*****    ***         *     *   ****   *****  *****   ***
         *       *   *        **    *  *    *    *    *      *   *
         *          *         * *   *  *    *    *    *      *
         ****      *    ****  *  *  *  *    *    *    ***     ***
             *    *           *   * *  *    *    *    *          *
             *   *            *    **  *    *    *    *      *   *
         ****    *****        *     *   ****     *    ******  ***

         Volume 2 Number 9        48/39             September 1977

                     Newsletter of the SR-52 Users Club
                                published at
                           9459 Taylorsville Road
                              Dayton, OH 45424
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

HIR Operations: A Major Discovery (58/59)
     Heinrich Schnepf (376), who publishes the German PPC newsletter
Display, discovered that p82 can be used to access the 8 pending
arithmetic registers.  TI acknowledges awareness of this capability,
but has no comment when asked why it was not announced.  A PC-100A
listing gives p82 the mnemonic: HIR, which I interpret to stand for
Hierarchy Internal Register.  I will henceforth use HIR mn to refer to
an operation, and HIR n to refer to HIR number n.  Heinrich found that
the sequence:  HIR mn does the following:  With n taking on the address
of one of the 8 HIRs (n=1,2,...8), m=0 produces STO, m=1: RCL, m=3:
SUM, m=4: Prd, m=5: INV SUM, and m=6,7,8 or 9: INV Prd.  Nested arith-
metic operations push operands first into HIR 1, then on into 2,3,...8;
print buffering assigns HIR 5 to Op 1, HIR 6 to Op 2, HIR 7 to Op 3,
and HIR 8 to Op 4; INV P/R uses the first 2 available HIRs, and HIRs
7 and 8; P/R uses the first available, and 7 and 8; D.MS and INV D.MS
use the first 2 available, and 8 ... which hold potentially useful
separations of some of the D.MS'd elements; sigma + and - use 7 and 8;
x (x-bar) uses the first available; Op 11 uses the first 2 available;
Ops 12 and 15 use the first 3 available; Op 13 uses the first 4 avail-
able; and Op 14 uses the first 3 available and 8.
     Unlike the analogous SR-52 Reg 60-69, the HIRs do not reformat
pushed operands, nor appear to attach operators to them.  Thus operands
used in nested arithmetic operations stay "clean" and may be retrieved
intact with the appropriate HIR recall (HIR 1n); stacked operands may
be changed (by HIR 0n) or modified (by HIR im, m=3,4,5,6) prior to
operator execution.  There is no (or doesn't seem to be a) fractured
digits potential.  Neither CLR nor CMs nor CP clears a HIR.  The only
way I've found that does (short of storing zeros or turning the machine
off) is by Op 00, which only clears HIRs 5 through 8, and even then
only if the 58/59 is plugged into the printer; without the printer,
Ops 0-5 execute as Int.
     HIRs 5-8 do reformat print code, and do not perform normal regis-
ter arithmetic on print code contents produced by Ops 1-4.  However,
print code can be synthesized and put into HIRs 5-8 in such a way as
to make possible normal register arithmetic.  This leads to the ability
to modify print code directly in a print buffer, obviating the need to
use ordinary registers for such modifications.  The printer appears to
act upon the ten LSDs of the 13-digit mantissa or a print-buffer HIR:
a situation similar to SR-52 fractured digits synthesis (V1N2p5).
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
  The SR-52 Users Club is a non-profit loosely organised group of TI PPC owners/users
  who wish to get more out of their machines by exchanging ideas.  Activity centers
  on a monthly newsletter, 52-NOTES edited and published by Richard C Vanderburgh
  in Dayton, Ohio.  The SR-52 Users Club is neither sponsored nor officially sanctioned
  by Texas Instruments, Inc.  Membership is open to any interested person:  $6.00
  includes six future issues of 52-NOTES; back issues start June 1976 @ $1.00 each.
Then synthesizing print code, it is necessary to fill the 3 MSDs with
non-zero numerals, then follow these with the desired ten digits of
print code.  For example, to artificially get the printer to print
ABCDE in the leftmost print sector, key 99913141 EE 5, store with
HIR 05, key 51617, SUM with HIR 35.  Now key Op 5, and see ABCDE
printed.  At this point, if you want to change the C to say an *, key
3.6 EE 5, and sum with HIR 35.  Key Op 5 and see AB*DE.  If you try
to start off with 1314151617 Op 1, HIR 35 register summing won't work;
anyone know why?
     Register arithmetic in the HIRs only works for integer, full
integer-fraction, or floating point operands.  The implied decapower
sign of a fixed point fraction gets changed from minus to plus (Posi-
tion B changes from 4 or 6 to zero or 2 (see V1N1p5 and V1N4p5)).
     User access to the HIRs will undoubtedly find many practical as
well as esoteric applications.  In an arithmetic teaching application,
Heinrich has written a program that zeros HIR 1 prior to halting for
the student's answer.  Following the input answer, HIR 1 is tested
for zero, as a cheating indicator.  In a biorhythm program, he uses
HIRs 2-8 for storage of intermediate results, freeing data registers
for print code.  This program also demonstrates the plotting of multi-
ple points, and makes use of CROM program ML-2O to perform calendar
calculations.  Fine tuning the partitioning to an effective 463.61
saves otherwise wasted space, but requires repartitioning during
program/data entry:  479.59 to key in the program, and 399.69 to pre-
store the last two print constants.  The program that follows contains
an English translation of Heinrich's German messages; run it by
pressing A and following printed instructions.  It appears that clever
accessing of the ML-20 program via the SBR nnn approach (V2N7p1) and
making appropriate register reassignments could save quite a few pre-
call data formatting steps, and perhaps some of you will want to try
this.  With the listing of this program, I am trying out some new
symbol conventions:  The * symbol for 2nd is dropped, L=Lbl, S=STO,
R=RCL, *=Ind; rtn=INVSBR, and groups of numerals requiring synthesis
will be bracketted by the " symbol.  For example, DSZ 56 182 is
written:  Dsz "56" 1 "82" indicating that the 56 and 82 must be arti-
ficially merged; p82 is written HIR which, ofcourse, must be synthe-
sized.  At least one member has confused xEt with x=t, so I am substi-
tuting xXt for xEt.  Leading zeros are omitted when address codes are
followed by non-numbers.
     Robert Snow (212) was probably within a few weeks of making
Heinrich's discovery independently, having reported same to me today
(1 Sept 77).  Also today, a question from A B Winston (707) pointed
to a use of the HIRs to provide a means for a running program to
detect whether the 58/59 is plugged into the printer:  Op 1 puts
printer-formatted display into HIR 5 only if the 58/59 is plugged
into the printer.  The sequence:  ... 0 HIR 05 7 Op 1 HIR 15 OP x=t
1'... transfers to L1' if the printer is not connected, continues if
it is.  This would allow programs to be I/O-optimized for operation
both with and without the printer... a feature that would especially
enhance many CROM programs.

