### FILE="Main.annotation"

## Copyright: Public domain.

## Filename: POWERED_FLIGHT_SUBROUTINES.agc

## Purpose: A section of Manche72 revision 3.

## It is part of the reconstructed source code for the final, flown

## release of the software for the Command Module's (CM) Apollo

## Guidance Computer (AGC) for Apollo 13. No original listings

## of this program are available; instead, this file was recreated

## from a reconstructed copy of Comanche 072. It has been adapted

## such that the resulting bugger words exactly match those

## specified for Manche72 revision 3 in NASA drawing 2021153G,

## which gives relatively high confidence that the reconstruction

## is correct.

## Assembler: yaYUL

## Contact: Ron Burkey <info@sandroid.org>.

## Website: www.ibiblio.org/apollo/index.html

## Mod history: 2024-05-19 MAS Created from Comanche 072.

BANK 14 # SAME FBANK AS THE FINDCDUD SUB-PROGRAM

SETLOC POWFLITE

BANK

EBANK= DEXDEX

COUNT* $$/POWFL

# CDUTRIG, CDUTRIG1, CDUTRIG2, AND CD*TR*GS ALL COMPUTE THE SINES AND

# COSINES OF THREE 2'S COMPLEMENT ANGLES AND PLACE THE RESULT, DOUBLE

# PRECISION, IN THE SAME ORDER AS THE INPUTS, AT SINCDU AND COSCDU. AN

# ADDITIONAL OUTPUT IS THE 1'S COMPLEMENT ANGLES AT CDUSPOT. THESE

# ROUTINES GO OUT OF THEIR WAY TO LEAVE THE MPAC AREA AS THEY FIND IT,

# EXCEPT FOR THE GENERALLY UNIMPORTANT MPAC +2. THEY DIFFER ONLY IN

# WHERE THEY GET THE ANGLES, AND IN METHOD OF CALLING.

#

# CDUTRIG (AND CDUTRIG1, WHICH CAN BE CALLED IN BASIC) COMPUTE THE

# SINES AND COSINES FROM THE CURRENT CONTENTS OF THE CDU REGISTERS.

# THE CONTENTS OF CDUTEMP, ETC., ARE NOT TOUCHED SO THAT THEY MAY

# CONTINUE TO FORM A CONSISTENT SET WITH THE LATEST PIPA READINGS.

#

# CDUTRIG1 IS LIKE CDUTRIG EXCEPT THAT IT CAN BE CALLED IN BASIC.

#

# CD*TR*GS FINDS CDU VALUES IN CDUSPOT RATHER THAN IN CDUTEMP. THIS

# ALLOWS USERS TO MAKE TRANSFORMATIONS USING ARBITRARY ANGLES, OR REAL

# ANGLES IN AN ORDER OTHER THAN X Y Z. A CALL TO THIS ROUTINE IS

# NECESSARY IN PREPARATION FOR A CALL TO AX*SR*T IN EITHER OF ITS TWO

# MODES (SMNB OR NBSM). SINCE AX*SR*T EXPECTS TO FIND THE SINES AND

# COSINES IN THE ORDER Y Z X THE ANGLES MUST HAVE BEEN PLACED IN CDUSPOT

# IN THIS ORDER. CD*TR*GS NEED NOT BE REPEATED WHEN AX*SR*T IS CALLED

# MORE THAN ONCE, PROVIDED THE ANGLES HAVE NOT CHANGED. NOTE THAT SINCE

# IT CLOBBERS BUF2 (IN THE SINE AND COSINE ROUTINES) CD*TR*GS CANNOT BE

# CALLED USING BANKCALL. SORRY.

#

# CD*TR*G IS LIKE CD*TR*GS EXCEPT THAT IT CAN BE CALLED IN

# INTERPRETIVE.

CDUTRIG EXIT

