That would work, but gives too many chances to get things wrong.
This version will reduce the chances (but still leaves the option of feeding 5V on the 3.3V rail - maybe add a zener to take care of this?) It’s based on using standard 0.1" headers and 2-pin shunts or the equivalent.
The 3-pin block will let you select the TTL connector or USB connector as the source of power (and prevent you from connecting both at once). The 4-pin block with jumpers placed vertically would tie in the regulator for 5V input, and with jumpers horizontally, would bypass the regulator for 3.3V input.
I want to use a power protection circuit on my board letting only 3.3V pass through (the other possibility is 5V that I want to block). Is the schematic below correct? Is it dangerous (especially for the Zener)? Is there any other suggestions/modifications or other advised schematics that can do the same job?
First, let’s keep this topic on one thread; spreading it out is frowned upon, and is hard to follow.
I’m not sure crowbarring the input rail is what you want. If you do use a crowbar, add a fuse so that it will blow before something else.
What about using a buck/boost regulator instead of all of this. The LM3668 will work on 2.8 to 5V in, and give you 3.3v@1A. Precede it with an diode OR (or better yet, a FET OR like the TPS2115A) to automatically select an input.
In fact, I will use the circuit u proposed with the 3-pins jumper and the 4-pins jumper. To prevent the option of feeding 5V on the 3.3V rail (5V via TTLV and horizontal 4 pins block OR 5V via VUSB and horizontal 4 pins block) I will use the circuit I posted above on the 3.3V rail.
For your second proposition do you have a shematic or something illustrating that?