                              V2N9p2 52-NOTES
TI-59/PC-100A Program:  Biorhythm Analysis   Heinrich Schnepf (376)/Ed

Program Listing:

000:  LE S7 rtn LE' 4 LD' S6 Op 0 R*7 EE INV EE Op*6 Op 27 Dsz6 014 Op5
029:  CLR rtn LA' 1/x X R4 = INV Int X 360 = sin X 7 = fix 0 EE INV fix
054:  INV EE - 7 = div 5 = +/- rtn LB' S7 INV Int EE 1 INV log = SUM*7
077:  CLR rtn LC' div 10 S42 S0 - INV Int Prd 0 + 1.2 SUM 0 = Int EE 2
103:  + R0 INV x≥t 112 + 2 = INV EE rtn LB 27 E E' E' INV stflg7 Adv
127:  rtn LC 35 E 1 D' R/S Prt EE +/- 4 HIR 2 2 D' R/S Prt EE 2 HIR "32"
150:  1 D' R/S Prt + HIR "12" = S0 Adv rtn LA 7 Op 17 Adv Adv 50 E E'
173:  Adv E' E' Adv 6 Op17 43 E 3 D' 2 D' C Pgm20 A Op19 INV ifflg7 202
198:  B GTO 180 2 D' R0 Pgm20 D S42 x:t 4 INV x=t 222 8 E GTO 230 10
224:  SUM42 R*42 S42 2 D' Adv 43 E 3 D' 2 SUM07 2 D' C Pgm20 B Pgm20 C
250:  Op19 ifflg7 260 CP x≥t 264 B GTO 233 S04 78 SUM5 Adv 17 E 2 D'
276:  2 D' 2 D' Adv Adv E' R5 - (365.25 X (1/x x (R5 - 122.1 )) Int)
310:  Int = S3 div R14 = Int HIR 4 X R14 = Int INV SUM03 1 x:t HIR "14"-
334:  13 = x≥t 344 + 12 = HIR4 10 x:t R3 C' EE 4 INV EE + HIR "14" C'
360:  + 2 EE 5 = S3 CLR S0 S1 S02 23 A' HIR8 x:t 33 A' HIR6 28 A' HIR7
390:  x=t 404 HIR"16" x=t 412 47 X HIR"18" B' HIR"17" x:t HIR"16" x=t
410:  418 50 X HIR"17" B' 51 X HIR"16" B' .24 x:t R1 EE div 6 INV log =
436:  INV Int x≥t 447 214 EE 4 SUM1 CLR E E' Op25 Op24 Dsz"42" 3"61"
458:  GTO 283

