Newsletter of the SR-52 Users Club
published at
9459 Taylorsville Road
Dayton, OH 45424
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Quasi-Alphanumerics: A Comparison Of SR-52 And HP-67 Capabilities
A few months ago an exciting HP-67 discovery (consequences of
which are still unfolding) opened up the possibility of displaying
quasi-alphanumeric messages on that machine... a capability with simi-
larities to SR-52 fractured digits and which got me to wondering about
using a mixture of fractured digits and numerals for SR-52 messages. I
don't think the value of displayed messages needs to be defended, but
as an insight into what matters most with limited capabilities, it is
instructive to examine what each machine can do. For purposes of this
discussion, display positions for both machines are numbered from left
to right as 0-13. An HP-67 user can store and recall special "numbers"
to and from data registers which when displayed can show the ten digits,
facsimiles of the letters: r, C, o, d, and E, the minus sign, and blank,
in accordance with what appear to be the following rules: Position 0
can only be either blank or the minus sign, each of the positions 1
through 10 can be any character except the minus sign, position 11 can
only be blank, and positions 12 and 13 can be any character except the
minus sign, but their combined value determines the position of the
decimal point (which cannct be suppressed). Once a complete display
line has been synthesized and successfully stored in a data register,
it cannot be merged with changing data; i.e. selected positions cannot
be masked. However, there is a way to modify pure data to be flicker-
displayed with minus signs, blanks, and decimal points. Messages can
be displayed effectively using the two built-in pause functions, a
software-generated flicker pause, and the data-entry-sensing capability
that flag 3 provides: automatic continued program execution when a
number is entered from the keyboard during a pause-displayed prompting
message. It is apparently by design that the 5 fractured digits look
like letters, since they are all used in the 2 words Error and Crd
generated by firmware and displayed as user cues.
So far as I know, in addition to the blank, the SR-52 can produce
only 4 fractured digits, only one of which (the square o) is very letter-
like. But before we dismiss the SR-52 message-display potential as
being too limited, let's see what can be done. I have experimented using
the following character assignments: A=°, B=8, D=0, E=3, F=', G=9,
H=", I=1, J=', L=1, N=", O=0, P=6, R=7, S=5, T=-, U=", Y=4, Z=2, and
find that when these symbols are grouped in the display to form familiar
words, intended meaning is generally conveyed. But it's probably best
not to hold to fixed character assignments, as specific juxtapositions
may suggest variations. For example, by the above assignments, the
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
word GOOD would be spelled 9000. But 9°°0 is probably more recognizable,
even though it might, in certain contexts, be interpreted as GOAD.
The message synthesizer program below may be used to see what var-
ious trials look like, and it produces the 2 numbers required to produce
the message by a running program using the sequence: RCL 98 + STO 60
RCL 99 HLT (with = pressed manually). After a desired message has been
synthesized, the contents of Reg 98 and 99 can be transferred to 2 pro-
gram registers (unused by a user program) for permanent storage. 14
characters are entered as single digits followed by A for the first one,
then either E (if to appear normally) or C (if to appear fractured). The
fracture codes are per V1N2p5. Positions 0 and 11 may only be either
blank or -, position 10 is only zero for the SR-52A, else must be one of
the ten digits (0-9), and zeros in positions 12 or 13 are displayed as
blanks. For example, try: 6A, 7E, 4E, 3C, 5C, 9E, 5C, 1E, 2C, 3C, 0E,
6C, 1E, 5E; then after you press =, see: TRY AGAIN OTIS. Or (SR-52 only)
3A, 9E, 3E, 6C, 3C, 5C, 2C, 3C, 3C, 5E, 7E, 6C, 5E, 2E; Press = and see
GET AN SR-52.
One can, ofcourse, mix messages with changing data. During the
synthesis process, substitute zeros for the positions to be dynamically
filled, and design the user program to put the desired data into these
positions via careful register summing for positions 1-10, and and multi-
plication for positions 12 and 13. Keep a copy of the original "template"
(with the pre-positioned zeros) intact, as a new template will be needed
for each data change.
