Selecting IC for auto-switch to battery, in case of mains(dc) absent

[Vindhyachal.takniki]

1. I have a system, where:
a) Input SMPS is 11-14V, where it is used to charge battery with 2.5Amps and supply 12.5Amps to load also.
b) SMPS is used to charge lithium battery via LM3478 and dedicated charging IC.
c) When SMPS is present, it should supply to both load(12.5Amps) & charge battery.  Battery should not be connected to load.
d) In case SMPS is absent, emergency , then load should shift to battery in that case only.




2. Buck-boost LM3478 is for maintiaing voltage 12.6V  & boost LM5155 for load voltage 12.3V/12.5Amps.

3. Precondition is:
a)  Shift battery to load in case of smps absent only.
b) GND need to be common for all, so high side fet need to be chosen.

4. I checked Linear Technology LTC4414,  it supply this diagram:
a) Can i avoid diode UPS840(or whatever equiavalent of 12.5A), since SMPS will not allow reverse current in any case?
b) It has P-channel mosfet for drive, my laod current is 12.5Amps, so is there any other IC or way where I can use N-channel on high side?
c) Any other better solution for this?

[rohmeooo]

4) a. SMPS may not allow reverse current when operating normally, but it may not have that functionality without input power. Basically, is it happy with voltage on it's output and nothing on it's input. 
4) b. You should be able to find a suitable PMOS for 12V/12A. eg SI7157DP-T1-GE3 has <2mOhm resistance for 288mW of loss at 12A (.2%), seems reasonable to me. Or browse for better here https://www.digikey.com/short/z280z7
However, these devices actually operate in a sort-of linear mode, so you can't just put a super low resistance FET down and expect it to operate at RDSON.
Look at V_fr and V_RTO in the electrical characteristics table. Basically, the FET will regulate to 10-32mV forward voltage--that's your "ideal diode"'s forward voltage, irrespective of min RDSON.  So whether it's got 1A or 12A, it'll drop 32mV.  at 12A that's 32mV/12A = 2.7mOhm.  Basically, it does you no good to get a FET lower than ~2.7mOhm.  In other words, at 12A you're losing 12A*32mV = 384mW--doesn't matter how many FETs you put in parallel. Seems reasonable to me (still ~.3% loss)


There are similar devices with a charge-pump for driving NMOS. Or with lower forward regulation voltage. Don't know off the top of my head but "power path controller" "power OR controller" "ideal-diode controller" are your search terms.
Also some with 2 outputs or you can just put 2 of the LTC4414 in parallel. See figure 1 of the LTC4414 datasheet. Bear in mind the higher voltage prevails so your SMPS will have to always output more than the battery.


Why not always run from the battery, letting the input power provide as much current as is needed to keep the battery topped off?