X12 Electronics Stack

aka v1 stack of Triton.

	Architecture Overview
Vehicle	X12, ROV Triton
Leads	Jacob Bobson, Cynthia Zatlokowicz
Heads	Ryan Pearson (Power)
Predecessors	X11, ROV Remora Stack
Successors	X13, ROV Triton Stack v2
Success?	Yes, but with flaws

Boards:

2 x Bricks Boards
Power Distribution
Power Conversion
Backplane
3 x QuadESCs
Pi Shield
Solenoid Board
Camera Boards

Summary

The X12 electronics stack was well developed. The schematics matched closely with the X11 stack, but the arrangement of boards was altered to better fit in the power box. The power box was chosen to fix many of the flaws with the X11 power tube, which it did. The box was machined from a single block of aluminum, which minimized the potential failure points to the lid seal and the binder panel jack seals. It is also an excellent heat sink for the quarter power bricks, which got hotter than desired when heatsunk to only the end of the X11 power tube lid.

With minimal changes to boards, relatively few issues were encountered. The backplane required a second and third revision to get right, and the Quad ESC, conversion, and solenoids boards were re-ordered for minor improvements. Limitations in removing and replacing the conversion board, backplane, and Quad ESC boards motivated many of the subsequent changes in the X13 electronics stack.

The discontinuation of the Afro ESCs and the difficulty in programming the existing ones also motivated the search for new ESCs in future stacks.

In its final weeks of use, Conner discovered the source of "brown outs" that rebooted the system when thrusters started too aggressively. The on/off circuit, which is designed to turn the bricks on simultaneously at ~45 volts was turning them off simultaneously when thrusters started and the input dipped from 48V to 45V. No error codes were present in the bricks since they were simply told to turn off and on again.

Scott also discovered the other source of brown outs. When going from a low speed to no speed, the Afro ESCs would short their positive and negative inputs together, leading to a massive current spike. Scott fixed the issue by modifying the Afro ESC firmware in assembly to skip the subroutine that caused the brown out.