Out of all this, it is evident that the limitations of each machine
are quite different. The HP-67 fractured digits are more letter-like,
but there are fewer "natural" multiple assignments (such as " standing for
H, N, or U for the SR-52); the SR-52 can mask selected positions, but
display isn't as automatic or fancy (pausable or flickerable); each
HP-67 line takes considerably more effort to produce, but each can be
coded into the contents of a single data register, recorded, and pro-
duced (displayed) with a single instruction (RCL). Neither the HP-97
nor the SR-52/PC-100 can print fractured digits.
Much of the incentive for trying to create messages on either machine
is due to the challenge: it is mildly difficult, and often considerably
frustrating. But, ofcourse, as soon as a PPC comes along with real
alphanumerics, nobody is going to bother, any more than we would spend
time now developing efficient manual keystroke sequences for the non-
programmable scientifics. But it should be fun for awhile! So send me
your best messages and/or applications, and perhaps quasi-alphanumerics
will become a lively part of the TI/HP users challenge.
SR-52 Program: SR-52 Message Synthesizer Ed
User Instructions:
1. Key position 0 code (3 or 6), press A, see 1.
2. Key position i code, press C for fracture, E for normal, see
i + 1; repeat for i=1,2, ...13.
3. Press =, see message.
4. To transfer message code to another program:
a. Write other program into memory
b. In RUN mode: Key RCL 98 STO ab RCL 99 STO cd, where ab is
address of program register to hold operator number and cd
the operand register.
52-NOTES V2N5p2
SR-52 Message Synthesizer Program Listing
000: - 6 = *ifzro 009 *stflg 0 CLR 9.00 STO 98 STO 99 .01 STO 69 9 STO 00
030: 11 STO 68 B *ifflg 1 053 *E' SUM 99 *dsz 035 GTO 061 *E' SUM 98
057: *dsz 035 B *E' SUM 99 B - 6 = *ifzro 076 *stflg 2 B *ifflg 1 096
082: - 1 = X 10 = SUM 68 GTO 103 EE +/- 12 SUM 98 B *ifflg 1 119 - 1 =
112: SUM 68 GTO 131 X 10 = INV *log *fix 0 EE *PROD 98 RCL 68 INV *log
136: *fix 0 EE *ifflg 2 146 EE +/- *PROD 99 *ifflg 0 159 1 +/- *PROD 99
159: RCL 98 + STO 60 RCL 99 HLT *LBL *E' X RCL 69 = *rtn *LBL E INV
181: *stflg 1 *rtn *LBL C *stflg 1 *rtn *LBL B .1 *PROD 69 1 SUM 67
200: RCL 67 INV EE HLT *LBL A = *rset
- - - - - - - - - - -
SR-52A/PC-100A
Peter Stark (321) notes that while the SR-52A has been alluded to
in 52-NOTES, it has not been properly identified. It is the "newer"
SR-52 I refer to on V2N2p6 under "Display Arithmetic Modification" and
so far as I know, such machinbs are not physically marked as such
("SR-52A" is not printed or stamped on the machine). Bob Edelen (100)
notes that Alan Charbonneau's Yahtzee program (V2N2p5) will not work
with an SR-52A, and suggests the following changes for use with the
newer machines: replace the 5 with 6 at step 98; replace the sixes at
steps 210-214 with ifflgs. All five dice will show, followed by an
irrepressible zero. While the TI trouble shooting manual (V2N4p4)
suggests a number of hardware differences between the SR-52 and SR-52A
machines, the only difference I am aware of that is noticeable to the
user is the 13th digit truncation.
