MOSFET Heating Problem

[rks96]

I want to create a power supply switching circuit using LTC4412 https://www.analog.com/media/en/technical-documentation/data-sheets/4412fb.pdf. I am using the reference example shown in figure 2.

My goal is to switch between 12.6v battery and 19v (90w) power supply from old laptop and design functions correctly except for heating mosfets. Right now I am using IRF4905 as P ch mosfets https://alltransistors.com/pdfdatasheet_international_rectifier/irf4905lpbf_irf4905spbf.pdf.

When testing with 15v on wall adaptor input (from figure 2) and drawing only 10w mosfet starts to heat up. From datasheet IRF4905 can dissipate 170w (that's a lot so must be with additional heatsink).

My question is why mosfet is heating up so much with such a little power draw? Or MYB mosfet isn't switched fully on or off? I have seen smaller mosfet packages switching 90w on laptop MB but they whee N ch? Can someone recommend P ch mosfet that wouldn't need additional heatsink to switch 90w in this application (figure 2)?

PS mosfets were bought on Aliexpress so they might be fakes (also they look "used").

[magic]

Connect the power supplies, measure voltages at sources, gates and drains of the FETs.

If some FET has 12.6V between source and gate and still has significant difference between source and drain with less than 1A of current and is getting hot then it has to be fake.

[HB9EVI]

a schematic would help understanding what's going on.

if the FET gets hot when being used as switch, it's not properly turned on or off.
if you switch 19V (voltage applied on the source) and you have only 12V or 15V on the gate, it won't be turned off but rather be somewhere in its linear, resistive area (with 15V) or almost completely on (with 12V)

[rks96]

LTC4412 datasheet fig2:

[HB9EVI]

if I understand right from the short look to the datasheet, the STAT pin is an open drain, so if you connect it to the FET on the wall adapter side, you have to bias the gate with a resistor to source.
the drawn 470k R belongs there, and not connected to the load side of the FET controlling the battery side

when running on that supply, STAT should go low and drive the FET fully open, so you should measure a few 10mV between GND and gate.

[rks96]

@HB9EVI
470kR already is between G and S on wall adaptor side FET!

I think unrelated noob question here: In datasheet there is other configurations in witch 470kR pulls up STAT pin to VCC. Does it mean VCC on the input side of the circuit (battery or AC) or it means output side that goes from circuit to load side and if it is so why isn't it fully connected like in figure2?


Also SENSE pin is connected to different places?

[rks96]

Also, when I run the circuit from battery side, FET also heats up. This leads me to believe FETS simply are fakes.

[rks96]

Bought different mosfets and everything is working fine. Also used T7-H to test old mosfets and test shows 2ohm Rds. One of 10 mosfets even show 8ohm Rds!