Narrative:

A few unrelated things that might have actually been related:1. During cruise flight we had a dual FMC reset which included audio and EICAS indications that the autopilot and autothrottles had switched off. However; in spite of the indications; the autopilot and autothrottles remained connected. The FMC's reset normally. 2. Aside from the FMC's resetting; on two other seemingly random occasions we received audio and EICAS indications that the autopilot and autothrottles had switched off; however; in each case; the autopilot and autothrottles didn't actually disconnect. 3. We were cleared to descend from cruise altitude via the neell 6. The PF selected 2;000' in the altitude window accordingly. We were assigned to descend at or above 310 knots. We selected 310 knots on VNAV page 3 (the descent page) and the automation appropriately maintained the airplane at 310 knots. At this point; LNAV and VNAV were both engaged and things were normal; including appropriately complied with altitude constraints. At approximately 12;000' ATC instructed us to maintain 290 knots and so we adjusted the VNAV page 3 accordingly. Continuing the descent; as we approached 10;000'; I queried the controller if he needed us to continue maintaining 290 knots with the idea of us slowing below 250 knots at 10;000' in mind. He said that his previously assigned 290 knots was no longer necessary. As such; we attempted to begin our slow down. The PF opened up the speed window and dialed the speed back to 250 knots; however; the airplane only slowed to 280 knots; showed no signs of any further slow down attempt; and proceeded to go through 10;000'. The PF switched the vertical automation from VNAV to V/south; dialed the vertical speed to 500/minute; and dialed the speed window further back to 220 knots in order to allow the airplane to slow down. However; instead of slowing down; the autothrottles then powered up in order to continue maintaining 280 knots. Thinking that perhaps; by some very strange occurrence; that the automation was still somehow using the speed in the VNAV page 3; in spite of the speed window being open and with us being in the V/south mode; I accessed this page and noted that the previously manually set 310 and then 290 knots had automatically switched to 245 knots; debunking this theory. The PF grabbed the thrust levers; attempted to keep them at idle; and further dialed the vertical speed to 100'/minute in order to expedite the slow down. With the PF manually maintaining the thrust levers at idle in spite of the automation trying to throttle them back up; the airplane automation then used 'pitch on speed' raw autopilot mode in order to maintain 280 knots; disregarding the 100'/minute set in the V/south window and the 250 knots set in the speed window. Seemingly unrelated to everything going on; it was at about 5;000' when we momentarily received a 'stab trim 2' EICAS message; however; this went away on its own. Seeing that V/south wasn't getting the airplane slowed down; the PF then switched over to flch and checked the thrust levers at idle. The autothrottle 'hold' mode displayed appropriately and the thrust levers stayed at idle; however; the airplane continued utilizing 'pitch on speed' in order to maintain 280 knots in spite of 220 knots still in the speed window. The PF then pulled the speed brake to attempt to slow down. But the automation overcame this by increasing the nose over angle; still utilizing 'pitch on speed' in order to maintain 280 knots in spite of the speed window being open and the first officer continuing to dial the speed back through 220 knots. At that point; in flch; the thrust levers at idle and authrottles in 'hold'; the speed brake extended; and the airplane pitching to maintain 280 knots; the aircraft had developed an abnormally high descent rate. Prior to reaching waypoint elias; still in flch; the automation proceeded to bust our aircraft through the 2;000' we had previously set in the altitude window; corresponding to the[/2000A] set on the legs page on the right side of waypoint elias; as if our preselected altitude weren't even there. The PF then turned off all automation; pulled back on the yoke to arrest the descent; and got the speed under control. The airplane descended to approximately 1;700' MSL during this process. ATC queried us to confirm our altitude and I informed them that we were having difficulties with our aircraft. The PF quickly got the airplane back up to 2;000' and we all figured that it would be best to continue visually to the airport. The airplane automation would have driven us straight into gompertz channel without hesitation had we not intervened in spite of the fact that we had everything properly set. In hindsight; we could have turned off the automation sooner; however; the autopilot sustaining 'pitch on speed' raw autopilot mode with the engines at idle coupled with the speed brake fully extended meant that the descent from 10;000' to 2;000' was very quick due to the excessive descent speed that had developed. The stab 2 trim EICAS message also distracted us a bit at an inopportune time. I really can't rule it down to any one entity with the automation that would have been the cause of all of our problems.

Google
 

Original NASA ASRS Text

Title: 747-400 flight crew reported a malfunctioning autopilot they later realized had been written up; in aircraft maintenance logbook; several times prior to their flight.