Bob Edelen has made some electric power measurements and reports
the following: The PC-100A charges the battery PAC at a constant 12Oma
even when the power switch is off. His SR-52A with adaptor charges a
full PAC at 100 to 110ma with SR-52 off; on: 50ma. A dead PAC draws
120ma with SR-52A off; on: 60ma. SR-52A turned on with 0 displayed
draws 160ma, with full display 260ma, which discharges a full PAC 45ma
during charge. SR-52A execution (with the 2 minus signs displayed)
draws 145ma. Since continuous charging is not good for batteries, Bob
has modified his PC-100A to allow the SR-52A to discharge the PAC while
connected to the printer. Members interested in details are invited to
send Bob a SASE (note his address change elsewhere in this issue).
TI Notes
Machine Exchange at TI Centers: Although individual retailers may
be changing their exchange policies, it is still TI policy (see V2N1p2)
for its regional centers to continue to exchange good machines (of any
model vintage) for bad, that are still in waranty.
Adding Chips (52): TI is now making available to the public the
TMC 0599 chip via Mr S Riggs, Box 53, Lubbock, TX 79408 for $17.77 each.
Reportedly, 2 of these will provide the missing SR-52 registers (20-59)
if properly installed. Michael Rak (502) has obtained details from a
TI source, and if you are experienced at assembling modern electronics
components and are willing to risk machine damage, send 25¢ and a SASE
to Mike at 1823 El Cerrito Pl #F Los Angeles, CA 90068 for a copy of
his info. Neither Mike, nor TI nor the SR-52 Users Club assumes any
responsibility whatsoever for any consequences which might arise from
chip ad-on mods. TI does not now, and does not plan to perform this
mod for the general public.
52-NOTES V2N5p3
Tips
Mag Card Marking: Noting that the first issue of the PPX-52 news-
letter mentions use of a "Sharpie" pen, Mack Maloney (246) has found
that the Sanford Sharpie #49 is permanent, and that a Vis-a-Vis 69 is
smearless with markings that can be removed with a moistened fingertip
or cloth, and which seem to work in place of the write-protect mag card
tabs.
Another One-Step Error Producer: Jared Weinberger (221) suggests
repeating an operator to create an error condition. For example, the
sequence: RCL 99 + + RCL 98 = performs the desired addition as well as
creating an error condition.
More on Trig Function Anomalies: Karl Hoppe (507) notes that
attempts to take the sin, cos, or tan of a number whose absolute value
is between zero and 3.6 D-97 in degree mode (or 6.283185307180 D-99 in
radian mode) result in an error condition. This may be due to the
creation of underflowed intermediate results by the trig firmware.
IND Prefix to a Subroutine Call (52): Several members have noted
that placing an IND before a subroutine call has not yet been mentioned
explicitly in 52-NOTES. An IND prefix maintains viability through a
subroutine call, just as an INV does (V1N1p2). Further, the sequences:
INV *IND or *IND INV are viable together. For example, the routine:
*LBL A SUM 99 *rtn prefaced with INV *IND will cause the contents of
the register pointed to by Reg 99 to be decremented.
Battery Charging PAC: Noting that TI does not supply a battery
charging PAC for any of its PPCs, Bob Edelen (100) claims that the HP
PAC will work with connector mods, and a rubber band to hold things
together.
More on the 0 Divide Error Condition (52): Howard Cook (556)
reports that an underflow resulting from dividing a small number by a
number ≥ 10 produces the 0 divide error condition (see V1N1p2). How-
ever, product by a number ≤ .1 does not do the same thing.
Fast Relative Addressing (52): John DeHaven (548) notes that the
machine will branch to a label located within the first five or so steps
faster than it will with an absolute branch.
Routines
Odd/Even Determiner (52): Jim Rosenberg (516) has devised a clever
way to determine whether a positive integer is odd or even: ... STO 00
50 *PROD 00 *IND RCL 00 *ifzro GTO ..., which branches to GTO on odd,
continues on even. Here is another way a non-register (50) can be put
to good use (see V2N2p6).