TC CDUTRIGS

TC INTPRET

RVQ

CD*TR*G EXIT

TC CD*TR*GS

TC INTPRET

RVQ

CDUTRIGS CA CDUX

TS CDUSPOT +4

CA CDUY

TS CDUSPOT

CA CDUZ

TS CDUSPOT +2

CD*TR*GS EXTEND

QXCH TEM2

CAF FOUR

TR*GL**P MASK SIX # MAKE IT EVEN AND SMALLER

TS TEM3

INDEX TEM3

CA CDUSPOT

DXCH MPAC # STORING 2'S COMP ANGLE, LOADING MPAC

DXCH VBUF +4 # STORING MPAC FOR LATER RESTORATION

TC USPRCADR

CADR CDULOGIC

EXTEND

DCA MPAC

INDEX TEM3

DXCH CDUSPOT # STORING 1'S COMPLEMENT ANGLE

TC USPRCADR

CADR COSINE

DXCH MPAC

INDEX TEM3

DXCH COSCDU # STORING COSINE

EXTEND

INDEX TEM3

DCA CDUSPOT # LOADING 1'S COMPLEMENT ANGLE

TC USPRCADR

CADR SINE +1 # SINE +1 EXPECTS ARGUMENT IN A AND L

DXCH VBUF +4 # BRINGING UP PRIOR MPAC TO BE RESTORED

DXCH MPAC

INDEX TEM3

DXCH SINCDU

CCS TEM3

TCF TR*GL**P

TC TEM2

# ****************************************************************************************************************

# QUICTRIG, INTENDED FOR GUIDANCE CYCLE USE WHERE TIME IS CRITICAL, IS A MUCH FASTER VERSION OF CD*TR*GS.

# QUICTRIG COMPUTES AND STORES THE SINES AND COSINES OF THE 2'S COMPLEMENT ANGLES AT CDUSPOT, CDUSPOT +2,

# AND CDUSPOT +4. UNLIKE CD*TR*GS, QUICTRIG DOES NOT LEAVE THE 1'S COMPLEMENT VERSIONS OF THE ANGLES IN

# CDUSPOT. QUICTRIG'S EXECUTION TIME IS 4.1 MS; THIS IS 10 TIMES AS FAST AS CD*TR*GS. QUICTRIG MAY BE

# CALLED FROM INTERPRETIVE AS AN RTB OP-CODE, OR FROM BASIC VIA BANKCALL OR IBNKCALL.

QUICTRIG INHINT # INHINT SINCE DAP USES THE SAME TEMPS

EXTEND

QXCH ITEMP1

CAF FOUR

+4 MASK SIX

TS ITEMP2

INDEX ITEMP2

CA CDUSPOT

TC SPSIN

EXTEND

MP BIT14 # SCALE DOWN TO MATCH INTERPRETER OUTPUTS

INDEX ITEMP2

DXCH SINCDU

INDEX ITEMP2

CA CDUSPOT

TC SPCOS

EXTEND

MP BIT14

INDEX ITEMP2

DXCH COSCDU

CCS ITEMP2

TCF QUICTRIG +4

CA ITEMP1

RELINT

TC A

#****************************************************************************************************************

