hard knock trigger

I am looking for a trigger or switch that needs good amount of force to activate. I have a large canvas - video screen that needs to be interacted by hitting several areas with a stick. I have used piezos so far, but I run into difficulties with calibration every time. since these are on the same surface hitting one place will trigger some othrers as well. I can reduce sensitivity OK, but sometimes the hits are too different in force and can not be predicted.
So if there would be a sensor-trigger-switch that would react just as on-of regardless the amount of force applied. Just the normal momentary switch will not work because there is too much movement of the canvas.

Presumably in the centre of the canvas there is quite a lot of movement and not much on the edges. What about the micro switches with a long wire actuator, maybe you could tweek each actuator to be more/less sensitive. IE the ones in the middle have a long wire, on the edge a short wire. The wire actuators allow some over travel which hopefully would accomodate those with a heavy hand at the stick.

Is it possible to use three peizo sensors mount around the edge and trianglulate based on the force registered by each?


Rob

This canvas is a video screen that is projected from rear mostly. http://vimeo.com/7834730 So this trianngulation idea is rather interesting. Could you explain more?

My first thought was to go with Richard's idea, it would be very simple to implement, then you said

This canvas is a video screen that is projected from rear mostly

So no things behind the canvas, including switches of any kind.

The triangulation was just a thought of the top of my head, it should be doable but probably above my level. However I do have some thoughts. I see two methods based on the resolution you need.

High resolution: You need to allow hits anywhere on the canvas. Read the magnitude from each sensor and come up with an algorithm that says OK we've got magnitudes of 2 8 and 23 so the hit is really close to sensor #3, quite a way from #2 and miles away from #1 so the hit was at X,Y. The readings may not be linear and there will be other factors I haven't thought of. I'm sure this could be done, it's probably easy for someone used to working with such things but above my level.

Low resolution: You only need to know about 3-4 hot spots (maybe 10-20 I don't know how accurate this would be). Set the rig up, hit the spots and record the readings. Each spot should have a diffferent set of readings. These then become a "signature" for each spot, but not the absolute values, just their values relative to each other because people will hit at different levels. Then when a hit is received compare the signature with the prerecoded data including some fuzziness to allow for drift and the imprecise nature of the real-world. If the readings are not linear then fudge factors would have to be applied based on the absolute values read to linearise the signature before comparing.

Would this all work? I've no idea but with a smart programmer I reckon it should. A VERY smart programmer could possibly detect things like the rise time of the hit to dertermine if it was a real wack or a gentle touch and modify the display accordingly. That's even further about my level. :)

Another thought, can you do an LED/LASER matrix across the front of the sheet?


Rob

Read all the sensors and mark the time each receives a vibration. Use that to calculate where the canvas was struck. These would actually need to be somewhat more sensitive but not so sensitive that sound would set them of. You may consider conditioning the peizo's output so that you can get a clean pulse out of them which you can then do some math on to calculate a location.

http://itp.nyu.edu/physcomp/sensors/Reports/Triangulation

I'm not very math literate.

Thank you. It seems that i need to find a good programmer to do this. But I am considering some LoFi solutions with mechanical devices too.

This could be pretty tough because you're trying to detect something that's more of a "thump" than a "hard knock" :)

Detecting an impact on glass, or even wood, is easier because the sound pulse is relatively narrow and much stronger than hitting canvas. It's like reading by candlelight: it's harder because the "contrast" is lower.

How fine a resolution do you need? If it's "not much", you might be able to do something like threading wires through the canvas, and putting a weak RF source in the end of the stick. Then you could detect which wires are picking up the strongest signal.