Narrative:

I was climbing the medium large transport to a planned cruise altitude of FL220. An intermediate level off altitude of FL210 had been displayed in the dfgc altitude select for several mins since set by the first officer following the most recent altitude clearance. At FL200 I reduced the rate of climb to less than 1000 FPM in anticipation of the autoplt altitude capture at FL210. At about FL206 the ZOB controller requested a descent to FL200. By FL208 I had reversed our climb and we continued the descent and leveled off at FL200. At this time cle advised us that we had been cleared to FL200 and we responded that we understood the clearance to be to FL210. Later phone discussions with ZOB controllers indicate the audio tapes reveal clearance issued to FL200 and pilot readback of FL200. The only reasonable explanation to this incident deals with the lack of human engineering in the design of the installed flight director on our medium large transport's. This system is currently being modified due to innumerable incidents of this type, but only a small fraction of the fleet has been modified and we will be using the old procedure until the entire fleet modification is complete. In addition our recurrent training does not specifically deal with techs to minimize exposure to the limitations of the system. In regard to this incident the particular limitation of the system that created the altitude discrepancy is the altitude arming and setting switch. For each change in altitude clearance the altitude set knob is rotated while the pilots visually monitor an adjacent altitude readout window. There are very slight detents between each altitude setting that can just barely be felt as the set knob is rotated. For the system to work the altitude set knob must finally be pulled out to arm the alert and capture modes. Unfortunately if the set knob is between the mini-detents or if the slightest rotating motion of the hand moves it out of the detent, then the altitude selected will change when the spring-loaded set knob returns to its normal position. At that particular moment the pilot's eyes are focused on the flight mode annunciator to confirm the arming function indicated by an amber altitude. Now one must immediately rechk the altitude readout window to confirm the altitude selected. There is complicated 6-STEP technique with an exact sequence that can be used to overcome the system limitations that I believe most pilots attempt to follow. If nothing else was going on in the cockpit, if there were no other important #'south (frequencys, courses, speeds, weights, cabin temperatures, etc) then I believe this technique would be effective most of the time until the dfgc and FMA are modified. In this case the altitude changed to FL210 and other cockpit activities precluded our noting the discrepancy. In this incident involving a short leg at a low altitude some intervening events that warn of an impending problem did not occur. There was no change of controller sector that would have revealed the erroneous altitude selected. There was no advisory ot eh location or altitude of the conflicting traffic that was causing the intermediate level off. The fortunate part of this incident is that the ATC alert system functioned as designed and an alert controller was able to communicate a corrective action in a timely manner. This type of incident will continue to plague the ATC system until: 1) ineffective dfgc systems are modified at an expedited rate. 2) training for all parties include system limitations and offsetting techs. 3) cockpit duties are limited in relation to the reality of the 2-MAN environment. Supplemental information from acn 144483: there are 2 probable causes. First, we are very aware of a problem with the present flight director. All of our airline medium large transport flight directors are being modified so the altitude arming switch will not have to be pulled. It must now be pulled to arm the system. When the arming knob is released it can be inadvertently turned which will change the altitude being armed in the flight director. This is probably what happened. The second contributing factor is radio discipline. Controllers are starting to give too many changes and not giving adequate time to give a clear response. This is especially true when there are a lot of #'sinvolved; i.e., altitude change, heading change, airspeed and radio frequency change. The controllers that do it most are the sharpest ones that handle a lot of traffic at the same time as in chicago, atl, ny and dfw. They may save time for themselves but are asking for errors from others.

Related reports:

Google
 

Original NASA ASRS Text

Title: ALT DEVIATION. THE DFGC (DIGITAL FLT GUIDANCE COMPUTER) AS SET BY THE ALT ALERT KNOB IS REPORTED TO BE VERY DIFFICULT TO SET AND THEN STAY SET FOR PROPER ALT READOUT IN CONTROLLING CLIMB OR DESCENT.

Narrative: I WAS CLBING THE MLG TO A PLANNED CRUISE ALT OF FL220. AN INTERMEDIATE LEVEL OFF ALT OF FL210 HAD BEEN DISPLAYED IN THE DFGC ALT SELECT FOR SEVERAL MINS SINCE SET BY THE F/O FOLLOWING THE MOST RECENT ALT CLRNC. AT FL200 I REDUCED THE RATE OF CLB TO LESS THAN 1000 FPM IN ANTICIPATION OF THE AUTOPLT ALT CAPTURE AT FL210. AT ABOUT FL206 THE ZOB CTLR REQUESTED A DSNT TO FL200. BY FL208 I HAD REVERSED OUR CLB AND WE CONTINUED THE DSNT AND LEVELED OFF AT FL200. AT THIS TIME CLE ADVISED US THAT WE HAD BEEN CLRED TO FL200 AND WE RESPONDED THAT WE UNDERSTOOD THE CLRNC TO BE TO FL210. LATER PHONE DISCUSSIONS WITH ZOB CTLRS INDICATE THE AUDIO TAPES REVEAL CLRNC ISSUED TO FL200 AND PLT READBACK OF FL200. THE ONLY REASONABLE EXPLANATION TO THIS INCIDENT DEALS WITH THE LACK OF HUMAN ENGINEERING IN THE DESIGN OF THE INSTALLED FLT DIRECTOR ON OUR MLG'S. THIS SYS IS CURRENTLY BEING MODIFIED DUE TO INNUMERABLE INCIDENTS OF THIS TYPE, BUT ONLY A SMALL FRACTION OF THE FLEET HAS BEEN MODIFIED AND WE WILL BE USING THE OLD PROC UNTIL THE ENTIRE FLEET MODIFICATION IS COMPLETE. IN ADDITION OUR RECURRENT TRNING DOES NOT SPECIFICALLY DEAL WITH TECHS TO MINIMIZE EXPOSURE TO THE LIMITATIONS OF THE SYS. IN REGARD TO THIS INCIDENT THE PARTICULAR LIMITATION OF THE SYS THAT CREATED THE ALT DISCREPANCY IS THE ALT ARMING AND SETTING SWITCH. FOR EACH CHANGE IN ALT CLRNC THE ALT SET KNOB IS ROTATED WHILE THE PLTS VISUALLY MONITOR AN ADJACENT ALT READOUT WINDOW. THERE ARE VERY SLIGHT DETENTS BTWN EACH ALT SETTING THAT CAN JUST BARELY BE FELT AS THE SET KNOB IS ROTATED. FOR THE SYS TO WORK THE ALT SET KNOB MUST FINALLY BE PULLED OUT TO ARM THE ALERT AND CAPTURE MODES. UNFORTUNATELY IF THE SET KNOB IS BTWN THE MINI-DETENTS OR IF THE SLIGHTEST ROTATING MOTION OF THE HAND MOVES IT OUT OF THE DETENT, THEN THE ALT SELECTED WILL CHANGE WHEN THE SPRING-LOADED SET KNOB RETURNS TO ITS NORMAL POS. AT THAT PARTICULAR MOMENT THE PLT'S EYES ARE FOCUSED ON THE FLT MODE ANNUNCIATOR TO CONFIRM THE ARMING FUNCTION INDICATED BY AN AMBER ALT. NOW ONE MUST IMMEDIATELY RECHK THE ALT READOUT WINDOW TO CONFIRM THE ALT SELECTED. THERE IS COMPLICATED 6-STEP TECHNIQUE WITH AN EXACT SEQUENCE THAT CAN BE USED TO OVERCOME THE SYS LIMITATIONS THAT I BELIEVE MOST PLTS ATTEMPT TO FOLLOW. IF NOTHING ELSE WAS GOING ON IN THE COCKPIT, IF THERE WERE NO OTHER IMPORTANT #'S (FREQS, COURSES, SPDS, WTS, CABIN TEMPS, ETC) THEN I BELIEVE THIS TECHNIQUE WOULD BE EFFECTIVE MOST OF THE TIME UNTIL THE DFGC AND FMA ARE MODIFIED. IN THIS CASE THE ALT CHANGED TO FL210 AND OTHER COCKPIT ACTIVITIES PRECLUDED OUR NOTING THE DISCREPANCY. IN THIS INCIDENT INVOLVING A SHORT LEG AT A LOW ALT SOME INTERVENING EVENTS THAT WARN OF AN IMPENDING PROB DID NOT OCCUR. THERE WAS NO CHANGE OF CTLR SECTOR THAT WOULD HAVE REVEALED THE ERRONEOUS ALT SELECTED. THERE WAS NO ADVISORY OT EH LOCATION OR ALT OF THE CONFLICTING TFC THAT WAS CAUSING THE INTERMEDIATE LEVEL OFF. THE FORTUNATE PART OF THIS INCIDENT IS THAT THE ATC ALERT SYS FUNCTIONED AS DESIGNED AND AN ALERT CTLR WAS ABLE TO COMMUNICATE A CORRECTIVE ACTION IN A TIMELY MANNER. THIS TYPE OF INCIDENT WILL CONTINUE TO PLAGUE THE ATC SYS UNTIL: 1) INEFFECTIVE DFGC SYSTEMS ARE MODIFIED AT AN EXPEDITED RATE. 2) TRNING FOR ALL PARTIES INCLUDE SYS LIMITATIONS AND OFFSETTING TECHS. 3) COCKPIT DUTIES ARE LIMITED IN RELATION TO THE REALITY OF THE 2-MAN ENVIRONMENT. SUPPLEMENTAL INFO FROM ACN 144483: THERE ARE 2 PROBABLE CAUSES. FIRST, WE ARE VERY AWARE OF A PROB WITH THE PRESENT FLT DIRECTOR. ALL OF OUR AIRLINE MLG FLT DIRECTORS ARE BEING MODIFIED SO THE ALT ARMING SWITCH WILL NOT HAVE TO BE PULLED. IT MUST NOW BE PULLED TO ARM THE SYS. WHEN THE ARMING KNOB IS RELEASED IT CAN BE INADVERTENTLY TURNED WHICH WILL CHANGE THE ALT BEING ARMED IN THE FLT DIRECTOR. THIS IS PROBABLY WHAT HAPPENED. THE SECOND CONTRIBUTING FACTOR IS RADIO DISCIPLINE. CTLRS ARE STARTING TO GIVE TOO MANY CHANGES AND NOT GIVING ADEQUATE TIME TO GIVE A CLEAR RESPONSE. THIS IS ESPECIALLY TRUE WHEN THERE ARE A LOT OF #'SINVOLVED; I.E., ALT CHANGE, HDG CHANGE, AIRSPD AND RADIO FREQ CHANGE. THE CTLRS THAT DO IT MOST ARE THE SHARPEST ONES THAT HANDLE A LOT OF TFC AT THE SAME TIME AS IN CHICAGO, ATL, NY AND DFW. THEY MAY SAVE TIME FOR THEMSELVES BUT ARE ASKING FOR ERRORS FROM OTHERS.

Data retrieved from NASA's ASRS site as of July 2007 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.