# THESE INTERFACE ROUTINES MAKE IT POSSIBLE TO CALL AX*SR*T, ETC., IN

# INTERPRETIVE. LATER, WHERE POSSIBLE, THEY WILL BE ELIMINATED.

#

# NBSM WILL BE THE FIRST TO GO. IT SHOULD NOT BE USED.

NBSM STQ

X2

LXC,1 VLOAD*

S1 # BASE ADDRESS OF THE CDU ANGLES IS IN S1

0,1

STOVL CDUSPOT

32D # VECTOR TO BE TRANSFORMED IS IN 32D

CALL

TRG*NBSM

STCALL 32D # SINCE THERE'S NO STGOTO

X2

# THESE INTERFACE ROUTINES ARE PERMANENT. ALL RESTORE USER'S EBANK

# SETTING. ALL ARE STRICT INTERPRETIVE SUBROUTINES, CALLED USING "CALL",

# RETURNING VIA QPRET. ALL EXPECT AND RETURN THE VECTOR TO BE TRANSFOR-

# MED INTERPRETER-STYLE IN MPAC; COMPONENTS AT MPAC, MPAC +3, AND MPAC +5.

#

# TRG*SMNB AND TRG*NBSM BOTH EXPECT TO SEE THE 2'S COMPLEMENT ANGLES

# AT CDUSPOT (ORDER Y Z X, AT CDUSPOT, CDUSPOT +2, AND CDUSPOT +4; ODD

# LOCATIONS NEED NOT BE ZEROED). TRG*NBSM DOES THE NB TO SM TRANSFOR-

# MATION; TRG*SMNB, VICE VERSA.

#

# CDU*NBSM DOES ITS TRANSFORMATION USING THE PRESENT CONTENTS OF

# THE CDU COUNTERS. OTHERWISE IT IS LIKE TRG*NBSM.

#

# CDU*SMNB IS THE COMPLEMENT OF CDU*NBSM.

CDU*SMNB EXIT

TC CDUTRIGS

TCF C*MM*N1

TRG*SMNB EXIT

TC CD*TR*GS

C*MM*N1 TC MPACVBUF # AX*SR*T EXPECTS VECTOR IN VBUF

CS THREE # SIGNAL FOR SM TO NB TRANSFORMATION

C*MM*N2 TC AX*SR*T

TC INTPRET

VLOAD RVQ

VBUF

CDU*NBSM EXIT

TC CDUTRIGS

TCF C*MM*N3

TRG*NBSM EXIT

TC CD*TR*GS

C*MM*N3 TC MPACVBUF # FOR AX*SR*T

CA THREE # SIGNAL FOR NB TO SM TRANSFORMATION

TCF C*MM*N2

# *NBSM* AND *SMNB* EXPECT TO SEE THE SINES AND COSINES (AT SINCDU

# AND COSCDU) RATHER THAN THE ANGLES THEMSELVES. OTHERWISE THEY ARE

# LIKE TRG*NBSM AND TRG*SMNB.

#

# NOTE THAT JUST AS CD*TR*GS NEED BE CALLED ONLY ONCE FOR EACH SERIES

# OF TRANSFORMATIONS USING THE SAME ANGLES, SO TOO ONLY ONE OF TRG*NBSM

# AND TRG*SMNB NEED BE CALLED FOR EACH SERIES. FOR SUBSEQUENT TRANSFOR-

# MATIONS USE *NBSM* AND *SMNB*.

*SMNB* EXIT

TCF C*MM*N1

*NBSM* EXIT

TCF C*MM*N3

# AX*SR*T COMBINES THE OLD SMNB AND NBSM. FOR THE NB TO SM

# TRANSFORMATION, ENTER WITH +3 IN A. FOR SM TO NB, ENTER WITH -3.

# THE VECTOR TO BE TRANSFORMED ARRIVES, AND IS RETURNED, IN VBUF.

# AX*SR*T EXPECTS TO FIND THE SINES AND COSINES OF THE ANGLES OF ROTATION

# AT SINCDU AND COSCDU, IN THE ORDER Y Z X. A CALL TO CD*TR*GS, WITH

# THE 2'S COMPLEMENT ANGLES (ORDER Y Z X) AT CDUSPOT, WILL TAKE CARE OF

# THIS. HERE IS A SAMPLE CALLING SEQUENCE:-

# TC CDUTRIGS

# CS THREE ("CA THREE" FOR NBSM)

# TC AX*SR*T

# THE CALL TO CD*TR*GS NEED NOT BE REPEATED, WHEN AX*SR*T IS CALLED MORE

# THAN ONCE, UNLESS THE ANGLES HAVE CHANGED.

