Summing up the HP 33S
Contents
1. g SU MS D Rmdr Ch RN AIG NI s 1 ATAN CHG H F G ASIN SGN F cos ey Gy i L LEAR K nPr c nCr Fig 2 The HP 33S RPN ALG Scientific Calculator Actual size 15 8 x 8 3 x 1 61 cm Summing up the HP 33S Page 5 OF PARTS EJ Fig 3 On the same scale the HP32SII as last seen in 2002 Actual size 14 7 x 8 1 x 1 5 cm Summing up the HP 33S IE aAa PACKAR RPN SCIENTIFIC RAD 6 gt in PSE R S a b c SPACE Page 6 ElI LcD1 aN y i k 4 BT2 m f SI ay erz__ SR2 PI 10 5 GTF 05 94VOAON Fig 4 Internal view of the HP 33S What s so wrong The early production batches that W odek Mier Jedrzejowicz referred to in his previous Chairman s Bytes have provided HP with a good many first hand impressions from contributors to the forum at hpmuseum org Paul Brogger and sharp witted Ben Salinas being the most active Leaving aside the chevron keyboard the most often repeated criticisms are that it is too big to fit comfortably in a shirt pocket that the decimal point is too small that there are too few new features and that users cannot take advantage of that much RAM As to the radix point size issue there s only one thing for it to upgrade the LCD which according to HP will be implemented this July They should not pass up the chance to get rid of the annoying shadows too By inspecting the new functions it becomes apparent that t2. D 7 of the user s manual C 8 in the 32SII owner s manual VIEW and STOP instructions in programs that are being solved or integrated are executed only once not each time the routine is Variables Numbers in program _ Integers 0 through 254 1 5 lines bytes All other numbers 9 5 3 bytes 1 byte for each character 255 maximum Equations in program lines Summing up the HP 33S Page 3 called by SOLVE or Integrate to monitor their execution intermediate steps can be displayed by the sequence VIEW var PSE refer to pages 14 2 and 14 10 of the manual returns exactly the same results as the 48GX integral function including the uncertainty of the result i e the 48GX IERR variable The uncertainty but no the actual error tends to be a bit larger than the one returned by the 32SII The following exception taken from the August 1980 issue of the Hewlett Packard Journal shows that the integration algorithm is clearly different from the 32SII or the 15C which return a wrong result in FIX 5 32SII 33S l Jx 1 Ie ox 1 lnx FIX 5 FIX 7 FIX 5 FIX 7 00364500 SEES 5198 10 Running time 2 min 4 sec 5 min 6 sec The 33S is powered by two CR2032 batteries which are held in place by Warning Warning the plates shown in Fig 1 Appendix Turn off before Turn off before A advises the user to remove the changing changing Tee ly anion batteries and lightly press a on Please replace Please repl
3. 31 x 3 21 and the resulting displays In RPN mode 3 5 ENTER 2 GZ ENTER 3 ABH EN CMPLX X all stack levels are used 21 HWee In ALG mode 5 LE 3 E CMPLX ENTER x A 03 2 amp Cv Enter a There is no provision for interchangeable overlays Somehow they do not seem to be favoured in the slightest any more For instance the five additional bezels for the 30S F1908A are only available in Australia Despite the 11 new keys there are still 14 menus whose names are no longer printed on darker backgrounds actually lighter on the black 32SII The functions in the 32SH PARTS and PROB menus are now on keys while there are two new menus the CONST menu CONST which displays an up to date list of 40 physics constants in metric units nearly all the constants on the 48GX plus a few more such as magnetic moments muon mass conductance quantum etc and the menu displayed by and fea Rt in ALG Summing up the HP 33S Page 1 mode to review the stack four registers that keep the values returned by SOLVEJand A Although the 33S is in general faster it does not put the 32SII to shame Simple loops with counters run about twice as fast on the 33S provided the ISG and DSE commands are not used if 0 999 is stored in A this program LBL A ISG A GTO A RTN runs in 34 seconds on the 33S whereas it takes only 16 seconds on the 32SII With some exceptions such as the integ
4. S look less cluttered and helps the user find functions more quickly others find it easier to use it with two hands but had the designers have that in mind they should have placed the key in the middle in the lower right corner at least as on the 308 or the 9S mA co ome believe the 33S will never fit in well in their workplace I N ge do find this calculator much slicker than the 32SII shown in Nae ar Fig 3 Alas there s no accounting for taste The HP 33S is the N Np most threatening looking calculator I ve ever seen wrote Erice S gt A k Smith the awkward angles of the keys are very disconcerting Sy and somehow bring to mind the Cthulhu Mythos A nightmare Mem y r or surrealism Oh well I d better stop trying to understand it erg i Summing up the HP 33S Page 9
5. Summing up the HP 33S Jordi Hidalgo 1046 Upon reading the last issue some of you I think must have got the impression the new HP 33S scientific calculator is good only for those who are a Chinese and b all thumbs Let me then give you more information about this neat machine Except to a few collectors details of my beta 33S are pretty irrelevant so it s better if I refer to my production unit which is the latest version available to date early May Its serial number is CNA41101044 I ve been told that the A in the site code not present in the first production units is used for differentiating the multiple suppliers What s new It is in the current context of resorting to redesigns of discontinued models that we must view this calculator but the result is quite good all the same In essence the 33S is a 32SII with much more memory 31KB more keys though somewhat creased as described by Craig Finseth in his HPDATABase a 2 line by 4 character dot matrix LCD a few new functions and the inevitable algebraic entry mode which is not so useless as in the 12C Platinum but equally unnecessary since the 32SII could already handle algebraic expressions as well Some users will prefer the ALG mode however when operating with complex numbers as calculations are being displayed all the time although the RPN sequence is shorter and is not required to view the imaginary part Compare the following ways of evaluating 5
