That is what R9 (the hysteresis setting) is supposed to take care of. As designed, if this circuit is connected between a discharging battery bank and a load, the load will be disconnected (PFET turns off) when the voltage decreases below Vlow. As the battery voltage increases, the voltage at V(in) will have to go up by some ΔV to Vhi (controlled by R9) before the load reconnects to the battery (PFET turns on) automatically. This is the way the circuit is now, but it sounds like I guessed too low for ΔV
If the load consumes many amps, the voltage drop from the internal resistance of the battery can be larger than ΔV, in which case the circuit will "chatter", meaning it will cycle on and off repeatedly. This happens because as the load is turned off, the battery voltage comes up because it got relieved of the large load. That higher unloaded battery voltage is higher than Vhi, causing the circuit to turn back on...
Seems like we have several choices:
1. Increase ΔV. You would measure the actual voltage drop at the battery terminals when the inverter load is turned on/off, and then we will modify the circuit to increase the ΔV (by playing with R9) so that Vhi comes only after the battery has been charged. i.e. Vhi must be higher than the voltage jump as the PFET turns off...
2. Modify the circuit such that it requires an extra input to start/arm it. That way, it would turn off the inverter load automatically the first time the voltage dips below Vlow, then just sits there until you push a "start" button. You would wait to push the start button after the battery bank has been sufficiently recharged so it is ok to run the inverter again.
3. Make the cycle totally automatic. Modify the circuit so that it has two-independently adjustable trip points: Vlow similar to now. Vhi has its own trim pot. You would adjust Vhi to turn on the inverter only when the battery has been recharged to a voltage like 14.2V
Please advise how you would prefer to proceed.
Friendly dig: You see now why I always breadboard a circuit before making it "permanent". Here is an example where the circuit design met the original specs, but you hadn't considered all of the requirements, so coming up with a final circuit is an iterative process. Much easier to move a wire or to plug-in a new resistor on a white board.
If the load consumes many amps, the voltage drop from the internal resistance of the battery can be larger than ΔV, in which case the circuit will "chatter", meaning it will cycle on and off repeatedly. This happens because as the load is turned off, the battery voltage comes up because it got relieved of the large load. That higher unloaded battery voltage is higher than Vhi, causing the circuit to turn back on...
Seems like we have several choices:
1. Increase ΔV. You would measure the actual voltage drop at the battery terminals when the inverter load is turned on/off, and then we will modify the circuit to increase the ΔV (by playing with R9) so that Vhi comes only after the battery has been charged. i.e. Vhi must be higher than the voltage jump as the PFET turns off...
2. Modify the circuit such that it requires an extra input to start/arm it. That way, it would turn off the inverter load automatically the first time the voltage dips below Vlow, then just sits there until you push a "start" button. You would wait to push the start button after the battery bank has been sufficiently recharged so it is ok to run the inverter again.
3. Make the cycle totally automatic. Modify the circuit so that it has two-independently adjustable trip points: Vlow similar to now. Vhi has its own trim pot. You would adjust Vhi to turn on the inverter only when the battery has been recharged to a voltage like 14.2V
Please advise how you would prefer to proceed.
Friendly dig: You see now why I always breadboard a circuit before making it "permanent". Here is an example where the circuit design met the original specs, but you hadn't considered all of the requirements, so coming up with a final circuit is an iterative process. Much easier to move a wire or to plug-in a new resistor on a white board.
Last edited: