I am interested in a circuit board to control a two digit 7 segment LED.  Information will be coming from my software.

If you need to route low-bandwidth data around a warehouse, use ultrasound.  The transducers are about a buck a pop.  They operate around 50KHz.  Easy to modulate, say DTMF or RS232 fsk.  No FCC.  Tolerant of dirty environments.  Time to wander out of the box.

I et the impression the displays are very close to each other, perhaps a rack of component draws with LEDs on the front to indicate count needed to be taken, think of a classic Postal Pigeon hole wall, lots of small holes in a square grid waiting for stuff to be removed, so running a high speed data link around the set should be well within reason (SPI im thinking or is it I2C... the one without addresses)... that's what I get for drinking a nice glass of wine before responding lol

I cant help thinking of a similar system to NEO PIXELS but using the 595 serial to 8 bit chips instead or even the TI 5940 serial to 16Bit chips, the latter providing way more capability than needed but if the cost is right...

I et the impression the displays are very close to each other, perhaps a rack of component draws with LEDs on the front to indicate count needed to be taken, think of a classic Postal Pigeon hole wall, lots of small holes in a square grid waiting for stuff to be removed, so running a high speed data link around the set should be well within reason (SPI im thinking or is it I2C... the one without addresses)... that's what I get for drinking a nice glass of wine before responding lol

I cant help thinking of a similar system to NEO PIXELS but using the 595 serial to 8 bit chips instead or even the TI 5940 serial to 16Bit chips, the latter providing way more capability than needed but if the cost is right...

@Peter and Don and Douglas

It needs a micro on each display because it needs intelligence on each display. The shift register solutions won't scale to 1000 devices and the clock and data lines can't be emc proofed. The LED driver chips don't support the button and all the shift register systems are addressed by position on the chain so that the system is um-maintainable.

It can be done with UART style comms (NOT RS232 because the hardware must be different.) Each module needs a unique address (can be coded by dil switch or solder link/track cut). All the displays listen all the time, the PC sends out addressed commands so only one display responds. It periodically asks display which are on of their buttons have been pressed (it's a feature of this system that only a small number of displays are lit at any one time.)

Just for housekeeping the PC can also poll all the devices on a slow round robin basis just to check that they are happy.

The timing for this works fine, at 9600 baud it can use, say, 6 byte messages, (2 address, 2 display data, 2 spare) and receive 4 byte messages - so it could interact with about 100 displays per second, max but perhaps more realistically with 25 - 50 allowing for latencies etc.

There is now way the human picker can keep up with 25 part bin interactions per second so it's easily fast enough.

So now all we need are three or four wires linking all the devices, all in parallel. Fastest signal is 9600 baud. The receivers on the displays need to be low load so that they can all be driven at once (but that's not too hard at this data rate). The design of the display data driver needs care so that there is no risk of it being permanently asserted because this would jam all the whole system.

Because each display has a unique code and they are wired in // you can add or remove displays and connect them to the wire at any point in any order and it all still works.

It's all quite doable but it needs a custom display design.

MK

I was initially thinking along the same lines, Michael, about the intelligence required at each display.  To reduce the complexity and to remove the need for two-way communications, the user could enter into the computer that they pulled the parts, at which point the display would get cleared.  If the design needs a clear button at each bin, two way comms may be necessary, for which a small microcontroller would be ideal and pretty easy to set up comms.  Addressing could be jumpers (wire or solder), or even software configurable.  I like your idea for comms, using a full 2 byte address leaves plenty of room for expansion.  This is definitely ideal for a custom display board, though I think some of the cascading options should be able to work since the signals are going to get buffered at each device, up to a certain point.  Multiple buses can be created using I/O expansion as well.  I wonder if there is a market for something like this?  There are already some which are expandable to a degree, but typically nothing more than around 16 devices per bus (at least using in I2C example).