RFC: low leakage UVLO (under-voltage lockout) for magsafe

[manekinekohello]

greetings humans,

My most recent hack was to place an Apple MagSafe 1 charging port into a Lenovo IdeaPad U430. It works, but I'm still working on controlling the charging LED over 1-wire, and making it compatible with Apple genuine/original adapters. This post is about the second part.

As described in http://www.righto.com/2013/06/teardown-and-exploration-of-magsafe.html, you need to follow a specific protocol to get Apple adapters to start: first, connect a 39k resistor, then wait for the voltage to climb before connecting your load. Essentially this is what a UVLO (undervoltage lockout circuit) does. But it also needs to not interfere with the detection part, i.e. the resistance should be much higher than 39k. In other words, it needs to have very low leakage.

I designed the following circuit and hope to get some comments. It works well in simulation. I set the threshold to around 15V using a 15V zener (D1). Reverse leakage current for a zener of this voltage is typically 5 uA.

C1 represents the MOSFET that would switch the actual load. It would be nice to have high-side switching so I still have to find/work out a circuit for that. The rest of the circuit is so simple really that it describes itself.

Simulation results: blue trace shows the input voltage: linear ramp to 20V and back down again with some noise. Green trace is output: it is practically zero (<1 uV) for Vin < 15V. Then it turns on and stays turned on because of the positive feedback (hysteresis). Because of the hysteresis, the turn-off happens at a slightly lower voltage than turn-on, this can be controlled with R5. The red trace shows the input current: < 1 uA even when close to threshold.

Comment if you would like updates :)

thanks for feedback!

[NiHaoMike]

Replace Q2 with a P channel MOSFET capable of handling the entire load and change R6 to 33k, then it's pretty much good to go. D2 is not necessary and can be replaced with a short.