Benchmarks

[Marctwo]

I'm using the 18f4550 (same family as your USB, Synthex) but practically any old chip with A/D would get the same results.

I've found 0.5ms to be comfortablly larger than a piezo half wave. This would likely depend on the piezo and how it's mounted though.

I'm just a bit shocked by this sudden realization that I obviously set myself a relatively very high target. I'd just assumed that everyone else would be working in that range.

Has anyone tested an e-Drum module, I wander?

[Synthex]

Has anyone tested an e-Drum module, I wander?

Of course. Roland module.

[Marctwo]

Of course. Roland module.

No, I meant one of these: http://www.edrum.info/. I may have confused you with the hyphen.

[Synthex]

Of course. Roland module.

No, I meant one of these: http://www.edrum.info/. I may have confused you with the hyphen.

Yes too, latency from eDrum is longer than MegaDrum !

[Lakedaemon]

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

Oh well...I was just stating facts (I got them there frequency of piezo tranduscer) and what I understand of periodic signals (I'm a mathematician and know next to nothing in electronics).
I didn't want to upset you.

Besides, I post lots of (annoying ?) messages, because I would really like to understand (electronically and theoritically) the board I'm trying to build.
(this has been really enlightening till now).

If all goes well, I might be able to build megadrum (and sell my TD20 thus getting fun, knowledge and money in the process).
And I might be able pass the knowledge/experience gained about electronics/the electronic world to my students (they study to become engeneers).

On a side note, what I like about megadrum is that it has a modular architecture. You can choose what it will become (LCD, keyboards,...).
I would love the source to be open too (that way more optional features could be made...), but it is not.
It is your choice not to open it and I respect it.
(If I really wanted it that bad, well then I should blunder around trying to make my own...like a real man).

[Lakedaemon]

As for benchmarks....getting a latency between 3ms and 6ms is quite good (excellent ?)

a 400+€ professional soundcards (like an Emu 1820m) give you at best 2ms of latency and I bet than in real world scenario, it is higher than that
(when I work in Cubase in a big project with lots of tracks and stuff, I have to raise the latency not to get clicks).

When you are playing megadrum...you have to add the latency of the sampler and of the soundcard
and it should be much higher than megadrum's meager 4.5ms mean latency ^_^.

When you are recording, between 3ms and 6ms should be good enough (as I'm a beginner, I don't even play the drums good enough to make a difference anyway) especially since the good sequencer compensate latency.

In that aspect, for recording purposes, what would be very nice for megadrum to have is consistent latency.
(i.e. a fixed latency.....but this should be quite hard to implement (must be done in software)).

[edit] dimitri says in this thread that the megadrum latency should be pretty constant....weeeeeeeeeeee[/edit]

[Lakedaemon]

Now.....concerning waves.

I would be very interested in what a piezo wave looks like.
Is is a sine wave ? Is it something more complicated ?

Does someone know ? has someone recorded such a wave ?
(maybee I should do it, instead of asking for it :p).

On a side note... If I'm not mistaken, megadrum process the positive part of the waves.

It should be possible to process the negative part too, with atmegas but I guess that it would require more passive parts :

if you have a tension V between -VCC and VCC,
you would have to transform it to V'=(V+VCC)/2 : it is positive and between 0 and VCC, so the atmega DA pins can sample it.
now, inside the atmega, you can get V by doing V=2V'-VCC so, you can work on V.
But, you don't even have to compute V : you can work directly on V'.


Now, this would of course require more parts for each DA pin of the Atmega (or of the 4051...we want to avoid that...to much pins).
As, I'm no electronician, I don't know how much but I guess than for simplicity and costs concerns, it is much smarter for megadrum to only process positive parts of piezo waves (I don't even know how they look like anyway, so this is just speculation).

[Goldie]

I would be very interested in what a piezo wave looks like.

This thread starts with picture of piezo wave.

[Lakedaemon]

indeed.....

it looks like the dampening factor of the wave is really very high (not that unexpected with the mecanical deformation of the piezo)
and the initial voltage differential is very high too.
(it makes the signal saturate.... for 1ms ^_^).

Somehow, it contradicts the 2-6 khz vibrating frequency (of the piezo specs...maybee I misunderstood that)
and confirms what dmitri said (i.e not even half a wave in 0.5ms).

Concerning megadrum. If the big voltage differential is negativ, that means that megadrum triggers on the second half-period (around 1.5ms after impact) ?

[edit] a funny thing with trigger, is that contrary to audio saturation....we can somehow guess the maximum of the saturated half-wave signal by computing the maximum of the next non-saturated half waved and applying the inverse of the dampening factor ^_^[/edit]

[Marctwo]

The picture at the start of the thread looks very much like a processed (and clipped) signal. Was this taken directly from the pad? Does your pad have passive componants between piezo and output?

The raw piezo signals I've been working with are just mid frequency audio. The exact nature of the signal can vary quite a bit depending on how it's hit just like a cymbal can produce a different sound depending on how hard it's hit, where it's hit, which part of the stick is used, etc.