3 ms·

Looking at photos on Google Images, it looks like the pagers in question have just two pins on the top, and two more mating pins on the bottom. Obviously, that

by ssl-3 2y ago

Looking at photos on Google Images, it looks like the pagers in question have just two pins on the top, and two more mating pins on the bottom. Obviously, that's for power delivery: They stack to charge. But it seems like data must be sent (in one direction) through the stack at the same time that charging happens -- to set IDs, and to let attract mode work. One way I can conceive of to allow this to happen is by reversing polarity: Send power with one polarity for a 0, and in the opposite polarity for a 1. Once a pager has some data, it can pass it along to the one stacked above it by doing the same thing -- in a manner not dissimilar to how things like RGB LED strings do. Charging can happen at the same time as this data is received (and potentially modified) and then retransmitted by using a bridge rectifier in each unit for the charging circuit, so that (as far as the battery bits can see) the polarity remains constant. (This sounds complex and slow, but using things like diodes and MOSFETs is cheap at mass production PCB stuffing scales, and the data rate needn't be fast. It may even be cheaper to do it this way than a more complex assembly with dedicated data pins.)

kragen 2y ago
this is a great idea. you don't even need a bipolarity driver, just an h-bridge. though if you do have plus and minus rails plus ground, you only need two transistors
- ssl-3 2y ago
  How do you mean? There are only two contacts between any device and the device below it, according to the photos. This trend continues upward for all subsequently-connected devices. An interface with +, -, and ground implies three contacts.
  - kragen 2y ago
    i mean you connect one wire to ground and switch the other wire between - and + with two transistors, thus reversing polarity on it in a way that can be used to signal information while also efficiently transferring power (though only a quarter of the power you could transfer with an h-bridge across rails of the same voltage) basically rs232 signaling with just tx and ground
    - ssl-3 2y ago
      I like it. That's related to, but more elegant than, my off-the-cuff idea was.
      - kragen 2y ago
        i think that to the receiving device it looks exactly the same as the h-bridge approach which i think was your initial version, unless it has some other path to ground, maybe a capacitive one. but if it does have such a path, it's probably useful for it to have such a genuine ground pin
fragmede 2y ago
Or you could have a fixed clock and sample based on that, so no having to deal with reverse polarity. This is the restaurant industry though, so assume they'll get wet and the the worst possible combination of pins get shorted, so design the circuit to be as tolerant of that as possible. Also consider the 4 pins getting into conductive fluid electrified with AC.
- ssl-3 2y ago
  Perhaps. Can you elaborate on how your own idea works? I think I know what you mean, but I'd like for it to be explained. My own idea implies a potential of ~100% charging duty cycle even as low-speed data is being transferred, and that's good because charging seems to be the primary goal here in this application. (My other idea, that I have not yet espoused, involves a waveform on top of a fairly-constant charging voltage. It can perform asynchronously, and might even be just amplitude modulation. But it probably means more parts and complexity than my first idea involves: Functional analog filters always seem to take a handful of parts to get the it right, whereas my polarity-swap idea only uses a couple of high-tolerance jellybean MOSFETs and some parts-bin diodes.)
  - fragmede 2y ago
    every 10 milliseconds (or whatever), the base goes high or low, and by having a clock on board that can count 10 milliseconds, the puck can sample the pin and read a high or low value, and then you specify patterns from there. You'd have to make it so it can be connected in the middle of a packet and ignore it and such, but those are solvable. (Technically you'd sample more frequently on the pucks due to Nyquist.) Ooh, just AM off the power pin itself is another clever of doing it. All comes down to price and what's the cheapest to implement so you can drive the BoM on the puck's down as much as possible.

by ssl-3 2y ago

kragen 2y ago

this is a great idea. you don't even need a bipolarity driver, just an h-bridge. though if you do have plus and minus rails plus ground, you only need two transistors

ssl-3 2y ago
How do you mean? There are only two contacts between any device and the device below it, according to the photos. This trend continues upward for all subsequently-connected devices. An interface with +, -, and ground implies three contacts.
- kragen 2y ago
  i mean you connect one wire to ground and switch the other wire between - and + with two transistors, thus reversing polarity on it in a way that can be used to signal information while also efficiently transferring power (though only a quarter of the power you could transfer with an h-bridge across rails of the same voltage) basically rs232 signaling with just tx and ground
  - ssl-3 2y ago
    I like it. That's related to, but more elegant than, my off-the-cuff idea was.
    - kragen 2y ago
      i think that to the receiving device it looks exactly the same as the h-bridge approach which i think was your initial version, unless it has some other path to ground, maybe a capacitive one. but if it does have such a path, it's probably useful for it to have such a genuine ground pin

fragmede 2y ago

Or you could have a fixed clock and sample based on that, so no having to deal with reverse polarity. This is the restaurant industry though, so assume they'll get wet and the the worst possible combination of pins get shorted, so design the circuit to be as tolerant of that as possible. Also consider the 4 pins getting into conductive fluid electrified with AC.

ssl-3 2y ago
Perhaps. Can you elaborate on how your own idea works? I think I know what you mean, but I'd like for it to be explained. My own idea implies a potential of ~100% charging duty cycle even as low-speed data is being transferred, and that's good because charging seems to be the primary goal here in this application. (My other idea, that I have not yet espoused, involves a waveform on top of a fairly-constant charging voltage. It can perform asynchronously, and might even be just amplitude modulation. But it probably means more parts and complexity than my first idea involves: Functional analog filters always seem to take a handful of parts to get the it right, whereas my polarity-swap idea only uses a couple of high-tolerance jellybean MOSFETs and some parts-bin diodes.)
- fragmede 2y ago
  every 10 milliseconds (or whatever), the base goes high or low, and by having a clock on board that can count 10 milliseconds, the puck can sample the pin and read a high or low value, and then you specify patterns from there. You'd have to make it so it can be connected in the middle of a packet and ignore it and such, but those are solvable. (Technically you'd sample more frequently on the pucks due to Nyquist.) Ooh, just AM off the power pin itself is another clever of doing it. All comes down to price and what's the cheapest to implement so you can drive the BoM on the puck's down as much as possible.