low voltage cutoff with TL431, many variants -- specific question about one of them

Hello there,

This is my first post here

I don't know yet if it's common and/or required here to introduce oneself, so I'll keep it short; please feel free to jump to next paragraph if I'm being irrelevant ; or to ask more if you want. I'm JF ; Simon is actually my surname. I'm 27 and live in France. I'm a mechanical engineer but always liked to tinker with electronics. In the recent years I began to learn more and more and (slowly) to be a little bit more proficient. But enough about me!

These days I was browsing the internet looking for low-voltage battery cutoffs and I found several designs, and particularly three of them by MikeML.
Here is the list I made, if, by any chance, some of you are interested : **broken link removed**

I would have two applications of such circuits in mind, one is to help me measure capacity of (fairly old) lead-acid batteries, by first charging them and then discharging them through one of these circuits (and 12v halogen lightbulbs as a load), while logging the current. This way I could stop the discharge at the same voltage with each battery and it would be easy to compare apples to apples.

(The other one would be to make a homemade UPS for my modem, using another lead-acid battery, and a LV cut-off would be a good feature, but let's not get into that for now)

One of the classical ways to do it can be, apparently, to use a TL431 as a comparator and to drive a P-channel mosfet on the high side. I've seen several forum members recommend this solution in several places.

The circuit MikeML gives us here https://www.electro-tech-online.com...ttery-discharge-protection.94128/#post-752640 drew my attention. Same idea but with a twist ; the battery voltage is sensed after the MOSFET instead of before. As such, in the Off-state, everything turns off. So I guess the quiescent current would be very low, if any ; maybe leakage through the 431. Also, there is no need for a hysteresis resistor because the switch-off is permanent and manual switch-on is required with a pushbutton. That would be useful for the first application I just mentioned.

So, my question is : Wouldn't there be a problem, if the load is not perfectly constant ? I guess (haven't tried it yet) that a surge in the load could trip the switch because of the increase in the FET's voltage drop , even if the battery voltage hasn't changed.

Do some of you have used this particular circuit in a practical situation, and is the observation I just made relevant ? If so, what can be done to allow precise switching even with non-constant loads ? I would really like to hear your thoughts about it. I asked Mike by PM and he wanted me to post on the forum so everyone can discuss.

Thank you very much for your time, and looking forward to hearing from you,

jfsimon

jfsimon said:

So, my question is : Wouldn't there be a problem, if the load is not perfectly constant ? I guess (haven't tried it yet) that a surge in the load could trip the switch because of the increase in the FET's voltage drop , even if the battery voltage hasn't changed.

Click to expand...

I haven't used that specific circuit, but something like it, somewhat overengineered with a PIC because I hadn't come across that one which is better. Size the FET big enough and it's not a practical probelm, yes, in theory if you are just above the threshold and a pump kicks in on the load side you will be edged over the threshold a bit faster than you otherwise would be, but so what? Your battery is already getting low, and the aim of the device is to save your battery.

Look at power PMOS RDS(on) - they can be amazingly low. If load glitches bother you, consider a capacitor at the junction of R2 and the top end of U2 (the 5k pot) down to ground. Guess you'll have to hold the switch closed a few milliseconds longer

jfsimon said:

...
These days I was browsing the internet looking for low-voltage battery cutoffs ...
I would have two applications of such circuits in mind, one is to help me measure capacity of (fairly old) lead-acid batteries, by first charging them and then discharging them through one of these circuits (and 12v halogen lightbulbs as a load), while logging the current. This way I could stop the discharge at the same voltage with each battery and it would be easy to compare apples to apples.

Click to expand...

Use a momentary push-button and a 12V relay such that the TL431 holds the relay pulled in while the battery is discharging. Put the relay coil and TL431 on the load side of the relay contact, so no current is drawn after the battery the voltage sags below the cutoff point.

