PID controller for a heater using a Raspberry Pi, SSR and Python

Question

I would like to PID control the temperature of a water bath using a RaspberryPi4, a solid state relay (SSR) and a heater (200W, 230VAC) using Python.

The IVMech library seems a good starting point.

As well as Simple-PID

I went for IVMech since I saw it quoted in other projects.

I can control my actuator by pulling my SSR low one of the RPI-GPIO pins.

Currently, I'm stuck with reducing the test_pid.py test code to the bare minimum required. Please see my code sample below.

Is it fine to just use the output of the PID calculation as the time to turn the heating element on?

In my case I don't want to cycle the ppwer so many times (sample time around 1-2s) since my application should last > 1year and I'm not sure if the SSR that uses a photodiode to for galvanic isolation can be cycled indefinitely. I read that the SSR supports around 100MIO cycles. For a switching frequency of 1Hz that would correspond to a lifetime of around 3years.

    import PID
    import time
    import numpy as np
    from scipy.interpolate import BSpline, make_interp_spline #  Switched to BSpline
    import RPi.GPIO as GPIO
    GPIO.setmode(GPIO.BCM)
    HEATER = 17
    GPIO.setup(HEATER,GPIO.OUT)
    
    P = 0.2
    I = 0.0
    D = 0.0
    L = 0
    sampletime = 2.0
    
    def measure_Temp:
        temp = 20.0
        return temp
    
    while True:
        pid = PID.PID(P, I, D)
        pid.SetPoint=25.0
        pid.setSampleTime(sampletime)
        END = L
        feedback = measure_Temp()
        pid.update(feedback)
        output = pid.output
        GPIO.output(HEATER, False)
        time.sleep(output)
        GPIO.output(HEATER, True)
    

Answer 1

Sleeping for a duration proportional to the output isn't the best way. The controller is "blind" while sleeping and you may overshoot.

To achieve what you want, simply sleep for 2 seconds after every measure / control cycle. Turn on the heater if the output is greater than 0, turn it off otherwise.

If you require a particularly high regulation accuracy, I recommend reading up more on how to achieve that, it is not a trivial problem but it has been well-researched.