Narrative:

It's not the autopilot. The first officer and I were discussing the 'capture problem' that exists when the FGC tries to lock on early to the pdx ILS runway 28L. After I described the symptoms; he said that he had seen it happen once; didn't understand what had gone wrong; and couldn't get the captain he was with to believe that it was the avionics; and not him; that had caused the airplane to turn away from the airport when it should be turning inbound. I explained further that the manufacturer thinks that one possible cause might be the autopilot. So he says; 'mind if I try it raw data; no autopilot?' I said sure; so we did. The conditions were: a clear day; 20 minutes before sundown. ATC gave us 'direct to hanah; cross at or above 2500 ft; cleared for the visual to runway 28L.' this gave us an intercept heading of 350 degrees. The autopilot is off; the command bars are removed. He's got raw data and he's hand flying. We have four miles to go to get to the extended centerline. He says; 'push navigation' what should happen in the FMA is to see heading in green (active) and localizer in white (armed). Instead; we instantly got localizer in green. That's bad; but it got worse. When the command bars came up; they commanded a five-degree bank to the left for five seconds; then gradually worked toward a twenty-degree bank to the right -- a turn in the wrong direction. During all of this; the localizer needle never came off the peg; since we still had not reached the centerline. Summing up: an early capture; a turn in the wrong direction; while hand flying. I don't know what's causing this; but it's not the autopilot.

Related reports:

Google
 

Original NASA ASRS Text

Title: A CRJ-700's FGC attempted to capture the PDX 28L ILS early. The flight director commanded a turn in the wrong direction to track the localizer. The crew was hand flying the approach having seen this event occur previously.

Narrative: It's not the autopilot. The First Officer and I were discussing the 'capture problem' that exists when the FGC tries to lock on early to the PDX ILS Runway 28L. After I described the symptoms; he said that he had seen it happen once; didn't understand what had gone wrong; and couldn't get the Captain he was with to believe that it was the avionics; and not him; that had caused the airplane to turn away from the airport when it should be turning inbound. I explained further that the manufacturer thinks that one possible cause might be the autopilot. So he says; 'Mind if I try it raw data; no autopilot?' I said sure; so we did. The conditions were: a clear day; 20 minutes before sundown. ATC gave us 'direct to HANAH; cross at or above 2500 FT; cleared for the visual to Runway 28L.' This gave us an intercept heading of 350 degrees. The autopilot is off; the command bars are removed. He's got raw data and he's hand flying. We have four miles to go to get to the extended centerline. He says; 'push NAV' What should happen in the FMA is to see heading in green (active) and LOC in white (armed). Instead; we instantly got LOC in green. That's bad; but it got worse. When the command bars came up; they commanded a five-degree bank to the left for five seconds; then gradually worked toward a twenty-degree bank to the right -- a turn in the wrong direction. During all of this; the localizer needle never came off the peg; since we still had not reached the centerline. Summing up: an early capture; a turn in the wrong direction; while hand flying. I don't know what's causing this; but it's not the autopilot.

Data retrieved from NASA's ASRS site as of April 2012 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.