Hi everyone,

I want to design a cheap system for a game that can detect a small object dropping on a wooden Table/Plate, i.e. find the position of the drop.
A combination of shock sensors on the table and the dropject, high-gain amplification of their signals, onset discrimination and then a time to tigital conversion would be my first attempt here.
If the TDC is a good idea maybe even super-cheap shock sensors might suffice. Amplifying the sensor signal strongly shouldn't be a problem, because I only need to know if there is basically any signal present at the sensor above the noise floor and driving into saturation is okay with me.
The onset can be read out by a high-speed quad comparator or so (depending on the #ofsensors required) and put into a nice TTL voltage for the TDC.
The tabletop is sitting on some rubber bumpers... so it can resonate a little bit and carry the shock waves to the sensors.

so thats the general idea...
Do you have any suggestions/wisdom or is there a veto?

thanks

 

Sounds like an interesting project.

I suspect wood has different sound transmission properties depending on whether it is with or across the grain. Maybe a ceramic/glass plate, plastic sheet, or even a water filled vessel might be a better thing to start with.

John

 

Seems to me such a system could work though as jpanhalt pointed out, you would be at the mercy of the sound transmission characteristics of the wood; if not uniform in all directions, you could get erroneous results.

I should think an acoustic system, with three or more microphones placed around the table, might be better because it would remove that uncertainty.

 

Those accelerometers would work. They are quite inexpensive now.

 

The only problem I see with the concept is that the speed of sound in w ood is about 156,000 inches per second or 6 usec. per inch.
So depending on the resolution desired you will need to be careful with processing speed and calibration of the amplitude of the sensors as the waveform may not be a square wave.

 

I've seen a similar project done with an Arduino. Of course, I saved the link and it cannot be found now. Try Googling "locate an object by sound Arduino" and several projects will appear. Triangulation by sound seems to be a common project.

 

thanks for all the feedback!
The acozustic system would always be at mery of the background noises e.g. somebody talking. Also the whole detection process is slowed down because the sound travels way faster in the wood than in air. But after all it might just better with mics...

@djsfantasi:
Thanks, do you mean this project right here? http://duino4projects.com/sound-localization-using-arduino/

 

That might work. I was thinking more of time delay analysis rather than phase delay. Like this! Oh, Nevermind... I can't find the project. Google has failed me. Sorry!

 

Use a material with a homogeneous structure, wood has grain etc. that will mess things up.
I have successfully used piezo ceramic disks from beepers glued to acryic panels to detect impacts of objects, but never tried to localize the impacts.

 

thank you! You got me pondering whether it is even feasible in wood with shock sensors...
The signal is much faster in the wood than in air and I want to directly play a miss-sound, kind of like a stone falling into water, when the target is not hit in a radius of about 2 inches.
So the mic-array has a delay disadvantage... and considering all the computation that needs to be done for detecting place of the impact I'm not quite sure whether that is not a deal-breaker for the microphone solution after all

 

Another question on that matter: do you know multichannel time to digital converters that would fit here? I can only find one by asam with 8 stop channels, the TDC gpx. But I do not need to be as precise as 80 ps like that TDC and also 8 channels are too many... 4 or 5 stop channels should suffice.
Or do you think it would be better to use a completely different method here? For instance a PIC with CTMU?