#

# AX*SR*T IS GUARANTEED SAFE ONLY FOR VECTORS OF MAGNITUDE LESS THAN

# UNITY. A LOOK AT THE CASE IN WHICH A VECTOR OF GREATER MAGNITUDE

# HAPPENS TO LIE ALONG AN AXIS OF THE SYSTEM TO WHICH IT IS TO BE TRANS-

# FORMED CONVINCES ONE THAT THIS IS A RESTRICTION WHICH MUST BE ACCEPTED.

AX*SR*T TS DEXDEX # WHERE IT BECOMES THE INDEX OF INDEXES

EXTEND

QXCH RTNSAVER

R*TL**P CCS DEXDEX # +3 --> 0 -3 --> 2

CS DEXDEX # THUS: +2 --> 1 -2 --> 1

AD THREE # +1 --> 2 -1 --> 0

EXTEND

INDEX A

DCA INDEXI

DXCH DEXI

CA ONE

TS BUF

EXTEND

INDEX DEX1

DCS VBUF

TCF LOOP1 # REALLY BE A SUBTRACT, AND VICE VERSA

LOOP2 DXCH BUF # LOADING VECTOR COMPONENT, STORING INDEX

LOOP1 DXCH MPAC

CA SINESLOC

AD DEX1

TS ADDRWD

TC DMPSUB # MULTIPLY BY SIN(CDUANGLE)

CCS DEXDEX

DXCH MPAC # NBSM CASE

TCF +3

EXTEND # SMNB CASE

DCS MPAC

DXCH TERM1TMP

CA SIX # SINCDU AND COSCDU (EACH 6 WORDS) MUST

ADS ADDRWD # BE CONSECUTIVE AND IN THAT ORDER

EXTEND

INDEX BUF

INDEX DEX1

DCA VBUF

DXCH MPAC

TC DMPSUB # MULTIPLY BY COS(CDUANGLE)

DXCH MPAC

DAS TERM1TMP

DXCH TERM1TMP

DDOUBL

INDEX BUF

INDEX DEX1

DXCH VBUF

DXCH BUF # LOADING INDEX, STORING VECTOR COMPONENT

CCS A # 'CAUSE THAT'S WHERE THE INDEX NOW IS

TCF LOOP2

EXTEND

DIM DEXDEX # DECREMENT MAGNITUDE PRESERVING SIGN

TSTPOINT CCS DEXDEX # ONLY THE BRANCHING FUNCTION IS USED

TCF R*TL**P

TC RTNSAVER

TCF R*TL**P

TC RTNSAVER

SINESLOC ADRES SINCDU # FOR USE IN SETTING ADDRWD

INDEXI DEC 4 # ********** DON'T **********

DEC 2 # ********** TOUCH **********

DEC 0 # ********** THESE **********

DEC 4 # ********** CONSTANTS **********

# ****************************************************************************************************************

# THIS SUBROUTINE COMPUTES INCREMENTAL CHANGES IN CDU(GIMBAL) ANGLES FROM INCREMENTAL CHANGES ABOUT SM AXES. IT

# REQUIRES SM INCREMENTS AS A DP VECTOR SCALED AT ONE REVOLUTION(DTHETASM,+2,+4). SIN,COS(CDUY,Z,X) ARE IN

# SINCDU,+2,+4 AND COSCDU,+2,+4 RESPECTIVELY, SCALED TO ONE HALF. CDU INCREENTS ARE PLACED IN DCDU,+2,+4 SCALED TO

# ONE REVOLUTION.

#

# * COS(IGA)SEC(MGA) 0 -SIN(IGA)SEC(MGA) *

# * *

# * -COS(IGA)TAN(MGA) 1 SIN(IGA)TAN(MGA) *

# * *

# * SIN(IGA) 0 COS(IGA) *

BANK 14

SETLOC POWFLIT1

BANK

SMCDURES DLOAD DMP

DTHETASM

COSCDUY

PDDL DMP

DTHETASM +4

SINCDUY

BDSU

DDV

COSCDUZ

STORE DCDU

DMP SL1 # SCALE

SINCDUZ

BDSU

DTHETASM +2

STODL DCDU +2

DTHETASM

DMP PDDL

SINCDUY

DTHETASM +4

DMP DAD

COSCDUY

SL1

STORE DCDU +4

RVQ