Help to troubleshoot RTL crash

Please help me identify plane crash cause.
I have several years experiance flying Ardupilot sinceAPM time, all crashes have been user error or hardware. Limited expeience in LOG analizing won’t allow me to determine cause for these crash.
This time Pixhack 1 was installed with Ardupilot 3.9.0 beta 6 installed. equipped with air speed sensor and external compas, two GPS units as well.
After takeoff in FWBA A , gained some attitude ,made couple cycles around myself and switched RTL to test before proceeding to make 20-30 km FPV flight . Plane sudenly made a sharp dive to the ground. I was able to switch to FWBA A for recovery and gave full up elevator but plane just dove down even stiper. during dive wings stayed level in plane flu in strait line.
It was not first flight with that set up. Had some strange behavior but was able to bring it back for safe landing.
Please help me identify issue , weather it is software related or setup/user error!!!
Autoanalisize tool gave me below results:
Log File C:\Users\votolok\AppData\Local\Temp\tmp90CF.tmp.log
Size (kb) 39172.88671875
No of lines 481225
Duration 0:07:43
Vehicletype ArduPlane
Firmware Version V3.9.0-beta6
Firmware Hash ce7b5345
Hardware Type
Free Mem 0
Skipped Lines 0
Test: Autotune = NA -
Test: Brownout = GOOD -
Test: Compass = FAIL - FAIL: Large compass offset params (X:-490.29, Y:-238.51, Z:-46.48)
FAIL: Large compass offset in MAG data (X:-490.00, Y:-238.00, Z:-46.00)
Large change in mag_field (119.45%)

Test: Dupe Log Data = GOOD -
Test: Empty = GOOD -
Test: Event/Failsafe = GOOD -
Test: GPS = GOOD -
Test: IMU Mismatch = FAIL - Check vibration or accelerometer calibration. (Mismatch: 1.95, WARN: 0.75, FAIL: 1.50)
Test: Motor Balance = NA -
Test: NaNs = FAIL - Found NaN in POS.RelOriginAlt

Test: OpticalFlow = FAIL - FAIL: no optical flow data

Test: Parameters = GOOD -
Test: PM = NA -
Test: Pitch/Roll = NA -
Test: Thrust = NA -
Test: VCC = UNKNOWN - No CURR log data

Link to bin file https://yadi.sk/d/-g_5lj19E9Ondw

If you have an airspeed sensor that is reading way low, it will cause RTL to dive in an attempt to gain airspeed. I experienced this last week for the first time, and was shocked by it. I flipped from RTL to FBWA in time, and eventually with help from @tridge realized the source of the problem. Replacing broken pitot tube solved this for me. Check airspeed vs GPS speed in your log.

Kelly

I used same air speed sensor an several setups w/o any issues. Air speed values look reasonable and correcpond to plane behavior observed. One thing I noticed that att. and AHRS pitch value do not agree with each other! Thanks for a tip anyway!