The US Air Force released its report on the November 16, 2010, crash of an F-22 fighter in Alaska
The onboard oxygen generating system (OBOGS) itself was not to blame, the report said. The OBOGS did stop working, due to an upstream fault in the air bleed system. (That's not to say that there have not been issues with OBOGS,
either, whether on the F-22 and other aircraft.)
The bulk of the blame in the report is laid on the deceased pilot, who the investigators believe failed to manage the fighter's flightpath while trying to activate the emergency oxygen system (EOS) - "channelized attention", as the report describes it.
The sequence of events in the report started as follows: The air bleed failure shut off the OBOGS - and, with oxygen mask on, the pilot started to suffocate. The first corrective action would be to activate the EOS, which is attached to the ejection seat. The EOS is also the post-ejection oxygen supply, and is activated by a lanyard on ejection. For emergency onboard use, the pilot activates the EOS by snapping a metal ring out of a holder on the seat and pulling on a cable.
The report then notes that control movements put the aircraft into a rapid descent. Recovery was attempted at 5,700 feet above ground but was too late to be successful.
According to the report it takes a 33-pound pull to get the EOS ring out of the clip and a 40-pound forward pull, while "minimizing inboard/outboard and upward motion" to start the oxygen flow.
EOS activation ring - Note close proximity of ring to cushion
Also, the pilot was wearing full Arctic survival gear including bulky gloves, which made it difficult to move in the cockpit without nudging the controls, and night vision goggles that limit head movement under the canopy - making it difficult if not impossible to see the EOS ring.
NVGs block view inside canopy
Personal view: this sounds more like the kind of activity involved in extracting an evil-minded spring clip from somewhere behind the engine block on a 1960s British sports car than operating life-essential emergency equipment. The report notes that the board members had no trouble operating the system in the simulator, even with a valve incorrectly installed. It does not specify whether they were suffocating at the time.
Enough of my highly unqualified opinion, though: Some interesting discussion, with various viewpoints,
over at PPrune.