Beta testing for a totally new sampling/triggering alg

[ignotus]

Was it with a Linear Curve?

Yes.

[dmitri]

Two more things: what were DynLevel and DynTime in the the double triggering screenshot and what was XTalkLevel on both inputs in the crosstalk screenshot?

[dmitri]

Anyway, updated the beta firmware in the first post.
Adjusted false triggering suppression (for lighter hits) and crosstalk suppression (making shorter but stronger).

[ignotus]

Sorry, I'm away from the kit, hopefully I'll get a chance to test it at the weekend.

[dmitri]

Updated the beta firmware in the first post.
Added an option to extend sampling period for pads with bad signal where second half wave produces higher signal than the first half wave. It is activated (for testing purposes) if MinScan is set to an odd number.

[dmitri]

Updated the beta firmware in the first post.
Fixed LCD normalized samples levels graph.

[dmitri]

Updated the beta firmware in the first post.
Dedicated a separate Menu page for the samples graph: it goes right after Firmware version page
Major changes (hopefully improvements) to XTalk suppression.

[ignotus]

Finally had time to give the latest beta a go. Xtalk is massively improved - I no longer have any problems with it. I still feel I had flams and pressure rolls working better before, but I think I need to redo my snare trigger so will wait to try that out.

[dmitri]

I no longer have any problems with it. I still feel I had flams and pressure rolls working better before, but I think I need to redo my snare trigger so will wait to try that out.

Are you using the same DynLevel/DynTime settings with the old and the new algorithms? If so, then for the new algorithm you can actually reduce both without negative affect. I have a mesh snare which produces bad signals and is very tricky to configure with the old algorithm but with the new algorithm it's pretty much perfect with DynLevel 4-8 and DynTime 20-32 - no double triggering and as far as I can judge all hits in pressrolls are registered.
Also make sure MinScan covers signals peaks (the samples graph is useful for this) - my bad snare has signals peaks at 5-6ms (50-60 MinScan).

[ignotus]

I re-did my snare trigger today and did some more testing. This may seem strange, but I can actually set DynLevel/DynTime lower with the old algorithm than with the beta. With the old one, I can set them at 1/4. With the beta, anything lower than 1/12 causes lots of extra unwanted notes. I can also get better flams and rolls with the old algo. Something I also noticed was behaviour with fast, hard hits on the mesh head - with the beta I get occasional low-velocity rim hits in between (but without hitting the rim), which makes it sound like there are dropped hits. This doesn't happen with the old one. If I try to dial them out with the 3rd zone parameters, I spoil the rimshot performance... And yes, I am positive that I am enabling/disabling the New Algorithm and Input Priority checkboxes to switch between versions.

It seems that my mesh pads are already well-tuned physically to avoid double triggering and for some reason the beta doesn't seem to agree with them.