B737 HAD ERRATIC LOC SIGNAL AND MINOR MAP SHIFT ON APCH TO DTW.

Narrative

APCH: VECTORED FOR ILS RWY 21L AT DTW CAPT FLYING. AS FO; I TUNED IDENTED LOC FOR RWY 21L. NON EVENTFUL AS WE INTERCEPTED AND TRACKED INBOUND ON THE LOC AUTOPLT FLYING THE SHIP; CLRNC RECEIVED FOR ILS RWY 21L. ENGAGED APCH ON MCP (MODE CTL PANEL); GS INTERCEPT NORMAL. SHORTLY THEREAFTER LOC BEGAN SCALLOPING OR MOVING ERRATICALLY CONSEQUENTLY THE AUTOPLT TRACKED THESE OSCILLATIONS IN AN EFFORT TO ELIMINATE THE ACFT OSCILLATION TRACKING THE LOC. THE CAPT ATTEMPTED TO SELECT HEADING MODE ON MCP THE AUTOPLT WILL NOT DEFAULT TO HEADING ONCE GS AND LOC ARE CAPTURED. CAPT THEN DISENGAGED AUTOPLT MOMENTARILY; RE- ENGAGED AUTOPLT AND SELECTED LNAV TO COINCIDE WITH LNAV WAYPOINT DATA SELECTED ON FMCS. ACFT BEGAN AGGRESSIVE TURN TO R WHICH RESULTED IN A DEP FROM LOC. FO WAS IN RAW DATA AS WAS CAPT AND CAPT QUICKLY CORRECTED WITH HEADING COMMANDS AND RESELECTED APCH. ALMOST IMMEDIATELY THE ACFT WAS REESTABLISHED ON THE LOC AND GS. NORMAL LNDG FOLLOWED. MY ONLY THOUGHTS ARE THAT IT IS VERY COMMON FOR CREW MEMBERS TO ATTEMPT TO CHANGE MCP MODALITIES WHILE AUTOPLT IS ENGAGED IN APCH WITH LOC AND GS CAPTURED AND FIND NOTHING HAPPENING. IT WOULD BE A NICE DESIGN CONSIDERATION TO BE ABLE TO DESELECT ANY MCP FUNCTION REGARDLESS OF PHASE OF FLT AND RESELECT THE DESIRED MODALITY WITHOUT EITHER TURNING OFF FLT DIRECTORS THEN BACK ON; OR WHAT USUALLY HAPPENS; AUTOPLT OFF A MOMENT OF DISORIENTATION THEN REENGAGEMENT RECAPTURED OR BREAK OFF.