# Switching between 5V Power supplies with minimal losses [duplicate]

I'm designing a circuit that allows one to switch between two different 5V sources, which is then fed into a 3.3V regulator IC. I'm targeting low cost and low board size, and I'll take a solution around $1-$2 in cost. Power supply is an integrated 5V rail on a system and the 5V USB rail for debugging.

Switch between 5V power supplies?

The answers there aren't satisfactory for me. The voltage drop on the diode brings the 5V input too low with fluctuations on the 5V rail (ie a lower voltage 5V rail may exceed the bottom limit of my LDO). I'm mainly just wanting a solution that prevents back-feeding of any power source. I also want power to only come from USB if it's plugged in.

• If you say those answers aren't good enough because you can't afford the voltage drop, what LDO are you using? What is the 'fluctuation' on the 5V rail? What circuits have you come up with so far and why are they not good enough? Does it have to be automatic? Or can a user manually change the supply rail? Also, note on the second paragraph of the answer you linked mentions "If you can't stand the voltage drop, then use FETs". What is wrong with that solution? – MCG Jul 2 '19 at 15:15
• How is a .4V drop from a schottky diode (or a FET for that matter) too much for a 3.3V regulator? If you're using a linear regulator all of the excess voltage is wasted in the regulator anyway. EDIT: MCG beat me to it. – Stiddily Jul 2 '19 at 15:16
• @MCG I'm using an MCP1812 and mainly want to prevent power draw from the internal 5V if USB is plugged in. I asked a friend and was told that just putting diodes on the inputs is suboptimal. – QuyNguyen2013 Jul 2 '19 at 15:18
• So a switch which acts as soon as a voltage is on USB? – Arsenal Jul 2 '19 at 15:23
• The MCP1812 has a dropout voltage of 600mV MAX. That means your input can go down to 3.9V and still have 3.3V out. This means diodes will be absolutely fine to use. The maximum output of the LDO is 300mA, and pretty much 99% of USB ports can handle a minimum of 500mA. So I am failing to see where the problem is? Did your friend explain why it is suboptimal? And if your friend knows this, surely they have a solution? – MCG Jul 2 '19 at 15:24