The coalescer on the old Poseidon compressor is elementary. However, it can be made to work if a back pressure valve set to 2400 psi is mounted to the can itself. The rest will take care of itself, hopefully.
Yes, at, for example, 150 bar back pressure the dwell is 150 times atmospheric. However, 150 times nothing is next to nothing. The air space inside a chemical cartridge is about 1 or 2 cubic inches. The rest of the space is occupied by solid granules. With a 4 cfm compressor this would equate to 2 or 3 seconds dwell. It might help a little but it would require care to measure the improvement. Using a precision scale, I can see a difference at about 15 seconds but it is small.
As the SCUBA tank fills, the back pressure in the final filter increases anyway so the overall dwell advantage of locating the BP valve at the final filter becomes even smaller. The same could be said for the opposite arrangement so that is why I say the differences could be debatable. As to the BP valve at the coalescer and ejection of water, this refers to prevention of residual water spewing out of the coalescer as well as immediate improvement on condensation by raising pressure immediately, eg especially when the compressor is started. WRT the Poseidon system, I really believe that mounting the valve at the coalescer will help with this. I said that I added a final filter to the old Capitano and that this resulted in a big improvement, but not just due to the chemical media in the filter. I noticed that the amount of water draining from the coalescer (identical to the Poseidon) suddenly increased, more water drained after I moved the BP valve to the coalescer from its original location on the filter. I'm not sure why. It could be due to the slightly higher pressure setting I had dialed into the valve or some other factor such as I've described. In any case, not one drop of water has ever drained from the water trap on the bottom of the final filter or any of the other similar installations which I've done. The SCUBA tanks look great.