www.megadrum.info

As you may have noticed I lately regularly have been making changes to the alternative false triggering suppression algorithm (AltFalseTrSupp). I just published the latest, 20100307, firmware version in which I made major changes to this algorithm. I'd like to think that the changes are very positive.
Can I ask everyone who is interested to test the AltFalseTrSupp algorithm and report the results and in particular in these tests:
1. Soft, medium, hard fast rolls.
2. Pressed rolls.
3. Free falling bouncing stick.
4. Single soft, medium, hard hits.

A few pointers when configuring pads for this algorithm:
1. Of course, set AltFalseTrSupp to Yes.
2. Set Retrigger to 2.
3. HighLevel should be properly set before doing tests.
4. The various pads I did tests with all needed different combinations of DynLevel and DynTyme. Start with DynLevel=4 and DynTime=32. In general, the lower both values the easier rolls are but more likely false triggering will happen and vice versa. DynLevel plays more role with hard hits, DynTime plays more role with soft hits.
5. You may have to raise MinScan to ~50-60 to avoid false double triggering (one hit but 2 (or more) notes with close velocity levels)

Thank you.

I also should mention that these changes are meant mostly for mesh head pads since other types of pads don't suffer from false triggering as much (less after hit ringing) and easier to configure and I believe that pressed rolls only possible with the mesh head pads.

dmitri wrote:I also should mention that these changes are meant mostly for mesh head pads since other types of pads don't suffer from false triggering as much (less after hit ringing) and easier to configure and I believe that pressed rolls only possible with the mesh head pads.

My hardboard Simmons stuff did pressed rolls to some extent?
Are you using cones?

I'm using a Pintech mesh head snare.

12" single ply I guess - I mention this because I have found it very hard to cope with even minor deviations in pad/sensor design.

The way I see your code is a parametric low pass filter stuck on top of a demuxer.
The demux stuff feels great and I'd try to differentiate that code from further developments as far as possible.

Done correctly (as you will) I'm going to suggest that positional sensing could give more scope for general development because you would be moving toward the frequency domain more and more.

Not a surprise but again I'm struggling to understand what you are talking about.

I just wondered if your ADC input channel sampling routine is separated
from these other code changes ...