Well... I spoke too soon.
Our drivers only reported 1 or 2 connection drops while practicing outside of the competition venue (at school, no other robots)
There were a few more drops during practice night at the hub, but not too many.
During the competition... our connection gremlins returned
I would not be exaggerating too much if I said the robot lost connection once per round. (easily more than 50%)
At times it lost connection while sitting in the starting box waiting for the round to start.
Other times, it was while simply driving forward (2 large motors on ports 5 & 6).
We had a driver stall a winch motor while also driving - the nylon cord eventually broke - did not lose connection (probably the largest power demand)
There was no consistency to when it failed, so I don't think it was an overheat/overload issue.
The cortex is different than the one we experienced issues with last year.
We switched out VEXnet keys.
The cortex is near 2 servos, but about 8 inches from the closest motor.
The cortex has power pole connectors wired directly to the cortex (no adapter).
Batteries were never used for more than 3 rounds before recharging, and drops occurred on fresh batteries.
tried to cause a disconnect by jolting the cortex/keys - could not
I was not aware of other teams having issues, but it would have been easy to miss. (The hub did ask for hot-spots to be turned off at one point, so...)
I really think it has to be an interference problem. I will give the team a brand new cortex for regionals (just in case) and have them test an extension USB, but I don't think we will be able to replicate the game day WiFi conditions very easily... we will have to try... maybe at a basketball game with robots from the engineering class running.
on the bright side... Our veteran drivers were used to the issue and handled it very well... and the taught the new drivers not to panic. They also designed the gripping manipulators to "lock" into place to resist damage - they never dropped an item when connection was lost.