Narrative:

We were being vectored by approach control for the visual approach. Our last assigned altitude was to descend and maintain 3;000 MSL (2;990 AGL in this area). I had the airport in sight and we were just in the process of reporting the airport and being cleared for the visual approach; when my copilot said; 'altitude.' the autopilot was engaged; but the altitude cap was not. We were at 2;700 feet MSL descending; we hit 2;550 feet MSL by the time I added power and pitch to arrest the descent and climb to 3;000 feet MSL. At that second we were; 'cleared for the visual approach- contact tower.' the autopilot on these 30 year old planes has a mind of its own. We operate [several] of the same model; and they all behave the same. They seem to lose the altitude cap function when they 'feel' like it. They do kick off the altitude cap when you move the pitch wheel to adjust your rate of climb/ descent within about 500 feet of level off. But; they also drop it occasionally at their own whim too with no inputs to the pitch wheel. Our normal operating procedure at the 1;000 feet to altitude tone/chime is; '1;000 feet to go (i.e. 4;000 for 3;000) altitude cap selected.' which I must have missed in our discussion of seeing the airport in VMC/visual conditions. Our descent rate was between around 500 FPM. Double/triple checking that altitude cap is on in the climb and descent; as well as; checking altitude cap is active at the 1;000 tone/chime without fail is key to solving this issue.

Related reports:

Google
 

Original NASA ASRS Text

Title: DA10 Captain reports an altitude deviation during vectors for a visual approach due to an autopilot capture anomaly that is well known.

Narrative: We were being vectored by Approach Control for the visual approach. Our last assigned altitude was to descend and maintain 3;000 MSL (2;990 AGL in this area). I had the airport in sight and we were just in the process of reporting the airport and being cleared for the visual approach; when my copilot said; 'Altitude.' The autopilot was engaged; but the ALT CAP was not. We were at 2;700 feet MSL descending; we hit 2;550 feet MSL by the time I added power and pitch to arrest the descent and climb to 3;000 feet MSL. At that second we were; 'Cleared for the visual approach- contact Tower.' The autopilot on these 30 year old planes has a mind of its own. We operate [several] of the same model; and they all behave the same. They seem to lose the ALT CAP function when they 'feel' like it. They DO kick off the ALT CAP when you move the PITCH wheel to adjust your rate of climb/ descent within about 500 feet of level off. But; they also drop it occasionally at their own whim too with NO inputs to the pitch wheel. Our normal operating procedure at the 1;000 feet to altitude tone/chime is; '1;000 feet to go (i.e. 4;000 for 3;000) ALT CAP selected.' Which I must have missed in our discussion of seeing the airport in VMC/visual conditions. Our descent rate was between around 500 FPM. Double/triple checking that ALT CAP is on in the climb and descent; as well as; checking ALT CAP is active at the 1;000 tone/chime without fail is key to solving this issue.

Data retrieved from NASA's ASRS site as of July 2013 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.