There is no reference in a 555 so it will only be as stable as Vcc.

@SpehroPefhany thanks 12V auxiliary supply(VCC) is well regulated coming from a very good buck converter. If VCC is stable will my circuit work in practice?

Yes, it should. You can get 4.02K and 1.00K 1% resistors, which is close enough, or you could trim the trip point to compensate for the resistors, 555 and 12V tolerance. But maybe that’s unnecessary.

Voltage reference + comparator?

@winny ah nice idea i did a quick google and most of the comparator ics seem to work at around 5V. (for example, LMV331W5-7 requires VCC of 5V). But I have access to only 12V and don't want to add another supply as it requires another buck converter. Will 741 opamp work as comparator? Can I give VCC=12V and VEE=0V. One input I tie to Vref and another to the battery voltage divider. Will this be better than 555 circuit? Cost wise/performance wise...

Hundreds, if not thousands of comparators available which will take 12 V as Vcc. DO NOT USE 741!

@winny thanks ill research more...

Because you do not say to what precision you need to determine your voltage cut-off point, any answer has an implicit "good enough" opinion buried in it. Also, as soon as you implement this you'll begin to appreciate the unique joys of trying to monitor battery usage, in that for almost all battery chemistries, as the battery dischages the output voltage sag is affected more by increasing effective resistance rather than decreasing effective voltage. So unless the load is very constant you end up building a "cut power on high load" device, not a "cut power on discharge" device.

Look at the LM339 for starters. Like @Winny said, there's hundreds. The LM339 is possibly the most easily available, it should work from either 12V or 24V, and while it's not known if it meets your requirements for precision, it'll probably add less error than using the 12V from a switching regulator as a reference.

@TimWescott your first comment is very very scary... so simply relying on voltage-capacity relationship doesnt work when the load is on? Here the load is not constant :( it is electric cycle. So I'll have to monitor the current also and work the actual discharged mAh? Looks very very complicated. Any shortcuts?

Btw my battery is 24V 24Ah LiFePo4 8s4p

@TimWescott I was referring to this voltage-capacity relationship footprinthero.com/wp-content/uploads/2022/05/…

@across Tim is (always) right. Except that I don't recall you saying you will be using it to cut off power. If this is just for "detection" (an LED indicator, for example) or if you will only be applying the test in a specific situation (cycle is off and you can apply a standard load) then you may find some inaccuracies acceptable. It doesn't cost a lot of money or time to try out something. I agree also with Tim about the LM339, except that just looking right now I'm finding the LM239 to be cheaper. Not sure why.

@across that's probably with no discharge current. With discharge current, the cell voltage will go below that -- possibly by a significant amount, if you're discharging the cell heavily. If your discharge current in low and constant, or if you can turn it all the way off and give the battery a minute to recover before reading voltage, then cell voltage is sorta-kinda reliable.

@jonk thanks yes I'm not actually cutting off the battery, just want to remind myself by beeping an active buzzer for 3 seconds every 30 seconds, after the battery capacity falls to 30%. I found many good ics: LM239, LM339, LM393. thanks to you, Tim and Winny. you all are awesome!! I'm going to try the one available locally soon and update my findings here if you're interested...

@TimWescott I see the complexity of estimating the capacity of an actively discharging battery. The current draw varies a lot. The cycle draws more current on ramps and less on flat roads. As a shortcut I'll try to use the worst case current draw and see if it works. If it doesn't I'll have to put a current sensor and monitor Ah using a cheap microcontroller. Yeah already appreciating the unique joys of monitoring battery usage haha xD

@across Towards the bottom of this answer I illustrate a circuit with hysteresis using the LM139. (Works same with the LM239/LM339/LM2901.) It provides WAY more hysteresis than you want. And it's not set where you'd want, either. But it gives an idea. Given that it uses an open collector output, you likely can get this to work directly okay with an LED (high efficiency, recommended though.) You should work out in your own mind a narrow range that balances noise immunity with other needs. (Resistor dividers have their own errors, too.)

@across There are other methods. For example, two BJTs and some resistors can get the comparison part done. It would need an added BJT for the LED driver part. Or you can use a relaxation oscillator that will blink an LED at different rates. You'd need to mentally calibrate yourself for reading the device. But I've used the idea before for similar purposes. And it works over a VERY wide range of input voltages so it may have more uses than you imagine, right now.

@jonk two BJTs and a Zener diode right? Ill try this first as this is the simplest of all. I have used exactly your same idea: Detect low voltage, then use it to remove reset of relaxation oscillator. In simulation it is fine with two 555 timers: One 555 for detecting low voltage. Other 555 for beeping active buzzer with duty cycle 2/30. prnt.sc/QWYJcX4yUXyx

@across No zener. The BJTs use their forward biased BE junctions for comparisons, instead. They become slightly out of balance when enough current arrives via an extra resistor to the base of one of them. That resistor would need to be adjusted based upon the actual BJTs you use. It's not a predictable circuit. But for cheap and easy, it's nice. If the BJTs are in the same package (duals are easily available) then that is better. I'm not recommending it unless cents count or you just like the idea.

@across The LM1/2/339 is better, though there a zener or TL431 is probably needed. I prefer a circuit powered by the battery you are measuring so that I never have to worry about replacing a battery in the tester, itself. So I'd really recommend you try to apply a circuit that is powered by your battery under test. The relaxation oscillator does that, beautifully. And it does it without a zener or a TL431 too. I really hate having to make a circuit larger than necessary for a battery and case to hold it, when it's not needed. I'd repurpose an inline fuse casing, perhaps?

@across Just crossed my mind. You could get a dual LM1/2/339 and use one of them for a voltage-independent current source (it's internally designed to operate that way from V+ to V-.) That, with a resistor (you'd need to calibrate it, of course) means you have a voltage-independent voltage reference for the other LM1/2/339 in the package. So just some resistors plus a package and an LED. It's a hack. But...