Capacitor delay circuit specific component selection

I’m looking for advice on how to design a time-delay relay circuit using discrete components (rather than, say, a CMOS or 555 chip). The goal is to have delay between the switch being closed and the relay closing. This is necessary because the switch shown is part of a 12-way rotary, but I don’t want the relay to trip if the switch is merely passing that position between two others. I'm guessing the maximum pulse time of such a 'passing' contact would be about a second?

The attached circuit is where I’ve gotten to so far, but there are a few outstanding elements.

How do I figure out which transistors this circuit needs, and whether or not they will need heatsinks?

Which resistor should be replaced with a variable resistor, to allow the delay to be manually keyed post-installation for a specific delay within a range? This is necessary both because I don’t know exactly how long it should be yet (probably somewhere between 1-10 seconds, at this stage I’m thinking four seconds but I don’t know for sure yet), and because the input voltage is nominally 12vDC but in practice is probably closer to 14-15vDC. (I'm guessing that the 6K8 could be replaced with a variable 10K, with the third lead straight to ground?)

At present, the relay correctly drops out when the switch is opened, but the capacitor retains some charge and so the switch being re-closed does not properly reactivate the time delay. Is it possible to use the free NC contacts in the HK19F relay (or some other subsystem) to briefly short out the capacitor, forcing it to re-start the timer from zero on the next cycle?



Detailed specs:

a) Close of switch to close of relay, between 1 and 10 seconds, I think four but I won't be able to confirm until the circuit is in-situ and I can adjust it manually.

b) The coil current draw, from memory, is between 17-20mA depending on which specific relay I grab out of the bucket (I bought about a thousand of the HK19F's, for a different project that has since changed and now they're spare). Nominally 20mA. Coil resistance is nominally 720Ω, +/- maybe 40Ω.

c) Yellow diode is an LED.

d) Capacitor, as shown on the diagram, is 470µF in this draft, but I'm open to swapping it for something else if needed. I arrived at this value by trial and error, using this simulation (lower left): https://tinyurl.com/2adgnrzy

No blatant topology errors. Please update your question with your math for the timing capacitor value.

One possible issue is that the relay coil will be energized very slowly as the timing capacitor (reference designators!) voltage approaches its exponential peak. This could cause the relay contacts to "chatter". The usual solution is to add hysteresis to the circuit, but for your circuit that could mean a third transistor.

Another approach is one transistor plus a zener diode. Changing the transistor to a small MOSFET allows for a much smaller timing capacitor. Because the relay coil current is so low, a little hysteresis can be worked in. Schematic later.

Try this circuit:



simulate this circuit – Schematic created using CircuitLab

Waveforms:

Operation: Assume initial conditions: the voltage on both C1 and C2 is zero.

When SW2 closes, the voltage on C1 starts to rise due to R3. When C1 voltage reaches the break-over voltage of zener diode D1 (5.1V), current flows through zener to the base of Q1. Q1 and Q2 form a self-latching pair; both Q1 and Q2 are rapidly turned on by mutual action, which forces current to flow through the relay coil. The relay is now turned on, and its contacts close. LED D3 lights up.

When SW2 opens, C1 discharges via D2 and the coil of RLY1 until the Q1-Q2 latching pair turns off; this will occur when the voltage on C1 is relatively low (between 1 & 3V). C1 then continues to discharge via D2 and R8. C1 should be completely discharged less than 0.01 seconds after SW2 is opened.

The circuit is now ready for the next cycle.

To adjust the time delay, change the values of R3, C1, and the breakdown voltage of D1.

C2 does not have much effect on the time delay, but it is there to prevent spurious turn-on of the Q1-Q2 latching pair (from radio interference, energy spikes, and the like).

Note that R6 may not be needed, it is there to limit the current flowing through the relay coil in the simulation. Please do your own experiments to ensure the transistors can safely carry the coil current at the supply voltage you intend to use (I used +12V in the simulation here).

You may notice that the value of C1 presented in the answer here (10uF) is significantly smaller compared to the value presented in the original question (470uF). This is one of the benefits of using a self-latching pair that is triggered by a pulse. The down-side is that self-latching pairs are susceptible to unwanted turn-on; hence why C2 is required.

Provided that: (a) the supply is 12V, and (b) the current in the relay coil is less than, say, 100mA, then none of the components here will require a heatsink. All resistors can be 1/4 Watt. D1 is a low-power zener (1/4 Watt), in the range 4.7V to 8V. Q1, Q2 can be low-power transistors as shown here, or equivalents such as BC547, BC557.

Refinements The time delay is sensitive to the supply voltage - the higher the voltage, the shorter the time delay. To make the time delay independent of supply voltage, replace R3 with a constant current source. An example of a suitable circuit is provided below; select/adjust R2 to set the current into C1.



simulate this circuit

This topology uses different paths to charge and discharge the capacitor:



V2 is just a control signal which closes the switch at 1 s and opens it at 10 s. V1 is fixed at 12 V.

R1 can be a trimpot which selects the charge time. Higher R4 values cause faster Q1 turn-off, but we are obviously limited since R4 is in series. These are simulations for 0, 10 and 22 \$\Omega\$ (\$V_b\$ and \$V_e\$ refer to Q1):



With the relay inductance we may run better simulations.

If a cheap TL431 fits your requirements of "discrete component", this can be made more robust (less dependent on the BJTs).