Enhancing false trigger detection on hard hits and rimshots

[daniele99]

Hi dmitri, i'm back!
Yesterday i was thinking about my only complain about MD snare triggering. Sometimes, but enough to be annoying, MD detects false hit hitting hard the snare skin or doing hard rimshots. Cause i don't know how the algorythm handle this shots i'm asking to you if it would be a nice idea to add 2 parameters and an "if" code line that tell to MD to ignore every medium/hard hit done after another very hard hit (above 90/100 midi velocity) in a 40-50ms mask. It's nearly impossible for even the fast drummer to need a flam so "fast". I believe that a hard flam doesn't have a hit distance <60-70ms.
In "maccaroni" Basic i'll write something like this:

X= threshold above we can classify a hit as HARD HIT (es. 90/100) (HrdHThre)
Y= settable mask time (es. 40/60 ms) (HrdHMskT)

if "VEL" is >X then ignore every hit for Y (ms)

Or you can incorporate a progressive math function to your algo modifing DynLevel and DynTime behaviour...
If i'm not clear enough, ask me please!

[dmitri]

This is what DynTime and DynLevel are meant to do - the harder first hit the higher threshold for a next hit is. How high this threshold is set depends on the first hit level and DynLevel. And this threshold is gradually reduced during DynTime.

[daniele99]

Ok, then i'm going to tweak some more with DynLevel. Not DynTime cause i'm at about 20 to 32 and i don't want to miss pressrolls. Another question: how do the PS algo detect the first zero crossing? Before or after the False triggering routine? and it is possible that due to interpolation limits the algo sometimes detects the 2nd zero crossing? I mean when the wave passes across the 0 volt is there an approximation range of detection or you always have a perfect 0 Volt sample reading?
If it's off topic I can move this question in another section.

[dmitri]

It is not zero crossing, it is Threshold crossing. For fast rolls it uses position of the first hit for all subsequent hits.

[daniele99]

I believed that PS depended from the time between first 0V sample and first 0V crossing (first half-wave wavelenght). Not so? Are you saying to me that depends from time between 1st above threshold sample to 1st below threshold sample? Am I right? And why you clustered fast rolls? To avoid too many CC data to process? So if I do a pressroll from center to edge and viceversa what i'd expect it will happen?
Too many question marks...sorry!

[dmitri]

MegaDrum ignores levels below Threshold, this is why it is Threshold crossing, not zero crossing.
Positional sensing becomes unpredictable if a hit follows a previous hit while a signal from the previous hit is still decaying. Hence fixing the position for fast rolls based on the first hit.

[daniele99]

Good news. Yesterday i fixed some dynlevel/dyntime parameters and i noticed that with this firmware i'm not loosing too much sensitivity. Good job dmitri. I tried to do pressrolls from center to edge and viceversa stating that PS works 90% well with light to medium hits. Very satisfying, near perfection. I'm wishing to make a video to show the results. Very rare false triggers on hard rimshots but negligible. I'm happy!
Only a strange phenomenon: when i play 3 or more hard sidesticks and suddenly try to do a press roll, often i have to pass a relativly hi threshold to restart triggering the head. Once reached, it starts to trigger normally. Any thought?

[dmitri]

I'm not really sure. Even after several seconds you need to hit the head hard to trigger it?

[daniele99]

Yes, no activity on snare head channel until i do a 50/70 midi velocity hit. Even after 2 or 3 seconds. But not always, apparently random. It doesn't happen on toms where PS is deactivated. Rim signal is not very hot (max 500) head signal is now with fixed voltage divider max 990. It doesn't happen after rimshots, only after sidesticks. I deactivated all crasstalk too.

[dmitri]

Is there a reliable way to reproduce it?
Is anybody else having the same issue?