Hello Dmitri!
Of course I tried the configuration that you had suggested a while ago... It was working but sometimes I get a bell event followed by a bow event around 14 ms (as per MIDI Log Viewer)... Some other times if I hit the bell I get strong bow midi events.
However this never happens when I configure bow and bell on the same input as head / rim...
My feeling is that is I set the bow note to 0 (that's disabled) on the first input, Megadrum would not process that signal anymore in order to decide make the separation between bow and bell and relies for this on the Xtalk algoritm; furthermore maybe the scanning time and/or the Xtalk not being "strong" enough can have something to do with it.
If you think it's useful I can get some midi logs for the two scenarios to see the separation improvement when using the first input as bow/bell... In fact I think I could split the signals and use simultaneously the two configurations, but I'm not sure how the midi data is going to look like
...
Maybe we can identify by the input number what's supposed to be what...
My opinion is that if 3way Roland "algorithm" would read a voltage from the rim of the first input (instead of looking for noise reduction) when set to piezo in MegaDrum Manager (like Dual Or Yamaha option) but still expecting a switch in the second rim input (current 3Way Roland implementation) everything will work like charm...
Please let me know if my explanation was coherent enough
Thank you so much Dmitri!