If you’re willing to experiment, try replacing one with EPDM and see if it makes a difference.
]I was just holding a ruler to my screen and expanding the graph to get a good measurement on a tablet. That’s why I was using lengths instead of pixels. I’d use GIMP if I was at my computer.
It is! My measurements say it’s 1.40 hours. 
OK, so now I think I better understand what you meant earlier. The difference between 1.4 hours and 1.33 hours is only 4.2 minutes, so I’m guessing it’s either actually one number maybe somewhere in-between, or if they are different the numbers are so close that for all practical purposes they might as well be one number.
So…If the the initial water load is different but the regeneration times are the same (i.e. some predetermined, single fixed number of minutes), then 1. what are the implications of that, and 2. does whatever number Sunlu picked as the fixed amount of regeneration time limit how dry this thing can ultimately get? i.e.Does it represent some kind of compromise as you were suggesting.
All good questions. However, I still tend to think the governing limit is is likely to be how high the i10 allows the humidity to go before initiating a regeneration cycle, because even if the desiccant were perfectly dried during regeneration, the cabinet humidity will go that high before the i10 does anything about it. On the other hand, I suppose the average humidity inside the cabinet could be lower if the regeneration finished with a drier desiccant bed. But… if the cost of attaining that drier desiccant bed is a longer regeneration time, as seems likely, then is the resulting longer unprotected exposure inside the cabinet to the high humidity released during regeneration ultimately self-defeating? So, perhaps that’s yet another trade-off you were referring to. That’s a tricky one to figure out.
I kinda doubt Sunlu thought about those considerations as hard as we are. More likely as soon as they got promising results it was a code freeze with pressure to ship it ASAP. But that’s just me being jaded. It doesn’t answer the questions, which are still valid.