I am working on a large interactive wall that will use around 2,000 hall effect sensors to register the placement of pegs into holes. For our design to work as intended, I really do need 2,000 unique inputs. I am hoping to use a hall effect sensor that reads magnetic flux intensity to create some differentiation between the pegs (stronger magnet is a different input than a weaker magnet, for example). I have gotten something close to working with about 77 sensors, but I am concerned about scaling up to the full 2,000 sensors. If I used typical shift registers, I'd be using over 250 of them. Also, I'm not sure how much issues like voltage drop will effect this size of sensor array. I'm also concerned there may be other issues I've not yet thought of.
My question is if this seems like a reasonable approach, or if I'm missing something big that will keep this from working? I also think I may be missing another input type that would work better than hall effect sensors. It is important it is not a mechanical switch and ideally would allow for multiple inputs per sensor. I considered RFIDs, but I think there might be more problems with that than are worth the hassle.
Any feedback would be very appreciated.
Analog signals require analog multiplexers, e.g. 40x50 or multiple smaller matrices for 2000 sensors. Electrical interference may be a problem. Smaller matrices with dedicated ADC are required if response time matters. Eventually a network of controllers with ADC is a reasonable choice for fast response.
did a similar project some years ago involving 20000 inputs
several sensors (8 I think) were connected to a small processor which were arranged as a shift register
can you give details of your hall effect sensor and your current prototype?
That means you would have 2000 analog inputs. There is no such thing as an analog shift register, you would need to use an analog MUX. You would need 125, 16 channel muxes.
I would connect 6 muxes to a Nano and duplicate that 21 times. Then connect the 21 nanos via RS485 to a master UNO R4.
Do you mean, like an image recognition type of thing? There is a lot of activity in front of the wall while it is in use, so it would have to be from behind, but it is wall mounted.
Ok, that sounds like a reasonable solution. I still may jettison the idea of an analog input and go with the hall effect switch, but this is a great recommendation. Thanks!
Yeah, currently it is a series of hall effect sensor switches which read a magnetic 'peg' which then drives a display. You might think of it as a giant lite-brite, or a giant game of battleship. the sensors are placed near where a peg is inserted and when it registers the magnet it triggers an animation.
Can you give me a reference to use processors as shift registers?
I had some thoughts about that at one point, but I really want to limit the sensors to the objects which are provided, this would allow people to just use their fingers which isn't really in the spirit of the project. Perhaps there is another type of transparent capacitive screen that doesn't pick up human touch?
Is there a way to easily convert an analog input to digital to then feed to the shift register? I am guessing that just means 2000 more components which seems silly.
I know practically nothing about MUX so I'll have to do some research there ,but the plan you laid out seems reasonable. Thank you.
it was a commercial project over 10/15 years ago and I have no documentation, schematics or code
I seem to remember we used something like a PIC10F200 - cost was crital as we were using thousands of them
used three GPIO pins - in, out and clock - on a clock pulse a bit was shifted in and a bit shifted out
AVR controllers typically have an ADC with multiple (6-16) analog input channels. As I understand your project each controller has to scan all connected hall effect sensors for insertion (and removal?) of a peg. On such an event it sends a message to the master controller.