Today I received PCBs of the 4A and 30A mag switch circuits. These boards are 2oz. copper and designed to keep PCB trace resistance to a minimum. I built all three of the 4A version with the following results:

1. Verified 20V operation.
2. Off-state current 22µA for 2 boards, 60µA on one board. The high off-state current did not increase with input voltage, which leads me to believe it is from a component on the 5V rail. Even this higher current draw is not really a big deal in the scheme of things. It doesn't seem to affect operation in any way.
3. Memory hold time is greater than 1 minute. This is way too long. I will decrease C4 to 100uF (1206 10V X5R ceramic) on future boards. I also increased R6 to 220Ω to get a bit more isolation for the LDO -- the larger resistor only causes a voltage drop of about 10mV.
4. Switch resistance measured at 11-12mΩ with a load current of 1.5-2.5A. I was expecting better, but it should be good enough. The power dissipation of the switch will be less than 300mW with 5A of load current. That's only 100mW for each output FET so temperatures should not get too hot.

A couple of days ago Digikey notified me that the AH180N hall-effect switch was going to be obsolete. I found a drop-in replacement -- S5716ACD1M3T1 by ABLIC. This new part has a CMOS output, 4uA typical supply current, and slightly better sensitivity with a 30Gauss trip point. The footprint is identical. The cost is $0.72 in lots of 25. I will use this hall-effect device after using up current inventory.

So this latest 4A switch seems to meet all of its objectives. I just have to install it into one of my planes and test it out in the real world.