RCA on cause of lost propulsion early in the race - bjornburton/champbot GitHub Wiki

Lake-side I identified the defective assembly by cross-wiring the larboard PWM into channel A of the H-bridge. The defective or damaged, component seemed to be the H-Bridge's channel A. I had no spare on hand. Its failure was confirmed later.

This is a 20 Amp H-Bridge assembly that I felt was sufficient. This was determined through testing, both in water and stalled, at full voltage—or so I thought.

The MOSFET drivers that were installed are listed here.

Designator	Part Number	Pb free	Date Code	Lot Number	Mold Cavity
M1	LR2905	P	215G	L6Z5	G
M2	IRLR2905		606P	K7L7	M (FAILED)
M3	LR2905	P	215G	L6Z5	T
M4	IRLR2905		407A	AF5C	M
M5	LR2905	P	215G	L6Z5	U
M6	IRLR2905		650P	26MH	M
M7	LR2905	P	215G	L6Z5	L
M8	FR2905		Z027N	Gl25	M

This is an odd mix of 5 lot-number/part-number pairs! Expectation is one part-number and no more than two lot-numbers.

This may mean counterfeits, seconds or pulls were used.

During the test I had a maximum of 14.7 Amps, but I didn't monitor the battery voltage. When I tested it, the battery was certainly in a less that fully charged state.

Prior to the race the battery was charged using the procedure recommended by the battery manufacturer (http://www.power-sonic.com/images/powersonic/technical/1277751263_20100627-TechManual-Lo.pdf). The battery was fully charged for perhaps for the first time.

During this RCA I retested with the fully-charged battery the current is about 37 amps!

In hind sight, if I had carried the spare H-Bridge I could have changed it but I would not have caught the fact that R3, the "SET" shunt, was installed. With this part installed on the bridge, the Trinket's processor gets 12 Volts and would be destroyed. The Trinket is a quick-change part, and I did have a spare, but I'm not sure I would have discovered and repaired it in time.

Root Conflict was due to a flawed test & validation, combined with doing something new: fully charging the battery.