| HP codename, series |
Paladin (HP-28C), Orlando (HP-28S), Champion/Clamshell |
| Type, Precision, Input Mode |
Scientific, 12 BCD digits, exponent ±499, Reverse
Polish Notation |
| Programmable |
Yes, in RPL
(Reverse Polish Lisp) with arbitrary program and variable names. In
fact, a program is treated as a variable containing program code (other
variable types are strings, numbers, vectors, matrices, lists).
It turns out that the HP-28S is a surprisingly
fast unit! It easily beats the Casio fx-880P, Casio fx-2000C, TI-74 and
the HP-32SII. Compared to the HP-15C/41CV/CX it is about 20 times as
fast. Only the HP-48GX/HP-49G are significantly faster.
The RPL programming model is similar to the HP-48GX and HP-49G. |
| Performance
Index |
HP-28S: 54 |
| Memory |
HP-28C: 2048 bytes, HP-28S: 32kByte
All variables, programs and the (infinite) stack are stored in this
permanent memory. |
| Display |
138x32 pixel black&white LCD, 23x4 characters |
| Special features |
Folding design, the top half can be flipped all the way to
the back of the calculator, 72 (!) keys.
User/system RPL (Reverse Polish Lisp) programming language.
Many data types: Real numbers, complex numbers, real & complex
vectors, real & complex matrices, lists, integer
numbers up to 64 bits, strings.
Symbolic equation manipulation (ie. differentiation, integration),
structured programming, graphics. A wealth of built-in unit
conversions. Numerical root finding (solver) and integration.
Directory-oriented file system, online help catalog of built-in
functions and unit conversions. Infrared printer port, works with the HP Infrared Printer. |
| Original Pricing, Production |
HP-28C: 5.1.1987 ($235) - 1/1988 ($235)
HP-28S: 4.1.1988 ($235) - 1.4.1992 ($195) |
| Batteries |
3xN |
| Dimensions |
Length 15.9cm, Width 9.4cm (closed), 18.9cm (open) Height
1.8cm (closed) |
| Links |
HP-28C
Reference Manual, Jan 1987 (PDF)
HP-28C
Getting Started Manual, Jan 1987 (PDF)
HP-28S
Owner's Manual, Nov 1988 (PDF)
HP-28S
Reference Manual, Nov 1988 (PDF)
A
number of interesting programs (PDF)
HP-28S
Software Power Tools (PDF)
Quick Reference (PDF)
Step-by-Step
Solutions For Your HP Calculator, Calculus, HP-28S, HP-28C
(English) 2.6.1988
Step-by-Step
Solutions For Your HP Calculator, Probability and Statistics, HP-28S,
HP-28C (English) 1.11.1987
Step-by-Step
Solutions For Your HP Calculator, Vectors and Matrices, HP-28S, HP-28C
(English) 3.6.1989
HP-28C
Hardware I/O Basics (MoHPC)
Utility Programs.
HP-28COM Serial Interface.
Infrared Interface Format.
Power Consumption.
Available Accessories. |
| Comments |
The HP-28C was Hewlett Packard's first unit using the RPL
programming language.
What a device! The designes really put a lot of thought in this
powerful unit, see special features above. Despite all this
power and the vast number of functions is the HP-28 fairly easy to use
- at least after spending a small initial effort needed to understand
the basic concepts.
There's really not much missing except maybe that
its response on inputs feels a bit sluggish. Also - more importantly - its
programming model is missing the C-style BREAK instruction
needed for early termination of FOR-NEXT, DO-UNTIL and WHILE-REPEAT
loops. And although it had an infrared printer port it lacked the HP-41C's famous I/O capabilties.
Nevertheless, it was HP's only true pocket-sized calculator that
offered the powerful RPL language. All other RPL models (HP-48/49/50)
were quite big. Only the HP-41C and HP-42S
offered a sufficiently powerful keytroke-based programming model with
arbitrary program names that allowed to build entire program libraries.
HP's later pocket-sized models (HP-32S/II, HP-33S, HP-35S) had a rather restricted "A...Z" program naming scheme.
Software:
- The HP-28C came with two different software versions: 1BB
and 1CC (late 1987).
- Select Hex mode and enter "#A SYSEVAL" to retrieve the
software version number on the HP-28C and HP-28S.
|
| Software Bugs |
Version 1BB bugs (thanks to C. Finseth):
- Turn a unit on and keep the ON key pressed for a
moment. Now press SHIFT. There is a 50% chance that the unit will turn
off.
- Using the Solver with a program that does not
return an object to the stack. Sometimes a system object will be placed
there.
- Trying to delete an empty row with shift INS while
editing a command line can lead to strange results and possibly a
Memory
Lost.
- Dividing a 64 bit number with its top bit set can
give wrong answers.
- Otherwise equal binary integers with different
word sizes do not necessarily compare equal. Try: #1 DUP 50 STWS 'A'
STO 'A ENTER ==
- If flags 57 or 58 are set the Solver might give an
underflow or overflow error while working to a solution.
- The Solver can also give incorrect Sign Reversal
messages.
- User-defined units that are multiples of degrees
Farenheight or Celsius give incorrect results.
- The unit "yr" has been defined as 365 days instead
of the correct 365.242 days. As "lyr" is defined in terms of this
unit, it, too is incorrect.
- RND does not work correctly in FIX mode (try "0
FIX .7 RND": you get 0 instead of 1, unless you have flag 51 or 52
set!). If you don't want to set one of these flags, use ->STR
STR-> to round the number.
- Actually, this demonstrates another problem.
Do: "0 FIX MAXR ->NUM" and you get 1.E500. Now, ->STR gives
"1.E500" then STR-> gives Syntax Error.
- String comparisons produce incorrect results if
the strings do not differ in their initial letter: "AA" "AB" < gives
0.
- EDIT and VISIT can give unexpected results if you
attempt to EEX or CHS an already-existing number. Use "-" and "E"
or delete and reenter the number.
- Memory can be corrupted by incorrect usages of the
derivative function.
- Merging left with the divide sign can give
incorrect results.
- Rounding of numbers in complex and matrix
operations is sometimes slightly different than that of the same pure
real numbers.
- When the origin is not on the screen, plot tick
marks are relative to the screen instead of the origin.
- The Out of Memory condition does not clear
typeahead: you may unexpectedly delete variables or programs.
- The Solver does not always clear all internal
flags. Do:
Memory Lost
0 1/X (gives "Infinite result" and
sets flag)
DROP
<< A DROP 1 >> STEQ
Solve for A without an initial guess. You will
get a System Halt.
- Non-real results can cause the Solver to store a
system object as the variable's value.
- Some special characters are not displayed on the
menu line properly:
character
displays as
&
e
|
Z
~
\
\.x
`
\:-
_
proportionality o]
(140 decimal)
- (Not really a bug) Turning a string into a name,
where the string refers to a name that already exists, will always hang
due to inifinite
evaluation recursion.
|
