Why is the Relay "Chattering"?

Question

I have this circuit which measures if there is under/over voltage from the AC mains voltage and it consist of two parts:

Part one: measuring the input voltage (I had it cropped because it is not the main issue).

Part two: driving the relay with a red,yellow and green LEDs.

Red LED: it means there is under/over voltage. Yellow LED: A delay to check whether there is under/over voltage. Green LED: after the delay ends this means the Voltage is stable and The relay is powered ON.

The problem with this circuit, is that the relay is "chattering" after the delay ends (it gets back to the first state of the yellow LED delay) especially with inductive loads like motors(even as small as 60 Watts)

so my question: why is the "chattering" is happening and what is the solution for it?

Notes:

• VZ is Voltage from a Zener diode which equals to 27 Volts.
• This device works on a capacitive power supply.
• Under/Over voltage are intended to be: 175 V and below is under voltage and 268 V and above is over voltage.
• The Button is used the kill the delay time by discharging the capacitor.
• I soldered on the PCB an SMD un-polarized ceramic capacitor (C3).
• Delay goes for approximately 3 minutes until the Green LED turns on

The Circuit:

Sorry if I forgot to state/clarify anything, please feel free to ask.

Answer 1

When your load is applied the measured voltage will drop a bit (or maybe a lot) so you can get chattering as a result of that. So it's important you have adequate hysteresis for the lower trip point. You should have some for the upper trip point too, but it's not as important since the overvoltage will further increase when the load is removed.

I would also suggest using a Schottky diode only for D7. A lot of chips behave rather badly when you drag any pin below ground. Or use a series diode, losing a bit of voltage. Or drive the relay with a transistor.

Answer 2

Seems you need what is known as a "window comparator". This is just two separate comparators, each looking at the same input signal and comparing it to separate thresholds, Vlow and Vhigh. This provides two separate outputs: one to tell you if the input is below Vlow, and another to tell you if the input is above Vhigh. The benefit of this compared to your existing design is that you get full control over all the variables involved: you can independently adjust both the threshold and hysteresis for each comparator.

Here is a link to a decent app note for a window comparator:
https://www.ti.com/lit/ug/tidub01/tidub01.pdf

You will note that the schematic in this publication has two problems in your application:
(a) The outputs are connected together. I suggest keeping them separate.
(b) Does not include any hysteresis around each comparator; I would suggest adding a small amount of positive feedback (hysteresis) to ensure stability; say, between 0.1 and 1% should be OK, this depends on precisely where you want the output to trip.

Some simple logic after the comparator outputs can then be used to control the three LEDs and the relay as required.

UPDATE AFTER OP IMPROVED THE QUESTION

Your circuit needs hysteresis on BOTH comparators. So you need to add four (4) new resistors, as shown in the mark-up schematic below:

Here are some values that I suggest you try for each resistor. These are selected based on providing 1% hysteresis for each comparator.

RH1, RH3: 10kΩ
RH2: 10MEG Ω
RH4: 10MEG Ω

NOTE: The op-amp you have used for the comparators (LM324S) may not be sufficiently accurate for this application; the input bias currents (90nA typ, 250nA max) are not insignificant when compared to the currents associated with the hysteresis (~2 uA). This may cause the hysteresis to be very dependent upon the op-amp input current, which is not a tightly controlled parameter.

You can improve this situation by placing a new op-amp as a voltage-follower buffer after C2 to provide a low-impedance for the signal to the window comparator, and reducing the impedance of the signal to each comparator to about 1kΩ (it is now at 40kΩ = R4 + RH1, & RH3). This change allows RH2 and RH4 to be reduced significantly to null out the effects of op-amp input bias current variation.

Since the LM324 package contains four (4) op-amps, you should have a spare op-amp available to do this. If you don't have a spare op-amp available now, then I suggest re-purposing U1:3 for this role, and replacing U1:3 with some simple transistor logic to drive the outputs (LEDs and Relay) as required from the outputs of the comparators.