Positional sensing on megadrum? a dream or a future reality?

[Synthex]

I think aquawicket's Edrummonitor has features on position dectection. In the future, when Megadrum Controller is integrated with his program, then position detection would be solved.

It may even be possible to do now if you run Megadrum Controller along with Edrum Monitor.

How can you do that??? A common mesh head snare pad can only send one signal from the mesh head hits... I can`t think of how to do that with a PC software

aquawicket's Edrummonitor waits 2 shots in a long time (Multi-zone wait time = 15 ms by default) and calculates the distance depending on the strength of one or another of the piezos.

[dmitri]

Positional sensing is really very interesting !
I don't think this is so complicated to do it !

Exactly! This is why we see this feature in every drum module out there:)

[elrules]

Maybe positional sensing stuff can be processed faster if you consider only steps of the full 127 CC range, I mean, having values 0, 10, 20, 30 till 127 instead of 127 different values. A ruder interpolation. I don't know exactly, it is a suggestion.

I understand you dmitri not wanting to make available your source code, but at least you could explain how you sample the analog signal (a common sampling maybe? quantizing the signal?) and what algorithm you use to determine which is a hit a which is not and its velocity. More minds thinking in the same problem makes it easier to solve it

I also understand that number of inputs is one of the main advantages of the megadrum, but that problem can be solved multiplexing more signals (can't wait to see the new 32 inputs version ) or building two identical 22inputs megadrums and have both connected to the computer to have 44 inputs. ...That problem is widely solved (I think). I offer you a more challenging problem to solve that once solved will convert your megadrum in the best diy drum module ever invented.

Synthex, you seem to also have a good mind for electronic stuff. Do you think in any efficient way to implement positional sensing?

PD: after seeing a PD-80r pad signal, I can see that the first half-wave is in the range of 1ms long. Which is the sampling speed (samples per second) the megadrum does?

[Synthex]

I think it's difficult to sample the real first half wave with megadrum hardware. Because if this sample is negative, it will be removed by a diode ...
The process is also highly variable depending on the type of pad used.

But we can dig the problem

[elrules]

I think it's difficult to sample the real first half wave with megadrum hardware. Because if this sample is negative, it will be removed by a diode ...
The process is also highly variable depending on the type of pad used.

But we can dig the problem

So the diode deletes the negative part of the signal...that's great!!. As you can see in the PNG picture included in the thread of the vdrums forum I said in my first post, the second half wave is nearly identical to the first. Its length decreases gradually, but the difference between first and second should not be determinant for the reliability of the positional sensing final CC value.

Besides, the first half wave seen by the atmega (in fact the second for pads sending first a negative impulse) is always positive and it is that half wave the one who triggers the module. So there shouldn't be problems with that. In fact,having the negative half wave set to zero makes it easier to implement a position detection algorithm.

[dmitri]

The whole position sensing thing is not a quantum mechanics science, it is doable. It's just a matter of balance of cost/construction difficulty/user demand. Of course it will only work with pads which produce different frequencies between a centre and an edge. I'll think about implementing it if enough people will really ask for it.

[elrules]

Thanks dmitri

[Synthex]

I think it's difficult to sample the real first half wave with megadrum hardware. Because if this sample is negative, it will be removed by a diode ...
The process is also highly variable depending on the type of pad used.

But we can dig the problem

So the diode deletes the negative part of the signal...that's great!!. As you can see in the PNG picture included in the thread of the vdrums forum I said in my first post, the second half wave is nearly identical to the first. Its length decreases gradually, but the difference between first and second should not be determinant for the reliability of the positional sensing final CC value.

Besides, the first half wave seen by the atmega (in fact the second for pads sending first a negative impulse) is always positive and it is that half wave the one who triggers the module. So there shouldn't be problems with that. In fact,having the negative half wave set to zero makes it easier to implement a position detection algorithm.

But piezo polarity is very important here.

On your PNG picture, the second half wave is not nearly identical to the first ...
It's only the real first half wave who gives the position

[Marctwo]

I'm just messing about with this idea myself. The most important part for me is how you organize your trigger sampling. If you have an efficient system then you should have no problems implementing this on 1 input (snare head). If your code is pushed to the limit with normal processing of 22 inputs then your current methods may not be suitable to incorporate this without additional hardware.

Position sensing would be nice on cymbals too (as mentioned) but most cymbal pads would not produce an appropriately variable frequency so may not be worth it.

[dmitri]

I'm just messing about with this idea myself. The most important part for me is how you organize your trigger sampling. If you have an efficient system then you should have no problems implementing this on 1 input (snare head). If your code is pushed to the limit with normal processing of 22 inputs then your current methods may not be suitable to incorporate this without additional hardware.

Position sensing would be nice on cymbals too (as mentioned) but most cymbal pads would not produce an appropriately variable frequency so may not be worth it.

Just so you don't bring this again and again - my algorithms are bad, the code is pushed to the limit and the whole system is inefficient. Also, if you need to advertise YOUR programming skills on this site, please do it in "Off topic" forum.