Zener + LM358 Low voltage detect and cutoff

[Stefflus]

Hi!

I'm working on something involving a water pump running off solar.
The pump is small, 300mA or so, and has some fancy internal circuitry so that the pump doesn't work properly if the voltage drops below 11v, when it reaches this threshold it suddenly barely moves at all. I've had a look inside one of these pumps, and it is "MOSFET-driven", so I'm guessing the producer did something deliberate to protect them from overheating.

Anyhow.
Some of the users aren't as vigilant as myself about not running the lead acid batteries too low, so 11v can happen.
So I want a low voltage detection and cutoff, I want it cheap and I want it now. 

Here's where I'm at:



R2 and R3 sets a Vref of roughly 0,2v @ Vin=11v.
So when Vin=Vzener+0,2v the outout goes HIGH.

But since this is solar things happen slowly giving noise ample time to create havoc. So this needs a Schmitt input.
Hence T1 in the drawing, it adds hysterisis to the rising part of the signal.
To get hysterisis both directions I tried a mirrored PNP approach on the high side of the voltage divider, but as feared it just oscillated violently until something broke.

Turning Vin up on bench supply rather quickly gives this rising edge without hysterisis (T1, R5)


With:


The remaining problem, the falling edge:


Any thoughts on adding hysterisis to the falling edge as well?

[Delta]

Put Vref on the + input and the sense voltage on -. Use the output to directly pull the hysteresis resistor high and low. Then invert the output for your cutoff signal.

[Marinated]

Not directly related to your question, but are you quite sure about tying off the unused opamp that way? You'd get better results by connecting the output to the inverting input, and the noninverting input to a well-defined voltage somewhere between the rails, e.g. V+ / 2.

[Audioguru]

An LM10 is an adjustable voltage reference plus an opamp in an 8-pins DIP package.

[Stefflus]

@Marinated: Nope, not sure at all. It's from a quick Google search. I've watched a Texas Instruments video on how to do that properly, but couldn't remember.
Ah well, now I got to use it anyways.

@Audioguru: Thanks for that, I'll get a handful of those next time I order.
This was a whatever-I-had-at-hand kind of thing.

@Delta: Thanks again, works a treat!

[mikeselectricstuff]

MCU reset chips are a very simple & cheap way to do voltage detection. They don't go up to that sort of voltage but you can power from a resistive divider - current draw is only a few uA.
e.g. MCP100, TC54 etc.

[Zero999]

The LM358 is not a comparator but an op-amp and using it as a comparator is not good practise. You should use a comparator, such as the LM393 instead.

I'd use the TL431 with a MOSFET..


https://www.eevblog.com/forum/projects/power-the-load-only-when-car-supply-voltage-is-above-14-volts/msg1140485/#msg1140485

[Cerebus]

Not directly related to your question, but are you quite sure about tying off the unused opamp that way? You'd get better results by connecting the output to the inverting input, and the noninverting input to a well-defined voltage somewhere between the rails, e.g. V+ / 2.

In general that's correct and the OP has clearly got his inverting and non-inverting inputs muddled up on the spare amp. But, in the case of the LM358 you can get away with wiring the non-inverting input to ground/the negative rail instead of the more normal mid-supply.

The LM358 is an oddity being designed ab initio as a single supply op amp and having an output stage that's unusual and won't pull any additional current if its wired like that. The output stage is a PNP darlington pair on top with a single NPN at the bottom and a 50 uA constant current sink connected to the output so that for light loads the output can be pulled all the way to the negative rail - which is the case if you wire the spare amp non-inverting input to the negative rail and the output to the inverting input. As an aside, the input common mode extends all the way to the negative rail too.

Don't try this at home with any old op amp until you've read the data sheet. In general, follow Marinated's advice if you don't know something specific to the op amp at hand that says it's OK.

[Seekonk]

Since you have that second opamp, I would add a time delay if at least 15 seconds. Any motor is going to cause a current surge at startup and this will prevent cycling when the battery is near the low cut off. Just feed the next stage with a large resistor and cap to common and repeat the circuit. Big fan of the TL431, got a couple hundred, but I don't build these circuits any more. I prefer to use a NANO where I can dictate exactly what I want and not mess with changing resistors. I have a fridge I run off solar and I won't even start it unless the battery is at 13.5V ie it is sunny. Too many drag a battery down too low.  And of course you can buy China  low voltage disconnects with two set points for only $9.