FET switch

A standard answer would be a p-channel MOSFET, or perhaps a PNP transistor. But the MOSFET would have to be a low-threshold- voltage type, and the PNP would require a base resistor. So we have excuses to keep looking.

I think a good candidate would be a 1-gate logic buffer, a modern version of the old 74HC125 drivers. Consider 1G125 gates in the LVC family. These have Rout less than 30 ohms at Vcc = 3V, so a 2mA load would drop less than 60mV.

TI's 74LVC1G125DBV uses a sot-23-5 package, with 0.95mm lead spacing. Philips (NXP) part of the same size, 74LVC1G125GW, says it's in a sot-753 SC-74A package. TI's 74LVC1G125DCK uses a miniature sot-70 package, with 0.65mm lead spacing. NXP's 74LVC1G125G says it's in a sot-353 (TSSOP5) package. Sigh, whatever, no doubt both small 0.95mm parts will solder onto those pads, and ditto for the miniature 0.65mmm parts.

The 74LVC1G125DBV and DCK are 39 cents at DigiKey qty 1. The NXP parts are 30 cents, six cents in a full reel.

Think of these as a low-cost low-threshold p-channel MOSFETs. These are tri-state gates, so if you tie the input to Vcc you can use the tri-state control as if it was the MOSFET's gate.

__ 74LVC1G125DBV | \\ Vcc ---| >--- OUT |__/ O (Vcc and GND supply connections not shown) | control ---'

Speaking of low-threshold voltages, offhand I don't know of any discrete MOSFETs that can do as well, here're the datasheet Ron values at low voltages, down to 1.65 volts:

Vcc R_hi R_lo ---- --- ---- 1.65 112 112 ohms 2.3 50 37.5 2.7 42 33 3.0 30 23 4.5 22 17