Maybe it was a moment of inattentiveness decades ago in a classroom, or that back then the whole world assumed that you had ±15 V rails on everything, but I can neither recall or find on the web how an NPN BJT behaves (large signal) when the gate is positive with respect to the collector. Clearly the transistor is "on", but with the C-B junction forward biased, what is the collector current?
The specific application is for the PPS output from a BG7TBL GPSDO that outputs ±6 V that I need to convert to a level appropriate for a 3.3 V GPIO (input). While a MAX3232 can handle the serial lines, its nominal 300 ns delay (with un-spec-ed jitter) is "bothersome" for setting up an NTP server. (Rising-edge timing is "important", I believe, but may be wrong on this.)
I had thought that a simple NPN switch could handle limiting the voltages to 0-3.3 V when driven off the MPU's 3.3 V supply.

With ±6 V on the IN, it seems like it "should just work"
- +6 V on IN, ~
+3 V +0.6 V on base, transistor turns on (PN2222 td ~10 ns, tr ~25 ns, so expect ~35 ns), OUT drops below 3.3 V logic threshold. - -6 V on IN, ~ -3 V on base (within PN2222 BVEBO of 5 V) transistor turns off, OUT rises above 3.3 V logic threshold.
Edit: As noted by @glarsson below, the B-E junction will conduct once forward biased, limiting the VBE to somewhere around a volt or less, depending on transistor characteristics and operating point. Thanks for dusting off the old memories. Suggestions for a "better" transistor still welcome.But what happens if the drive on IN is greater? RS232 receivers need to be able to accept ±25 V and I've got other timing devices that run off 24 V, so this may be more than just a desire for filling in my understanding.
Now, when the IN goes high, it raises the base above the collector, forward biasing the C-B junction. Does this then drive current into the V+ rail, or is the physics of the transistor such that the current across the B-E junction keeps the current flowing as it does when the C-B junction is more typically biased?I assume that in operation at these voltages (putting -12.5 V on the base) that I'll need to clamp the base to ground such that BV
EBO of ~ 5 V is not exceeded.
Finally, any better suggestions either for an NPN transistor that is notably faster than a PN2222, or a different circuit topology? I'll admit that my circuits courses were at the dawn of FETs and TTL still reigned as king, so FET switches aren't second-nature for me.