I'm wondering if a 3.3 to 5 volt level shifter would provide more robust serial communication between the GPS and APM.   I don't see one, looks like the tx/rx lines just go straight between the 5v 1280/2560 atmega and the 3.3v gps.  Certainly may not be required, it has worked fine, usually? 

    I recently added a switchmode BEC that has a little higher voltage supply, about 5.5v and that seems to cause the issue.  Could the higher VCC of the APM prevent the serial data high (3.3v at the most) out from the GPS from being recognized reliably by the APM?

     Long story short, have flown a bixler successfully with the following powered from the ESC BEC:  the 1280 APM, Oilpan IMU, MediaTek GPS, Xbee, magnetometer, turnigy receiver, two servos and a beeper.   The voltage at the APM and servo rail was 4.8 or so.    (the bixler with a Radian 2meter wing floats well, but doesn't go as fast and can't get upwind too well)

     I put the bixler aileron wing back on, and the load of the two extra servos causes the older PPM to detect brownout and servos twitch.  (my old 1280 board has the brownout detection level set higher, I ordered a 2560 board and will use this old one for HIL, or reprogram the PPM/328 atmega).   In the meantime, I unplugged the ESC BEC and added a separate switchmode BEC, which keeps the voltage sufficiently high to avoid that issue.     But then GPS wouldn't lock at all, or the light on GPS would show lock but the APM would not give the red led and groundstation would not show GPS fix.

      I checked the GPS a couple ways, and pretty sure it's fine.   I re-updated it to the 1.6 firmware and it works over FTDI cable with the MiniGPS test program.  Not sure what it had before, must have been 1.6 because that's the APM code I have been using (compiled 2.24 with only the MTK16 selected so will fit in the 1280), and it worked fine.

     I also checked the voltages on the GPS adapter board:  the 3.3 supply and VCC pin on the GPS reads 3.3v,  regardless of the '5v' supply from APM (4.8 or 5.5)

    Today I flew successfully again, using both a 4cell NiMh pack (not fully charged) and the old ESC BEC plugged in, supply voltage about 4.8 or 4.9,  and the GPS locked and worked correctly. 

     All worked correctly, except the wimpy bixler spar broke.  If the wings had not been taped on, it would have been a sad flight over the swamp.....

     Although simple solutions abound  (reprogram the PPM so it is fine with a lower system voltage like even 4.5,   get a BEC that puts out 5 or so and not higher),   would it be best to have a transistor on the GPS adapter board to shift the 3.3 level back up to the 5v on the tx (gps out) line?   The other direction would work as long as doesn't hurt the gps, which it seems not to.