Adding current limiting to a buck converter (MT2499) with INA213 and TL431

[Tungsten]

Hello, somewhat junior designer here.

I am designing the power supply section for a circuit that will be mounted on a sensor node up a pole. I have chosen sealed lead acid batteries for solar backup power, and the circuit will also be connected to grid power.

There is an AC to DC converter in the design (HLK15M15C) that outputs 15V. To charge the battery, I selected the MT2499A IC because I’m using it elsewhere in the circuit, though I'm open to swapping it for a better IC. It is a generic 3A buck converter. The MT2499A will be configured to output around 14.2V.

To prevent the battery from drawing too much current when it’s empty, I want to have some current limiting at the output. I’m aiming for a CC/CV charging process. I’ve explored current limiting with transistors and shunts, but voltage drops make it difficult to possibly maintain 14.2V at the output. I also considered adding just a current limiter after the 15V and charge the circuit, but inconsistencies in voltage (e.g., possibility of reaching 14.6V) could compromise the battery’s longevity.

My current solution is to add a second feedback signal to the buck converter's feedback pin. This way, I can implement current limiting without losing voltage. I have seen similar approaches discussed in videos and forums.

I plan to integrate an INA213 to sense current. Once the current exceeds the threshold, I want to send a signal to the feedback pin. For simplicity, I intend to use a TL431 comparator, which compares the input to its internal 2.5V reference.

The INA213 is a current sense amplifier with a gain of 50. The voltage drop across the shunt gets amplified by the INA213, once this exceeds the 2.5V threshold, the TL431 will trigger. This will turn on the optocoupler, and the optocoupler’s output will use a voltage divider to reduce the supply voltage to a level the feedback pin can handle. The goal is to ensure the voltage is high enough to bias the diode (exceeding the MT2499 feedback voltage by at least 0.7V).

Schematic attached at the end.

As said, this is just the power section of the PCB, there are other items already in the design and on the same PCB. I can't order the components right now and test them separately.

My questions/concerns:

Is there any reason this approach wouldn't work with the MT2499A?

Will the TL431 work correctly when there’s an optocoupler LED in its path?

What disturbances or surprises could the optocoupler introduce on the voltage divider?

I’m open to alternative suggestions or IC recommendations, as I’ve invested significant time in this design but am not attached to it

Thanks in advance!

[HackedFridgeMagnet]

My thoughts.
Why the opto coupler? Im not sure you need it.
Or the reference.
There is the reference in the FB input.
So you just need voltage control or current control depending upon which is too high. Hence the diode.

I think you need a buffer on the voltage. The current sense amp is good.
Then you have 2 low impedance outputs, massage them to limit the fb pin interdependently. something like full voltage gives 0.8 volts (whatever it is) full current gives the same.
Then you need to 'or' them. Diodes might be ok but you have to account for the junction voltage. Maybe using ref pin on current sense amp.
Otherwise some sort of ideal diode would work.
Whatever you do I would simulate it first. Use one of LTSpices switching converter circuits to imitate your converter chip.

Hope that makes sense

[HackedFridgeMagnet]

Also if you are charging a battery that is already on a load. You probably want to measure the net current to the battery. Not the current to the whole system.

[Tungsten]

Hey,
About the having a buffer on the voltage sensing, I have that part as recommended on the MT2499A data sheet. So I think it would be ok.

Reference & optocoupler: Having just a diode makes it difficult for me to get some decent precision on the current sensing. I have not found a diode datasheet with a more or less constant forward voltage (are there ones with this property?). So I opted to have a comparator that will compare the output of the INA213 which I settled on using the simple TL431.
The optocoupler is my attempt at a crude way of interfacing the TL431's comparator output with the rest of the circuit.

But I had not thought of ideal diodes and have never used them. Are they available and cheap, are they as ideal as their name suggests?

[Tungsten]

Found this
CN3768: Dedicated lead acid charging IC that is based on SMPS approach. It also has the needed CC current setting capability.