Benchmarks

[Synthex]

Thanks.
I will test it this night !

[Synthex]

Synthex, try these two versions for 22 inputs

Yes, your new versions are much better !
The response time varies between 3 and 6 ms !

What changes have you made ?

Good job !

[dmitri]

Corrected a semi-error and did some code optimization:)

[dmitri]

I've updated all firmware versions on the Downloads page with corrected versions.

[Synthex]

You could increase the version number on the firmware name, for example: megadrum32_16_v10XX.hex ?
To be sure to be up to date.

[Marctwo]

Interesting. I never realized that these devices suffered such large delays.

How much of this delay is in the software? I suppose, pretty much all of it as the midi signal would start when you write to the output?

I'm currently working on the firmware for my own device and I'm working with software delays of ~0.5ms due only to the nature of the piezo signal. I need to test with a wider variety of pads but so far I'm getting very reliable results.

Are you using a complex sampling algorithm?

[Synthex]

Interesting. I never realized that these devices suffered such large delays.

How much of this delay is in the software? I suppose, pretty much all of it as the midi signal would start when you write to the output?

I'm currently working on the firmware for my own device and I'm working with software delays of ~0.5ms due only to the nature of the piezo signal. I need to test with a wider variety of pads but so far I'm getting very reliable results.

Are you using a complex sampling algorithm?

On what chip do you work ?
I am not on that delays of ~ 0.5ms be realistic.
Even the best professional modules are 2.5ms ...

[dmitri]

Interesting. I never realized that these devices suffered such large delays.

How much of this delay is in the software? I suppose, pretty much all of it as the midi signal would start when you write to the output?

I'm currently working on the firmware for my own device and I'm working with software delays of ~0.5ms due only to the nature of the piezo signal. I need to test with a wider variety of pads but so far I'm getting very reliable results.

Are you using a complex sampling algorithm?

0.5ms is not even half wave of a piezo signal:) I must say you found a much better algorithm so you'd better not spoil yours with others. Good luck with that!

[Lakedaemon]

0.5ms is not even half wave of a piezo signal:) I must say you found a much better algorithm so you'd better not spoil yours with others. Good luck with that!

It depends of the kind of piezo trigger you use :
20mm piezo transducer vibrate at 6.6Khz -> 3.3 waves for each 0.5ms
27mm piezo transducer (Roland Pad meshhead trigger) vibrate at 4.6khz -> 2.3 waves for each 0.5ms
35mm piezo transducer (Roland Pad Rim trigger) vibrate at 2.9 khz. -> 1.45 waves for each 0.5ms

to have such a low delay, I would guess that his triggering algorithm is pretty straightforward (i.e. very simple and quick, with probably very few influence of the past signals)

With a very simple triggering detection algorythm (i.e either you are over a threshold, or you are under) and with 16mips, it should be possible to have only 0.5ms of delay (yet...probably without crosstalk/retrigger rejection)

[dmitri]

0.5ms is not even half wave of a piezo signal:) I must say you found a much better algorithm so you'd better not spoil yours with others. Good luck with that!

It depends of the kind of piezo trigger you use :
20mm piezo transducer vibrate at 6.6Khz -> 3.3 waves for each 0.5ms
27mm piezo transducer (Roland Pad meshhead trigger) vibrate at 4.6khz -> 2.3 waves for each 0.5ms
35mm piezo transducer (Roland Pad Rim trigger) vibrate at 2.9 khz. -> 1.45 waves for each 0.5ms

to have such a low delay, I would guess that his triggering algorithm is pretty straightforward (i.e. very simple and quick, with probably very few influence of the past signals)

With a very simple triggering detection algorythm (i.e either you are over a threshold, or you are under) and with 16mips, it should be possible to have only 0.5ms of delay (yet...probably without crosstalk/retrigger rejection)

Alright, you know better, I'll shut up:)