting out content below 40Hz, whether the subwoofer's crossover is set at 40Hz or 400Hz. For there to be bass "doubling", the processor would have to actually generate some duplicate bass content.

With matrix derived sources, it's a different story. Here the processor IS generating additional channels (including the sub). Mind you, the contents of these channels are 100% from the original 2-channel recording (i.e., no generated sounds, echos or reverbs), but the channels themselves are generated by the matrix decoding process.

I used to think that modern matrix decoders (like Neo:6 and PL II) used to first generate 5 full range channels, which would then be bass managed into a 5.1 speaker set-up. However, it turned out to work a little differently than that. In receivers and processor I've had experience with, the matrix decoders did not generated full range signals for every channel unless they were all set to 'large'. If any of them were set to 'small', they got a derived signal that only contained sounds above the crossover point. It's not like that channel was full-range content that existed discretely on the recording. Since it this derived channel never had full-range content, there's no bass to move around. Each channel gets only what's above the crossover point.

So what comes out of the subwoofer output? It's another generated channel, made by summing the original 2 channels of the recording. As with the other derived channels, it will only receive content based on where its crossover is set. There's no need to take frequencies above this crossover point and move them to the main channels because there were no sounds above that crossover point to begin with; i.e., this isn't discrete content from the original recording, like a .1 LFE channel is.

In both cases, I think it's best to set the subwoofer crossover to the same setting as the highest crossover point of the other channels. If the processor is performing bass management correctly, you shouldn't be getting any duplication.

Best,
Sanjay