Efficient Temperature Units Conversions (52): John DeHaven (548)
has an algorithm for conversions between Centigrade and Fahrenheit
which combines the 2-way calculations efficiently and which can be
mechanized by: *LBL A *E' *LBL B + 40 = X 1.8 *iferr *LBL *LBL *1/x
- 40 = CE HLT. With temp in degrees F in the display, press A for
conversion to C; B for C to F. (E' must not be defined).
Membership Address Changes
100: 1632 Yosemite Denver CO 80220; 128: 356 Dart Dr Hanover
PA 17331; 252: 2111 Bidwell Rd #C1 Muscatine, IA 52761; 353: 2725
Monte Mar Terrace Los Angeles, CA 90064.
52-NOTES V2N5p4
Friendly Competition
The version of Barbara Osofsky's 5 X 5 Matrix program currently the
challenger for HP-67 users lists below (see V2N4p2), and needs no
further comment (see V2N2p3), except that I take credit only for using
Rick Wenger's error-flag approach; credit for eliminating more keystrokes
and the potential wipe-out danger (V2N3p3) goes to Barbara.
Another challenger candidate is Karl Hoppe's (507) "Square of 70
Digit Factor" SR-52 program, which follows. While there might possibly
be a sufficiency of data registers, ten-digit arithmetic will be a
hurdle for HP-67 mechanization. One construct which Karl uses identifies
an EE viability of which I was unaware: Decapower display softening by
EE carries through a subroutine return into main-program continued
execution. For example, the call to D at step 112 returns 1D21 in the
display register, which then gets modified to 1D16 by the 6 at step 113.
SR-52/PC-100 Program: 5 X 5 Determinant and Inverse Osofsky/Ed
User Instructions:
1. Initialize: Key 95, press C
2. Key ei, press E; repeat for i=1,2,...25 with column-wise
catenation; see 1 displayed.
3. Get determinant: with 1 in display, press D; first col of
adjoint printed; determinant displayed.
4. (optional) Print Determinant. On printer: press ADV, PRINT, ADV
5. Get Inverse: With determinant in display, press D; first
column printed, 1 displayed.
6. Get ith column: Key i, press RUN, repeat for i=2,3,4,5;
ith column printed, 1 displayed. For new problem, go to step 1.
Program Listing:
000: *LBL *B' STO 68 (*IND RCL 66 X *IND RCL 68 - 1 SUM 66 +/- SUM 68
024: *IND RCL 66 X *IND RCL 68) X *LBL *E' 1 SUM 66 *rtn *LBL C STO 66
047: *rtn *LBL *C' (8 C 4 *B' *E' 17 *B' +/- + 19 *B' C 7 *B' - 14 *B'
071: *E' 17 *B' + 19 *B' *E' 2 *B' + 9 *B' *E' 2 *B' - 4 *B' 6 C 12
095: *B') div RCL 67 +/- STO 67 X *prt RCL 95 + *rtn *LBL *D' 10 +/- A
118: *LBL B 95 C *A' *A' *A' *A' *LBL *A' *IND RCL 66 *IND EXC 69
137: *LBL E *IND STO 66 5 *iferr 149 1 SUM 66 *LBL A SUM 69 *rtn
158: *LBL D STO 67 CE 0 STO 69 *C' B *C' B *C' B *C' B *C' 5 A
180: *D' *D' *D' *D' *pap 0 = HLT STO 69 94 A *x! B GTO 164
- - - - - - - - - -
SR-52 Program: 70 Digit Square Karl Hoppe (507)
User Instructions:
1. Initialize: Key 99, press B
2. Input up to a 70-digit positive integer (no decapower);
Key blocks of ten digits, each followed by RUN; begin with
MSD; see prior block of ten digits (from earlier problem).
3. Get square: Press A; after about 200 seconds, see the first
ten MSDs of the square.
4. Get remaining digits: Key 6 B; then key RUN for successive
blocks of ten digits. For new number, go to step 1.