Prestored Data:

08:  3616134500 4317163117 3613374135 364131 303231 37411736 30.6001
15:  3723413536 213524 3624151327 4764332345 3713270000 5064301731
21:  3524376527 5164363324 3332364000 4000010000 311722 1613371700
27:  3033413717 6537001532 1632173631 1613371700 2213243140 24370013
33:  1700373545 3327171336 4517133520 7100000000 3032313723 1613457100
39:  13000000 2437002436 1613454000 364131 24310000 1700261745
45:  3327171336 1613371720 1424353723 16133717 3637133537 4536243651
51:  13311327 2345372330 5114243235 3723351717 3600243100 16133717
57:  1731371735 35633640 43243723 4017131523 3313353736

Printer Head Cleaning (PC-100/PC-100A)
     Since TI reports that running one of its printer head cleaning
programs with heavy paper cleans the head by heating the dot-matrix
elements, it would appear that programs energizing the most dots would
be best.  The TI-58/59 program (page VI-12 of the owner's manual)
prints repeated lines of 20 8s, which exercises only 340 of the 700
dots.  A routine along the lines of:  LB Op1 Op2 Op3 Op4 Op5 rtn LA
3232323232 B 7676767676 B 2424242424 B GTO A cycles through the symbols
0 (code 32), capital pi (code 76), and I (code 24) which together
exercise all 700 dots, when run by pressing A, and stopped with R/S.
With regular printing paper installed, this program may be used to
check out proper functioning of all the print dots.
     The comparable 52 and 56 programs can also be substantially
improved.  Here the apparent objective is to exercise all the dots
that can possibly be energized by these machines.  At any position a
numeral can be created (2-12, 14-16, 18-19), all 35 dots should be
exercised; one should try to find the minimum set of trace mnemonics
that covers all position 20 possibilities.  I'll publish the best
SR-52 and SR-56 printer head cleaning routines you send in.

                              52-NOTES V2N9p3
Magnetic Cards (59)

