08-28-2022, 12:10 PM
(08-27-2022, 08:22 AM)XORadmin Wrote: I can understand that E0 and D5 might behave slightly different even they (theoretically) shouldn't. But here is whats happening:
For E0, there is only one iteration needed and the next timestamp is calculated for when the gate should release again.
For D5, a timestamp is calculated for each tick and everytime it reaches the timestamp a counter is decreased until it is zero = release the gate.
Because of the scanning of the timestamps it might be that these times are not 100% the same. It might indeed be different with different BPM. The scanning (like sampling) rate is lose from the sequencer timing. (Just FYI, increasing the scanning time = less CPU power for all other future stuff, less smooth UI etc... And we are already talking about microseconds here)
If you want to hold a gate, then use a FE which is the only 100% trustable function that you would use for this usecase (keeping the gate on for X steps).
A step is indeed assumed to be 6 ticks (track clock divider/multiplers are included in this), but that is mainly because the triggers can be fired from everywhere at any time, even not step related. So this is the best assumption I can do.
I won't change anything here since I think it works well within the possibilities.
Agreed, that all makes sense, I do use FE and FD for this kind of thing but wanted to point it out if it was fixable.
Like you say I can see the difference, one is setting the gate end at the start, the other is analysing the state.