Determening the location of a ball contacting a surface

Question

I'm a student currently working on a project. I have two objects, one of which is a plastic ball. The other is a rectangle which is being hit with said ball. My goal is to detect a position of a hit. The rectangle is about 5x5 cm. What would be the least intrusive way of sensing a position of a hit?

Perhaps I could use some kind of sensors on the rectangle to try and triangulate the position? Maybe put a small magnet on a ball (d=3cm) and with a help of small coils on the rectangle to calculate the position (could that work? or is it just stupid?) Small current would induce on each coil. How about putting a thin metalic film on the ball and an array of small (+) pads and an array of detection pads (grounded via pulldown resistor)? Like a matrix.

Answer 1

Not stupid idea, but use soft low remanence iron ball instead of a magnet. It's position is arbitary, nothing needs to point to some specific direction. Have an external DC magnetizing coil and some small pickup coils. Measure the pulses induced to the pickup coils.

The hard part: You must match some complex function that gives the hitpoint coordinates from the pulse amplitude proportions. It can be a table of premeasured values and the intermediate values can be got by interpolation. Triangulation may be useful or not.

DC magnetizing might be better to replace with high enough frequency AC to get stronger signals. That would eliminate all effect of remanence and earth magnetic field. Also the eddy currents can increase the detectability. You must explore with hf magetization, which phenomena gives the strongest effect.

Iron powder+glue or ferrite ball has no eddy currents, a copper ball has nothing, but eddy currents, solid iron ball has magnetic effect that is affected by eddy currents.

Totally different approach: Learn how resistive or capacitive touch panels work

The next: have some microphones or schock sensors. Triangulate from timing differences. Actually, to get the result fast, compare with a table of prestored timing patterns and iterate or interpolate.

And still one more: One schock sensor gives the hitting time and triggers 2 perpedicular line cameras or lightport arrays. The software finds the mass centres of the shade patterns and these are your coordinates

Good luck!

Answer 2

What would be the least intrusive way of sensing a position of a hit?

If this is for a temporary project rather than for a product - then I'd mount the square that's being hit on a flexure made out of a PCB, with SMT ceramic capacitors acting as piezoelectric strain gages. The signals should be strong enough to connect directly to MCU inputs. The higher the multiplexing ADC sampling rate, the better, since ideally all four capacitor voltages should be sampled at the same time.

The approach is not fit for a product, since the sensor will have a limited lifetime due to fractures likely to develop in the multi-layer ceramic capacitors (MLCCs) and the solder connections. It probably could be engineered to last a long time, but then there are easier approaches that are guaranteed to work.

With the same PCB flexure approach, a strain-gage-like trace pattern on top and bottom of the flexed area will work as a nice deflection sensor. With four such patterns on each flexure you can have a bending moment sensor instead of just a deflection sensor - it may be more accurate. Many low cost approach to measuring the small bridge imbalance voltages are possible, without having to use an instrumentation amplifier.

Answer 3

One fairly simple and robust method would be to use a keyboard scanning matrix but using vertical and horizontal grids of wires separated by a millimetre or so. One grid could rest on the surface of the box - maybe with a sheet of rubber.

^{simulate this circuit – Schematic created using CircuitLab}

Figure 1. Scanning matrix.

• The arrangement shown in Figure 1 would give an 8 × 8 resolution. This could be increased by adding additional rows and/or columns.
• It requires something like an Arduino with ≥ 16 GPIO pins.
• The code could drive each column high in rapid sequence, disabling other column outputs.
• The inputs would then be scanned to check for contact.
• The contact point could be reported back to an LCD or serially to a computer.

You'd want wires that would take a bit of stretching and abuse. Guitar strings do this all the time!