This is a new circuit (designed in December 2015). Consequently I have not built a prototype and so it remains untested; I therefore cannot guarantee that it performs as described below. If you build it, please let me know the results.
The signal itself, shown in the shaded area of the diagram below, consists of two reverse-parallel pairs of LEDs. This configuration offers the advantages that each LED protects its partner against inverse voltages and the signal itself requires only a three-wire connection. Disadvantages are that none of the three wires to the signal goes to the common return and the driver needs three push-pull outputs. In the 1980s I devised a simple TTL-based driver in which a pair of inexpensive ICs provided the logic, the push-pull outputs and even the current limiting resistors. The demise of TTL and the advent of LEDs offering better colour but requiring higher working voltages led me to create a similar driver using discrete components.
The circuit is shown in the following diagram:
Deliberately no supply voltage, transistor types or resistor values are specified. You choose your own supply voltage depending on the types of LED you are using in the signal—and also of course on convenience! All transistors are general-purpose silicon types such as BC107 (NPN) and BC177 (PNP). There are two current-limiting resistors, R12 and R13. Again you choose appropriate values depending on the LEDs, supply voltage and brightness required; suggested starting values are 1K to to 2K2. Unfortunately each resistor limits the current for two LEDs of different colours; hopefully you will find a compromise value that suits both LEDs. The other 14 resistors are all bias resistors and their value is not critical, but to save current, should be kept high; a suitable starting value is 100K.
The current paths for the four aspects are as shown below:
|double yellow: as above plus||T9||R13||yellow 2 LED||T6|
This circuit eliminates the ‘sneak current’ that in the TTL version could cause the yellow 1 LED to glow feebly when green was on and yellow 2 to do the same when red was on. In this circuit when green is on, T3 and T4 are both ‘off’ and when red is on, T9 and T10 are both ‘off,’ so there can be no unwanted glimmers.
Inputs 1, 2 and 3 are normally connected to the train detectors in section 1 (that is, the section which this signal guards), section 2 (that is, the section after section 1), and section 3 (that is, the section after section 2). If input 1 is activated the signal shows red, irrespective of the status of other inputs. If input 1 is clear, activation of input 2 causes the signal to show yellow. If inputs 1 and 2 are clear, activation of input 3 produces the double-yellow aspect. If all inputs are clear, the signal shows green.
The inputs assume negative logic; that is, to activate an input that input must be connected to the supply negative line, which forms the common return. The output of the Tektor track circuit shown elsewhere on this site is ideally suitable. Additional inputs are easy to provide. If, for example, your signal guards a trailing junction, you will need an additional input 1 to turn the signal red if the turnout is set against the train. Connect the new input via a suitable resistor to the base of T1.
Most of the logic for the circuit is provided by the circuitry around T1 and T2. Essentially inputs apply bias to those transistors which in turn bias others on and so on.
With no inputs activated (that is, when the next three sections are clear), only two transistors in the circuit are conductive: T10 receives bias via R16; T10 in turn biases T7 on via R15, creating the path that lights the green LED.
When input 1 is activated, bias is provided for T1 via R4. T1 in turn provides bias for T11 via R5, shorting out the bias normally provided for T10, thereby inhibiting the green aspect. T1 similarly provides bias for T5 via R7, clamping the base of T6 to ground and inhibiting yellow aspects. Finally T1 provides bias for T3 via R6 and T3 provides bias for T7 via R10, creating the path that lights the red LED.
Assuming that input 1 is not activated, activating input 2 or input 3 provides bias for T2 via R1 or R3. T2 in turn provides bias for T11 via R8, inhibiting the green aspect and it provides bias for T6 via R9. T6 in turn provides bias for T4 via R11 and T9 via R14, creating the paths that light both yellow LEDs. But activating input 2 also provides bias for T8 via R2 which shorts away bias supplied to T9, inhibiting the yellow 2 aspect. In this way correct four-aspect signalling logic is conserved, the double-yellow aspect appearing only when input 3 alone is activated.
© Roger Amos 2016