IFR TRAINING FLT AUTOPLT GIVES HARD OVER SIGNAL RESULTING IN UPSET.

Narrative

WE WERE NBOUND ON 343 DEG R OFF ARDMORE AT 4000'. TURNED ON AUTOPLT FOR ABOUT 3 MINS; THEN ALL OF SUDDEN WERE IN A 90 DEG BANK GOING DOWN AT 3000 FPM. TURNED AUTOPLT OFF; PULLED PWR; PULLED FUSE FOR AUTOPLT; TURNED ON CARB HEAT; LOST 80% ENG PWR; TOOK BANK OUT SLOWLY; BROKE OUT ABOUT 1500' MSL LEVEL AIRPLANE PITCH UP TO LEVEL FLT ABOUT 300' ABOVE GND; GOT BEARING; AND TURNED LEFT TO 350 DEGS. 3 MI TO PAULS VALLEY ARPT. FLEW PLANE WITH 1200 RPM TO ARPT AND LANDED. I THINK AUTOPLT ON SMA SHOULD BE CHKED AND TESTED MORE OFTEN. CALLBACK CONVERSATION WITH RPTR REVEALED THE FOLLOWING: DID NOT HAVE PITOT HEAT ON AT TIME OF INCIDENT. GAVE THE ACFT TYPE AND AUTOPLT MODEL. DID NOT FEEL HE HAD ANY PROB WITH THE STATIC SYS. ADMITTED HOWEVER WHEN IN THE DSCNT HE PUT THE CARB HEAT ON AND KNEW THAT WAS PROBABLY THE TROUBLE WITH THE ENG. AIR WAS EXTREMELY SMOOTH; BUT HE WAS FLYING IN CLOUDS FOR TRNING FOR THE STUDENT WORKING ON AN INSTRUMENT RATING. WHEN HE TOOK THE ACFT THE STUDENT CALLED THE CENTER AND ADVISED THEY WERE IN 90 DEG BANK DSNDING AT 3000 FPM. DID NOT DECLARE EMER OR SET TRANSPONDER TO 7700. APPARENTLY NO LOSS OF SEP. MAINT CHKED THE ACFT OVER AND FOUND NO IRREGULARITIES AND FLEW THE ACFT TO HOME BASE USING THE AUTOPLT AND EVERYTHING WORKED PERFECTLY.