OK, here you go. I guess I just have a simple mind and like to keep things simple.
Actually, if 24 volt relays were used, there is no need for a regulator beause the chips can operate at this voltage too.
Three reed switches are arranged along the axle line so that a wheel spinning a magnet will close the switches when the magnet passes. If three reeds can’s fit along one wheel they can be divided up between the two wheels on the common axle assuming they are fixed and always turn together (i.e., non-ball bearing) but this would require two magnets; one on either wheel.
For the sake of discussion, the “start” reed will be situated at 0 degrees along the wheel (straight up). The “forward” reed will be placed at 120 degrees and the “reverse” reed will be placed at 240 degrees. The actual placement is unimportant just so that when the car moves in a forward direction the wheel magnet first encounters the “forward” reed from the “start” reed, and vice versa in a rearward direction.
The “start” reed triggers a dpdt relay. There is a capacitor mounted in parallel with the relay coil such that when power is removed, the relay stays triggered for a moment (~1/2 second). If this capacitor presents too much “load” for the reed switch, a resistor can be used to limit the current through the reed.
The “forward” and “reverse” reeds serve as triggers for a pair of 555 (one 556) timing chips. The triggers are set in series with the contacts for the “start” relay such that if the “start” relay is active and a trigger relay is activated, its current can flow to the 555 chip. This triggers the chip and it throws its relay. If the “start” relay is not active, no current can flow to trigger the forward and reverse chips.
The “forward” and “reverse” chip relays are wired so that one armature and set of contacts are used to power the LED or whatever you desire. The other set of contacts are also in series with the trigger reed for the opposite direction. I.e., if the forward relay is active, the reverse reed switch becomes disconnected and vice versa. Additionally, the contacts allow the reed switch to close the circuit that is connected to the active chip. So if the forward relay is activated, it also lets the forward reed “communicate” with the forward chip when the magnet passes regardless of the state of the start relay.
The 555 chips are wired for “one shot” (monostable). There are two possible configurations. In one, the chip accepts a trigger pulse but will then ignore all further trigger pulses until it “times out.” In the other configuration, the chip will restart its timing cycle from the beginning when it receives a trigger pulse before it times out. This is the configuration we will use.
So, let’s run it through in a forward direction and say the magnet starts at 180 degrees or straight down between the “forward” and “reverse” reeds. The magnet begins to spin and first encounters the “reverse” reed. This pulses the reed, but the circuit is incomplete because it is broken by the “start” relay so nothing happens.
The magnet moves on and encounters the “start” reed. This closes the “start” relay and the magnet moves on. The capacitor in parallel with the “start” relay holds the relay open as the magnet triggers the “forward” reed switch. Current flows through the reed then through the “start” relay to the “forward” 555 chip. The chip activates its relay powering the LED, breaking the path between the reverse reed and its chip, and making a path between the forward reed and its chip. We’ll set the “forward” and “reverse” chips to time out at 2 seconds. If no activity detected within that time (e.g., its reed is not triggered) it will reset its relay to the open position. The capacitor on the start relay eventually runs out (e.g., ½ second later).
The magnet proceeds around and encounters the reverse reed. This closes the reed, but it cannot communicate with its chip because the series contacts on the “forward” relay is interrupting the path so is ignored. The magnet proceeds around and encounters the start reed and triggers it. But the forward chip is already active and its reed is already connected to its chip, so it’s a moot point.
Eventually the car stops and the chip times out (at two seconds) and whatever direction the car restarts will dictate its behavior until it stops again.