The relay coil current adds to the load current while discharging; zero current after cutoff. Use an extra set of contacts on the relay to run a wall clock to log the hours that it takes to discharge the battery to get an estimate of Ampere Hours. I think I posted that somewhere...

(The other one would be to make a homemade UPS for my modem, using another lead-acid battery, and a LV cut-off would be a good feature, but let's not get into that for now)

One of the classical ways to do it can be, apparently, to use a TL431 as a comparator and to drive a P-channel mosfet on the high side. I've seen several forum members recommend this solution in several places.
...
So, my question is : Wouldn't there be a problem, if the load is not perfectly constant ? I guess (haven't tried it yet) that a surge in the load could trip the switch because of the increase in the FET's voltage drop , even if the battery voltage hasn't changed....

Click to expand...

Since the voltage divider on the gate of TL431 is relatively high impedance, why not just turn it into a low pass filter by shunting the gate to ground with a large capacitor. You could create a time-constant of several hundred ms there to bridge across any momentary sags created by an inductive load.

Hello, thank you for your replies

Ermine,
you are right, in a practical use, with the battery being already mostly empty, I guess that it wouldn't make much of a difference. My question was probably more theoretical than practical . Anyway, thanks for the tips, yes I will look deeper into FETs, I have very limited experience with them for now.
As for glitch filtering with a capacitor, that sounds good. I should have included a schematic in the first place, but in the circuit I was referring to, U2 is the 431 and the pot is U1, so probably you mean U1 ? Out of curiosity, why not connecting the cap between the wiper of the pot (which is also the REF pin of the 431) and ground, instead of the top end of the pot as you said ? Thanks

Mike,
That's great, because I also thought about using just a relay, a 431 and two resistors for my battery discharge testing, as that looked like the simplest solution ; in this case the relay coil current isn't an issue, as the goal is to discharge the battery anyway. I doodled it the other day, I think you mean to have it wired **broken link removed** ? It feels good to hear someone with much more experience considering the same solution
As for the low pass filter on the REF pin of the 431, that's neat, thanks!

Well I think my concerns are pretty much addressed, I have to breadboard things a little now. Still, if someone else has built a low-battery cutout (or alarm) by other means, I would love to hear about how you did it

jfsimon said:

but in the circuit I was referring to, U2 is the 431 and the pot is U1, so probably you mean U1 ? Out of curiosity, why not connecting the cap between the wiper of the pot (which is also the REF pin of the 431) and ground, instead of the top end of the pot as you said ? Thanks

Click to expand...

Absolutely, I meant U1, my bad. I didn't know the input to the 431 was high impedance as Mike said, although it's obvious in hindsight looking at the resistor values above and below U1, so Mike's location is a better place for the cap, though it'd work okay where I meant it to be.

I did mine with a PIC, comparing the divided down main battery ref with a couple of diode drops ISTR, sampling every time the main loop went round. Had a low trip threshold VLO and a higher restart threshold VLO+about 1V, there was a solar cell on the battery. I worried about the load dropping off letting the battery rise on charge, 1V was good enough so it didn't chatter. I cut the battery off at about 11V, it was a wet leisure battery 80Ah. In the end I just pulled the power and shut it down, we had to send a runner out to swap the battery as the solar panel didn't cut it for recharging the battery for a remote WiFi repeater. It was cheaper to swap batteries than buy more solar We had to swap other leisure batteries for the electric fences on site anyway. The PIC also shut the WiFi rig down overnight as it was relaying pictures of our pigs taken with a Pi camera, which stopped working at night because there was no light. At least the pigs had the decency to sleep in the pig ark at night

jfsimon said:

...
That's great, because I also thought about using just a relay, a 431 and two resistors for my battery discharge testing, as that looked like the simplest solution ; in this case the relay coil current isn't an issue, as the goal is to discharge the battery anyway. I doodled it the other day, I think you mean to have it wired **broken link removed** ? It feels good to hear someone with much more experience considering the same solution...

Click to expand...

Just make sure that the relay coil resistance is high enough that the TL431 is within its max current rating.