     Exploring Mag Card Protection:  Lou Cargile (625) has found that
he can get a protected card to read without setting the protection flag
by deliberately causing it to misread.  The types of misreads that work
may be machine-dependent; others trying this are invited to share
results.  In the meantime, a few questions from Jared Weinberger (221)
bring up an important aspect of card protection not clearly covered in
the owner's manual:  reading any protected side of a card sets the pro-
tection flag which affects the whole machine until cleared with a manual
CP or by switching the power off.  For example, if the machine is parti-
tioned "959." (no data registers) and Bank 1 recorded as protected on
one side of a card, when that side is read, all of memory is protected.
Repartitioning and/or p31 instructions encountered during program execu-
tion of protected code are ignored.  However, any bank that is not fully
partitioned for program memory can be copied normally on a separate card
even though the protect flag has been set.  The protect flag can then
be cleared, the card read, and the code examined.  For example, cycle
your machine off-on, then key 10 Op17 LRN LA 123 R/S LRN 999 S99 888
S98.  Now key 1 +/- Write and feed a card through.  Cycle off-on, key
10 Op17 CLR, and read the card just written; see -1 displayed.  Press
A, see 123; RCL99, see 999, R98, see 888.  But press LRN, and the machine
stays in RUN mode, and any attempts to repartition fail.  Now key 1
Write and feed a blank cardside through; see 1 displayed.  Cycle off-on,
then key 10 Op17 CLR and read the card; see 1 displayed.  Press A, and
the RCLs, and find things the same as for the protected card.  But now
you can get into LRN mode and/or repartition.  The first cardside was
sort of partially protected:  once read, the program part could not be
examined in LRN mode, but it could be copied to a second cardside
normally, which could then be read and examined normally.  This partial
protection situation also applies to all of memory, not just to the one
bank that has been read from a protected card.  For example, a 319.79
partition says that all of Bank 1 is program, Bank 2 is a mix of pro-
gram and data, and Banks 3 and 4 are all data.  A cardside protect-
recorded with this partitioning from any of the 4 banks, when subsequ-
ently read denies only Bank 1 from being recorded.  A consequence of
all this is that the read-in of one protected cardside, by virtue of
its having set the protection flag and frozen the partitioning, pre-
determines the rules by which all of memory will behave, regardless of
whether subsequent card-reads are by protected or unprotected sides.
     Incidently, although resident code affected by a protected card-
side read cannot be SST'd, you can access it selectively by SBR nnn.
This may make it possible in some cases to synthesize protected code
by observing results when SBR nnn is executed repeatedly for all nnn
from the end of the program partition back up through step 000.  Then,
if you're lucky, you'll intercept a sequence like:  ...GTO 182 A rtn,
and a SBR call to the step containing the 82 will return with the con-
tents of HIR 1 in the display!
     Mis -reads -writes:  Joel Rice (3) reports considerable difficulty
getting cards to read properly.  We may find that because of the greater
information density, 59 card read/write won't be as reliable as the
52's.  Others having serious read/write problems are invited to share
them.  I've found that clearing program and data registers before a
read, or a write key-in appears to help, as does wiping each card with
a clean lint-free cloth, which should also be used if after read or
write you can't remove the card without getting a fingerprint on it.

                              52-NOTES V2N9p4
     Mag Card Read and Write Under Program Control:  The owner's
manual (VII-5) notes that mag cards can be read under program control,
but doesn't say what happens when executing code is overwritten by a
read, or that card write can also be effected under program control:
a feature brought to my attention by Bob Moore (488).
     It turns out that if a cardside destined for Bank m is read by
an INV Write instruction being executed in Bank n, n≠m, execution
resumes at the step following the INV Write after the card has been
read.  But if m=n, and this bank number precedes the INV Write,
execution of the old code continues past the m INV Write until that
part of the read has been completed.  Then execution continues on into
the new code at the step following where the old code last executed.
This amounts to as much as a second or so of parallel processing: card
read and code execution occurring simultaneously.  Duration depends
upon where the INV Write is located within its bank:  the closer to
the end of the bank, the longer the parallel processing.
     SR-52 Cards for the TI-59:  Mack Maloney (246) has found that
SR-52 mag cards are the same thickness as the 59's (.008") and that
if they are trimmed to the 59's card width (.635") they appear to
read and write properly.  The .025" extra length doesn't seem to
matter.  As more users try this, we should be able to determine whether
there is any problem with TI-59 re-writes over SR-52 code, which may
not erase all the old code (a reported problem for HP-67 users when
attempting to use old HP-65 cards).  To be safe, you can first demag-
netize the SR-52 cards.

A PC-100A Typewriter (58/59)
     Thomas Cox (9) suggests that "It would be interesting if some-
one could devise a useable routine for the TI-58/59 that would use
the Rausch keyboard to generate the corresponding characters on the
PC-100A."  The following program does what Thomas has asked for, run-
ning somewhat like a slow typewriter, providing analogues of the hori-
zontal position indicator, carriage return, end-of-line bell warning,
and replaceable characters of real typewriters.  This is the sort of
program that is worth refining to cut execution time, since there are
broad practical applications.  So send me your better versions...
perhaps TI could be persuaded to put the best one in a new CROM.
     Incidently, the rather surprising coincidence (or did TI plan it
that way?) that the last column number (20) is also the print code for
the hyphen symbol saves a couple of keystrokes.  For the 58, partition
159.39; for the 59 turn-on will do.

