Like Rod said, you can't send on 16 channels at once, but you could send on 4 or so! Anyway, here's the setup I use to sequence a song while listening to the parts I've already sequenced.
I start with a 4 zone performance in master keyboard mode. In the master Utility mode I set Midi receive channels to "omni" so all the channels are received, and transmit to channel 1.
In the 4 zone performance edit menus, hit the "A" key to edit zone 1. Set it to Midi = "on" and and TG = "off". This separates the CS's tone generator from the keyboard, so pressing a key isn't sent on to the tone generator, instead it goes out on midi 1 (or you can set it to some other channel in the Zone 1 menu) to your sequencer. Make sure that the "layer" command for parts 1-16) is set to "off". Now depending on the sequencer you may have to set the send channel to the receive channel (1-16) that you are working on at the moment (eg. working on moog on voice 2, set Zone 1 send channel to 2). Then, as you work, just hit record on whichever channel you are sequencing on and make sure your zone is sending on that channel, and start playing, the software will send the midi back on that channel and the board will make noise!
Now something that you'll notice is that some voices don't have the depth/effects they had in voice mode. This is because performances have their own 5 band EQ settings that affect the sound in Perf mode, and only 1 voice can have insert effects in Perf mode (yeah, it's the thing everybody hates about this synth, Why Yammy, Why?). To get around this I set the insert effects to whichever part I'm working on at the moment in the "Common" menu of the performance, and then record the audio of each part in voice mode when the sequence is finished.
-Casey