Need help in designing a 3.3V input voltage astable multivibrator circuit that would generate a 1-second duration pulse periodically every 3 hours for microcontroller reset. Need it to be power efficient as it would be used for solar set up.

 

hi AT,
Welcome to AAC.
Have you considered using a low cost ESP32 module as a Reset sleep wake-up reset
Check this link.
E
https://randomnerdtutorials.com/esp32-deep-sleep-arduino-ide-wake-up-sources/

 

hi AT,
Welcome to AAC.
Have you considered using a low cost ESP32 module as a Reset sleep wake-up reset
Check this link.
E
https://randomnerdtutorials.com/esp32-deep-sleep-arduino-ide-wake-up-sources/

I need a physical hardware reset separate with the MCU.

 

Hi,
The ESP32 I am suggesting would not be the main MCU, but it will be dedicated to generating the Reset pulse every 3 hrs, nothing else.

It is the Reset generator.
E

 

Hi,
The ESP32 I am suggesting would not be the main MCU, but it will be dedicated to generating the Reset pulse every 3 hrs, nothing else.

It is the Reset generator.
E

It's an overkill I guess? What's in my mind is using a capacitor to charge/discharge then trigger a BJT to generate a pulse when a given voltage level is met during charging of the capacitor.

 

A multivibrator using RC components to set the frequency and pulse width is likely to have a wide tolerance (can be adjusted with pots or by selecting resistors). A digital solution, using a couple of counters and some gate(s), clocked by a quartz crystal (such as a 32768 Hz watch crystal) would do a much better job and would not need programming.

Would you consider a digital solution or does it need to be analog?
If analog, how much "slop" can the timing have?

 

A multivibrator using RC components to set the frequency and pulse width is likely to have a wide tolerance (can be adjusted with pots or by selecting resistors). A digital solution, using a couple of counters and some gate(s), clocked by a quartz crystal (such as a 32768 Hz watch crystal) would do a much better job and would not need programming.

Would you consider a digital solution or does it need to be analog?
If analog, how much "slop" can the timing have?

I need it to be as simple(less component) as possible and super power efficient. I'm considering using a LMC555 timer.

 

I need it to be as simple(less component) as possible and super power efficient. I'm considering using a LMC555 timer.

Or we can start from the circuit below. I tried to replace some RC values in Sim but I can't seem to get the result that I wanted.

 

hi,
Have you seen this option.
E
https://www.amazon.co.uk/tongfeng-W...e&qid=1734600527&sprefix=esp32,aps,550&sr=8-3

 

I don't think you will be able to get a multivibrator to generate a wave form with a 10800 : 1 duty cycle. I think using an 8 pin microcontroller such as 12F1840 or an attiny13 would be the lowest device count solution.
Les.

 

3 hours are impossible with RC itself or multivibrator.
Maybe 5min multivibrator + clock divider could be a solution.
But one stage of frequency divider costs about 18 parts. You will need 5 stages.

Here is one stage driven from multivibrator:

 

Using an 8 pun PIC requires only two parts. The PIC and a decoupling capacitor. Using its watchdog timer, it would be in sleep mode nearly all the time and use as little as 1uA. You cannot beat that with any BJT circuit. And if you need more accuracy, running off a 32KHz crystal is possible for maybe 10uA.

 

Using an 8 pun PIC requires only two parts. The PIC and a decoupling capacitor. Using its watchdog timer, it would be in sleep mode nearly all the time and use as little as 1uA. You cannot beat that with any BJT circuit. And if you need more accuracy, running off a 32KHz crystal is possible for maybe 10uA.

That's almost as funny as an "8 PIC pun"

 

It’s early in the morning. I get U and I confused a lot, not to be confused with getting you and I confused.

 

I'm an analog guy, but a tiny micro is definitely the simplest way to get the long time period with the short pulse you want.
The next simplest would be a timer using a CD4060.

 

Agree. Didn't someone used to make a uC in a 3-pin TO-92 package?

Anyway, back to geezer parts, a total of 18 connected pins -

1 - CD4060
3 - resistors (delay period, output period, oscillator protection)
3 - capacitors (delay period, output period, power supply decoupling)

Varying one resistor sets the long delay period. Varying another resistor sets the output pulse width.

ak

 

I don't think you will be able to get a multivibrator to generate a wave form with a 10800 : 1 duty cycle.

Depends on the multivibrator. Two transistors or 2 MOSFETs, yes but difficult. The ultra-slow edge rate would be a problem for the single-transistor stages. But a +/-10%, 22uF ceramic cap, a 4.7 M resistor, and 1/6 of a CD40106 Schmitt trigger hex inverter should be close to the desired delay period with very stable operation. Add a differentiator pulse former and another inverter and you have the desired output. As above, the delay period and output pulse period are independently adjustable. With this approach, the oscillator can be symmetrical, with each half-cycle being 540 seconds. 16 connected pins.

Another approach would be to replace the differentiator with a diode and resistor in the oscillator stage feedback path. Same number of components, but the oscillator capacitor would have to be twice as large because it is doing the entire delay period in one "half"-cycle.

ak

 

Varying another resistor sets the output pulse width.

Where would that be with the CD4060 circuit?

 

On the oscillator, unfortunately it would adjust the timing of everything.

 

I also join the chorus that a small microcontroller, those optimized for nano-power operation, is by far the simplest and cheapest way to go.

I used one to flash a LED to locate a light switch on a dark basement. With the help of a joule thief, the circuit would be operating continuously for over a year on a single AA cell.