Pixhawk on Octo: Des Roll vs Roll way off, Help with logs

I’ve been testing my droidworx x8 machine and have logged about 10 hours flight with no problems so far.

About the aircraft:

Skyjib HL X8
ESC - Herkules III
Mot - T-Motor 3515-15
Props - 16 x 5.4’'
PIXHAWK - mounted on 3D omnimac anti-vib mount
Jeti Rx/Tx
Battery: Zippy 8000 35C

The Incident:

I took the aircraft up to fly with no changes to previous days good flight (BIN #38), midway through the flight (attitude mode) the aircraft twitched a little with no stick input, about 3 times within a short period. I flicked to manual mode to bring it back to base and land, and whilst enroute the aircraft started to pitch uncontrollably and throttle was responsive only every so often. I was very lucky and eventually managed to land without roll-over.

Attached are the logs from SD card (.BIN #39+#38), I didn’t run the GCS on the flight unfortunately so I don’t have the tlog.
It’s the first time I’ve looked at the logs closely, and what I can observe is listed below: (BIN #39)

Main loss of control occurs in last minute of flight, and you can see the change in flight control mode from attitude to manual.

Vcc to pixhawk is good. Stayed within 5->5.3 volts for the flight

Flight log doesn’t cut out until batteries are disconnected

IMU Acc Z axis has large spikes, however I believe these are due to multiple impacts and ‘bounces’ off the ground when I was trying to safely land.

Battery Voltage - Gets a bit erratic at the moment of incident, I think due to throttle pulses while trying to control the aircraft.

Des Pitch vs Actual Pitch: The controller looks like it’s sending stabilise signals (shown by the straight line) however the Actual Pitch varies quite a lot.

That’s most of my observations, I have eyeballed all connections, they look okay but I haven’t taken it apart yet to have a close look, which is my next step. My interpretation from the logs alone is mechanical error, but I also am concerned it might be sloppy batteries aren’t provided enough juice or are old. According to the wiki it could also be ESC sync problems. The herkules III is a very good ESC I would be surprised to nail it with that, but definitely needs to be considered…

If anyone can please help me pinpoint my problem that would be a great help.

Log #38 is a sample of a good flight
Log #39 is when I struck trouble.

Cheers,
Cam

Motor 6 was having intermittent problems.