TI-58/59 Program:  PC-100A Typewriter                         Ed

User Instructions:

1.  Initialize:  Press A, see 1 (ready for first letter)
2.  Key ith letter via Rausch overlay (V1N3p2), see i=1; display
    flashes when i+1=20, in which case either Rausch-key last letter
    to fill line, or press E to end line with a hyphen.  Repeat step 2
    until message is complete.  To end an incomplete line with blanks,
    follow last letter with R/S.  To generate a non-letter character,
    key the 2-digit print code and press E (instead of Rausch keying).

                              52-NOTES V2N9p5
TI-58/59 Program:  PC-100A Typewriter                         Ed

Program Listing:

000:  LA Op00 20 x:t 4 S28 1 S31 S33 L2' 5 S00 1 EE 8 S30 CLR S32 L1'
030:  R33 INV x=t 3' Op55 L3' R/S R32 Op*31 GTO 4' LD + 9 LC + 9 = LB
057:  S29 CE R*29 LE X R30 = SUM 32 100 INV Prd 30 1 SUM 33 Dsz0 1' R32
084:  Op*31 1 SUM 31 Dsz"28" 2' L4' Op5 GTO A

Prestored Data:

01:  36 42 45 25 30 33 13 16 22 37 43 46 26 31 34 14 17 23 41 44 0 27
23:  32 35 15 21 24

Membership Address Changes
     3:  230 W 107th St #6J New York, NY 10025; 48:  3611 Wyatt Dr
Holiday, FL 33590; 75:  Box 699 College Station, TX 77840; 281:  1502
42nd St #3 Brooklyn, NY 11219; 306:  4523½ Avocado St Los Angeles, CA
90027; 343:  1325 Quaker St Golden, CO 80401; 346:  37 Winding Way
Madison, NJ 07940; 347:  VQ-3 Box 65 FPO San Francisco 96637; 440:  148
Transmitter Rd Oak Harbor, WA 98277; 449:  807 Ruggles Hall Columbia
Univ New York, NY 10027; 471:  Box 461 Storrs, CT 06268; 584:  266 Main
St #325 Windsor Locks, CT 06096; 587:  501 Loring Ave Los Angeles, CA
90024; 594:  45A Hillview Ave Rensselaer, NY 12144.

     How the PPCs Work:  Gene Werner (120) has kept a record of some 20
or so books and magazine articles published over the past few years
covering various aspects of calculator and PPC design.  Send Gene a
SASE and a few stamps for his bibliography.
     TI Trig Algorithms:  Joel Pitcairn (514) reports having mechanized
the CORDIC technique (see Byte Aug 77 p 142) for calculating the tangent
function and comparing results with the SR-52 built-in tan function.
For base ten and Max=13, Joel finds such close agreement, especially
for critical values, as to conclude that this is the approach used by
TI in the firmware.  Execution time comparisons suggest that the sin
and cos functions are derived from the tangent; it also appears that TI
also uses the CORDIC algorithm to get arctan.  Joel notes an error in
the Byte article:  arctan xo=zi+xi, not just zi.  Joel's findings are
likely to apply to the other TI PPCs as well.
     Owner's Manual Versions (58/59):  Mack Maloney (246) reports that
there are currontly 2 versions:  1014983-1 and -2 (look on the lower
right corner of the back outside cover to see what yours is).  The -2
reportedly corrects most of the -1 errors, and has its own addendum
sheet (#1019286-4).
     Numerical Solutions To Partial Differential Equations:  Larry
Gearhart (442) would like to get in touch with anyone pursuing numeri-
cal solutions to PDEs via one of the TI PPCs.
     More on INV and Ind Viability (52):  Claude Belanger (254) notes
that while INV and Ind viability are maintained through direct sub-
routine calls (to A-E') (see V2N5p4), viability ts lost if the INV or
Ind is followed by GTO or SBR.
     TI Notes:  Check with your dealer or recent TI ads for details
concerning the free Leisure Library CROM for all TI-58/59 owners.
     PPX-59 is expected to get under way 1 Oct 77 with an initial
catalog covering CROM programs and conversions of "the top 200" PPX-52

                              52-NOTES V2N9p6 (end)