ON few hours OFF few hours timer circuit

Question

I am very new at this, but I am looking for a simple circuit (no microcomputer please) that will turn ON an LED for 4-5 hours, then turn OFF for 20 - 19 hours. Then it will automatically repeat this loop for as long as there is power. This will be battery-powered (3.7VDC). I would appreciate any help anyone can provide—many thanks in advance.

Answer 1

As others have mentioned, the best and simplest solution would use a microcontroller. There are many microcontrollers, like the Arduino, that are extremely easy for a hobbyist to program.

But, just for fun, could we do it without a microcontroller, or any digital logic at all?

Sure.

For a starting point, we could use an opamp-based relaxation oscillator (https://en.wikipedia.org/wiki/Multivibrator#Astable_multivibrator_using_an_op-amp)

From the derivation in that article, the length of either phase of the oscillator is \$T=RC \ln[\frac{1+\beta}{1-\beta}]\$

In this topology, the positive and negative phase of the oscillator have the same length, but we can fix that: just use a MOSFET to change either the feedback resistance or the load capacitance based on which phase we are in.

For example, if we call the phase in which the voltage at the negative terminal is increasing the "charge" phase and the other the "discharge" phase, and we have a different feedback resistance in each phase, then:

\$T_{charge}=R_{charge}C \ln[\frac{1+\beta}{1-\beta}]\$ and \$T_{discharge}=R_{discharge}C \ln[\frac{1+\beta}{1-\beta}]\$

Here's an example circuit with \$C=400nF\$, \$R_{charge}=5k\Omega\$ and \$R_{discharge}=10k\Omega\$.

Simulate it

As we expect, the positive and negative phases of the output are ~ 2.2 ms and 4.4 ms respectively.

To make your timer run for hours, the passives should be a few orders of magnitude larger. With \$R_{charge}=1 M\Omega\$, \$C=10mF\$, and \$\beta=0.8\$, you can have a charging time of ~ 6 hours.

At this point, you have some other problems... you need to check that the MOSFET's gate is rated for the opamp's output voltage, make sure you don't have any leakage resistance on the order of a megaohm, and take extra care that your 10 mF supercapacitor doesn't explode in your face.

But I'm hoping you don't build this at all, because as mentioned, an Arduino can do it better. :)

Answer 2

This is how I would have done this in 1976. You can only expect about 1% accuracy with a 555 timer circuit. So, if 14 minutes per day accuracy isn't good enough, you may need to use a crystal oscillator.

You can calculate R2, R3, & C1 from the datasheet. C1 must be a stable cap, a mylar or something similar.

U1 and U2 could possibly be combined into the chip that I referenced in the comments.

^{simulate this circuit – Schematic created using CircuitLab}

Edit: R2 should be a fixed resistor in series with a pot. You don't want to use a 1M pot, it will be too hard to make fine adjustments.

Edit2: I built the circuit, it works fine.

Calibrating is tricky. Ideally, you want to use a quality lab frequency counter with a period measuring function. My scope will measure period, but with less accuracy. Without an instrument, you need to get creative. Temporarily change the U2 output to Q1. Then measure the time to get 10 LED pulses with a stopwatch. It should be 105.5 seconds. You should be able to set the initial accuracy to a few tenths of a percent with this method.

The 7555 is slightly sensitive to power supply voltage. When I changed the voltage from 5V to 4V, the period was reduced by 0.5%.

The period will also change with temperature. How much will depend on the quality of your components. But, if your circuit will be used inside, this should not be a major factor.