Narrative: A few unrelated things that might have actually been related:1. During cruise flight we had a dual FMC reset which included audio and EICAS indications that the autopilot and autothrottles had switched off. However; in spite of the indications; the autopilot and autothrottles remained connected. The FMC's reset normally. 2. Aside from the FMC's resetting; on two other seemingly random occasions we received audio and EICAS indications that the autopilot and autothrottles had switched off; however; in each case; the autopilot and autothrottles didn't actually disconnect. 3. We were cleared to descend from cruise altitude via the NEELL 6. The PF selected 2;000' in the altitude window accordingly. We were assigned to descend at or above 310 knots. We selected 310 knots on VNAV page 3 (the descent page) and the automation appropriately maintained the airplane at 310 knots. At this point; LNAV and VNAV were both engaged and things were normal; including appropriately complied with altitude constraints. At approximately 12;000' ATC instructed us to maintain 290 knots and so we adjusted the VNAV page 3 accordingly. Continuing the descent; as we approached 10;000'; I queried the controller if he needed us to continue maintaining 290 knots with the idea of us slowing below 250 knots at 10;000' in mind. He said that his previously assigned 290 knots was no longer necessary. As such; we attempted to begin our slow down. The PF opened up the speed window and dialed the speed back to 250 knots; however; the airplane only slowed to 280 knots; showed no signs of any further slow down attempt; and proceeded to go through 10;000'. The PF switched the vertical automation from VNAV to V/S; dialed the vertical speed to 500/minute; and dialed the speed window further back to 220 knots in order to allow the airplane to slow down. However; instead of slowing down; the autothrottles then powered up in order to continue maintaining 280 knots. Thinking that perhaps; by some very strange occurrence; that the automation was still somehow using the speed in the VNAV page 3; in spite of the speed window being open and with us being in the V/S mode; I accessed this page and noted that the previously manually set 310 and then 290 knots had automatically switched to 245 knots; debunking this theory. The PF grabbed the thrust levers; attempted to keep them at idle; and further dialed the vertical speed to 100'/minute in order to expedite the slow down. With the PF manually maintaining the thrust levers at idle in spite of the automation trying to throttle them back up; the airplane automation then used 'pitch on speed' raw autopilot mode in order to maintain 280 knots; disregarding the 100'/minute set in the V/S window and the 250 knots set in the speed window. Seemingly unrelated to everything going on; it was at about 5;000' when we momentarily received a 'Stab Trim 2' EICAS message; however; this went away on its own. Seeing that V/S wasn't getting the airplane slowed down; the PF then switched over to FLCH and checked the thrust levers at idle. The Autothrottle 'HOLD' mode displayed appropriately and the thrust levers stayed at idle; however; the airplane continued utilizing 'pitch on speed' in order to maintain 280 knots in spite of 220 knots still in the speed window. The PF then pulled the speed brake to attempt to slow down. But the automation overcame this by increasing the nose over angle; still utilizing 'pitch on speed' in order to maintain 280 knots in spite of the speed window being open and the FO continuing to dial the speed back through 220 knots. At that point; in FLCH; the thrust levers at idle and authrottles in 'HOLD'; the speed brake extended; and the airplane pitching to maintain 280 knots; the aircraft had developed an abnormally high descent rate. Prior to reaching waypoint ELIAS; still in FLCH; the automation proceeded to bust our aircraft through the 2;000' we had previously set in the altitude window; corresponding to the[/2000A] set on the legs page on the right side of waypoint ELIAS; as if our preselected altitude weren't even there. The PF then turned off all automation; pulled back on the yoke to arrest the descent; and got the speed under control. The airplane descended to approximately 1;700' MSL during this process. ATC queried us to confirm our altitude and I informed them that we were having difficulties with our aircraft. The PF quickly got the airplane back up to 2;000' and we all figured that it would be best to continue visually to the airport. The airplane automation would have driven us straight into Gompertz Channel without hesitation had we not intervened in spite of the fact that we had everything properly set. In hindsight; we could have turned off the automation sooner; however; the autopilot sustaining 'pitch on speed' raw autopilot mode with the engines at idle coupled with the speed brake fully extended meant that the descent from 10;000' to 2;000' was very quick due to the excessive descent speed that had developed. The Stab 2 Trim EICAS message also distracted us a bit at an inopportune time. I really can't rule it down to any one entity with the automation that would have been the cause of all of our problems.

Data retrieved from NASA's ASRS site and automatically converted to unabbreviated mixed upper/lowercase text. This report is for informational purposes with no guarantee of accuracy. See NASA's ASRS site for official report.