Robot drummer

#21

Research notes:

physics of drum strokes
http://acoustics.org/pressroom/httpdocs/155th/dahl.htm
~60-75N peak force
http://www.sofiadahl.net/pdf/DahletalForumAcusticum2011.pdf
peak forces ranging from 1.8 to 106.8 N
strike velocity ranging from 1 to 10m/s
(interesting f~v² relationship between velocity and force)
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.416.3669&rep=rep1&type=pdf
this one’s good. they stuck a piezo crystal and accelerometer on the end of a drumstick :smiley:

image.png939×655 48.4 KB

mmm data

So, the drum stick is moving at 10m/s and decelerates to 0 in 3ms. That’s 3333m/s²

If the contact force is 100N, that means the effective mass of (drumstick+residual effect of drummer’s arm) is 100/3333 = 0.03kg - which is in the same ballpark as the weight of an actual drumstick (about 50g)

That’s good. It means the drummer’s arm is not pressing on the drum - it’s just a fast moving bit of wood, free to rebound. All I have to do is accelerate 30g of wood to 10m/s as quickly/easily as possible.

So now I can make a quick spreadsheet so I can figure out what sort of solenoid I’m going to need. Let’s assume a 50% duty cycle - ie, every other 8th of a 180bpm beat. That’s 80ms. Assuming 1/2 that time is needed to make the hit (the other half for returning from it), I need a solenoid with…

7.5N, 20cm throw.
or, to make the hit in only 4ms
75N, 2cm throw.

These are outside the limits of physics and my budget respectively.

Someone check my sums, but I think there has to be a lever to get the required acceleration.

#22

Absolutely not! I intend to insist on a really robotic feel :slight_smile:
The second video was more to illustrate that techno can be played on “acoustic” instruments and still sound very much like techno…

#23

I assumed he meant the amount of gradation you could achieve in the force?

1 Like
#24

Moving coil should be able to do it and model different types of drumsticks and moving iron will probably have to be set up for a single type of drumstick.

I love the idea of hitting the drum with just a field from inside the drum .

#25

Yes and you could change it on the fly so modelling the weight of the drumsticks .

#26

I plan on hitting all sorts of non-drum items, eg frying pans, metal shelves, plastic buckets. Needs to be a physical hit.

Not sure you could get a short and hard enough impulse with pure magnet anyway. It takes time for the field to build up as you feed it power.

#27

Some sort of 3D printed clamp-on device could be applied to anything to make it part of your drum kit.

#28

The more I read, the more I like the idea of this project !

Levers are good and can be used as an amplifier but only on one force at a time - add speed you lose weight of impact, add weight of impact you lose speed. i’m sure there is a sweet spot there though - may have to build a rig to test the ideas.
My minds eye has the military style drummer holding the stick at 90deg to his arm and using wrist rotation to move the stick. there can’t be a Huge force but the speed is high as is the rebound and repetition. is that what you were looking at for a lever ?

Courty

2 Likes
#29

10m/s is 25mph / 35kph - complete guess without any foundation but it sounds a bit high

Courty

#30

A field can build up very quickly

#31

Data doesn’t lie!

image.png978×564 130 KB

Paying closer attention to the position/time chart, it takes a human ~100ms to bring the drumstick down, and about the same to lift back up (this example was slow paced so the rebound could be tracked separately). That’s an acceleration at the stick tip of 100m/s² (we can do better!)

There’s something screwy with the X-axis. The caption says slow tempo, but the scale goes from 0.195 to 0.22 seconds. 3 hits in 17ms is not “slow” !

EDIT: This movement trace from the 1st paper linked (PS: Sophia Dahl is a contributer on all three - she’s your woman for drum physics!). Each straight line represents 25ms.

Shows about 100-150ms for the downstroke, so as expected. So the fact remains: we must do better!

The discussion of contactless direct drive is interesting, but I think the performative aspect is important. If the audience can’t see any moving parts, it may as well be pre-recorded. Dozens of drumsticks going hell for leather is much more fun :slight_smile:

2 Likes
#32

Oooh! https://simphy.com/ Free 2D physics simulator!

1 Like
#33

Real drumsticks would be best, I think, because:

  • They are easily and rapidly replaced. Given your parameters, you will wear out strikers quickly.
  • You can interchange with different kinds of stick: batter sticks, gong beaters etc. Or go even farther out there (like we aren’t there already :grinning:) and try non-drumstick strikers: rods, whatever.
  • You can use levers to get good velocity at strike point - i.e. what real drummers do.
  • It will sound more human - a bit of robotic feel is fine but it should sound like a musical instrument.

However, the kick drum is probably the easiest place to start the project I suspect. Fewest variables and pedals are designed to take abuse.

1 Like
#34

I was going to use a big plastic tub for the kick. More sonorous, like an 808 or 909 kick. Also cheaper :smiley:

#35

Some links of interest:

https://dadamachines.com/
http://www.bastl-instruments.com/modular/solenoid/

#36

There are no new ideas :smiley:

However, I want mine to be loud enough to dance to.

#37

Cool idea, this may be of use to your project…

8 channels midi to relay, can drive large solenoids, motors, actuators whatevs…

hth + happy neu yuuur

1 Like
#38

Nice, although I think the solid-state one is more suitable:

Only thing missing is the tunable delay. That can be handheld in the sequencer, but it’d be nice to have each instrument calibrate itself…

#39

Full circle - are you heading back towards “sod the subtlety, go for volume” ? I can see pneumatics on the horizon again :joy:

Dual acting air rams with basic needle valve regulators and solenoid air valves ?

Courty

1 Like
#40

I have visions of this contraption… would liven up the maker space :smiley:

3 Likes