52-NOTES V2N5p5
70 Digit Square Program Listing:
000: *IND *LBL E RCL 98 *rtn *IND *E' 1 +/- C *rset *LBL D - (STO +
019: *LBL *D' 1 EE 21 *rtn *LBL C +/- SUM 98 *LBL *C' SUM 68 *rtn
038: *LBL *B' RCL 68 - 6) *rtn *LBL *A' 18 STO 68 19 *LBL *E' STO 98
061: *rtn *LBL A 105 STO 69 *A' 0 *IND *E' *IND STO 68 2 C *B' *ifpos
082: 071 9 *C' 6 *C' 105 *E' *IND RCL 69 D 6 - *D' 6) STO 67 = STO 66
110: *IND E D 6 - *D' 6) STO 64 STO 65 = STO 63 X RCL 67 *PROD 64 +
136: RCL 65 X RCL 66 *PROD 63) D - *D') SUM 64 + RCL 63) *IND *C' 1 C
163: RCL 64 div *D' 0) *IND *C' E - 99) *ifpos 110 SUM 69 *B' INV *ifzro
187: 087 *A' *IND E + *D' - *D') *IND INV SUM 98 div *D' 0) *IND *C' 1
210: C *B' *ifpos 191 CLR 6 *LBL B *E' *rset
PPX-52 Gems
#300032A: 4 X 4 Matrix Multiplication, by Barbara Osofsky (420),
facilitates chain multiplications by replacing each successive multi-
plicand by the next product. The PPX program analyst's note (2) is
in error, likely to be misleading, and should read: "Insertions and
deletions in the program will change the first 10 entries of matrix 1".
As Barbara notes, warnings of this type should be made for any program
that uses program registers for data, since data stored in program
registers are equivalent to unrelocatable instruction code.
#910050A: Two to Nine Pile NIM, by R L Floyd (82) effectively
mechanizes the modulo two addition algorithm that handles up to nine
piles, each of which can be up to 1D1O in size. Richard was the only
one to respond to my invitation (V1N3p3), but noted at the time that he
was submitting his program to the PPX-52 library.
Book Review: Computers At Large, by C J Sippl (239) and R Bullen,
Bobbs-Merrill, 1976, 222 pages.
Here is a readable discussion for the layman that covers a broad
spectrum of man-computer relationships. Emphasis is on the sociological,
psychological and philosophical impact computers have made and will
make on man and society. The text is liberally sprinkled with quotes
which the reader can pursue in greater depth via a substantial biblio-
graphy. Discussion of the varied uses to which computers have been put,
and the broad range of consequences to humans provides good food for
thought.
Meeting the Sum-of-the-Digits Challenge (V2N3p2)
Howard Cook (556) beats Jared's shortest by one with: *LBL C EE
STO EE 00 - 1 SUM 69 = INV *ifzro C *EXC 69 *rtn, and notes that Jared's
routine A and his routine C effectively sum the rounded 10-digit display,
and that Jared's routine B sums up to 12 of the most significant digits
in the display register. Howard goes on with a new challenge: beat
his 56 step routine D which sums all 13 digits: *LBL A (STO - .5) EE
INV EE *rtn *LBL D STO 01 *fix 0 *log A INV *log A *1/x *PROD 01 *LBL
RCL RCL 01 *ifzro *EXC A SUM 69 INV SUM 01 10 *PROD 01 GTO RCL *LBL *EXC
*EXC 69 HLT, which he says is clumsy, but which has some clever constructs.
Subject/Author Cross Reference
Bob Edelen (100) has kindly volunteered to work up an index covering
the first 12 issues of 52-NOTES, and will cut mimeo stencils under
computer control, in much the same way Mike Brown did with the alphabet-
ized membership list. I'll be asking Mike for a follow-on soon, that
will cover only currently active members.
52-NOTES V2N5p6 (end)