6. ace against both battery contacts 1n batteries one batteries one case of malfunction I ve never seen ata time ata time a Chinese coin but I rather suspect it s just a vestige of the manual for the 32SII As Leif Harcke wrote in V22N6 the manual is a one to one mapping from the 32SII manual 388 pages printed in black and blue but in a sans serif typeface this time There are indeed a few errors in the adaptation that need fixing but that s what revisions are for Although the ALG mode is wisely confined to Appendix C some sample programs have been transcribed to ALG This is not so bad as it sounds because programming in ALG has its own tricks e g how to use or which willing users can learn by studying the listings in the manual and in the learning modules there s no other way I m afraid In a posting to the comp sys hp48 Usenet newsgroup Bill Platt regrets that Horner s method is now written in ALG Too bad they did not keep the RPN version so that users could compare 5x 2x in RPN assuming x is on level one ENTER ENTER ENTER 5 X 2 4 J K X to the ALG routine supplied QIK ROLLY 42 x RCL DO SJ ENTER assumes x has been INPUT Still the ALG program for the general case Ax Bx Cx Dx E on page 12 25 is one line shorter but 25 5 bytes larger than the 32SII version Fig 1 The battery compartment Summing up the HP 33S Page 4 hp 33s scientific calculator
7. he designers did not have the slightest intention to undertake massive changes to the 32SII firmware Writing a complete simulation Summing up the HP 33S Page 7 was challenging enough I guess Gene Wright says it all in this posting to the forum My guess is HP said something like this We re really moving the 32SII to a new package and aren t going to be making many changes to it at all We can t sell the old one because we can t make them any more We face a choice guys the cost of a 2K chip is the same within pennies as a 32K ram chip which do we put in They went with the 32K ram Will it be entirely useful I don t know I expect perhaps not But it is there to try to use With all the room for improvement however it is quite a good remake What s wrong with more memory and an optional ALG mode Nothing Well nothing much Trouble is that there are up to ten thousand program lines at our disposal but only 26 program labels and 33 variables in total Seen from the outside giving the user more data registers even if only indirectly addressable does not appear to be too much of a hassle but Kinpo must know best A trick to create lists of constants with the ISG command and Paul s article on how to minimise label use both posted to the forum will help us get by with what we have The algebraic mode is neither fish nor fowl To quote Dave Johnson a prefix postfix mix characteristic of low end TI calculator
8. l The 33S user s manual omits such details On the 33S SHOW readings for equations length do not match differences between two consecutive MEM readings amount of memory used Notice that nibbles no longer come into play Amount of Memory Used Data 32S 33S 8 bytes per non zero value No bytes Variables including No bytes for zero values i and the statistics registers Saen ee data 48 bytes maximum 8 bytes _ take up no user memory for each non zero register Always allocated 15 bytes No short form for frequently used values but constants in program lines take 3 bytes unless they are in an equation lines 1 5 bytes 1 5 for each function 9 5 or 1 5 for each number Each and each uses 1 5 bytes except the for prefix functions Integers 0 through 254 1 5 Each entry in the equation list bytes All other numbers 9 5 takes 6 bytes 1 byte for each 1 5 bytes character 255 maximum Still more telling is the fact that the algorithms for and LO have been rewritten There is a detailed introduction to the new solver in the document Advanced Use of the Formula Solver one of the numerous training guides for the 33S available at HP s website SOLVE first tries to give a direct solution by rearranging the equation The new message INVALID VAR is displayed when attempting to solve an equation for a variable not present Incidentally SOLVE now fails to find the pole in the example on page
9. module Applications in Medicine It s fascinating to compare the two approaches as both authors carry out a careful presentation for different audiences If I may digress for a moment By bringing expert users in HP take advantage of the vast knowledge and experience of the user community We know only too well that the dissolution of the Corvallis group meant a lack of continuity in too many things mastery innovation responsiveness etc True continuity appears to be in Summing up the HP 33S Page 8 the hands of users who know how older products were designed where function keys were placed or the best angle to view the display at wrote Bruce Horrocks in his first editorial V19N6 Whether such users agree or not is another matter While many enthusiasts swear by the 42S others just don t understand why it is such a sought after model on eBay Quite telling is a posting to comp sys hp48 by Hydrix CEO Jean Yves Avenard software team leader as was preserved at groups google com like a fly in amber In my opinion you d have to be a nutcase to buy a 42S at that price 320 399 I didn t want to buy it when it was still being sold why would I pay so much for something that has so many problems and such a poor user interface HP Lovecraft vs Salvador Dali It s about time we talked about the keyboard The following thought that Diana Byrne R amp D project manager for the 48GX and the 38G wrote in her ar
10. ral below solving and integration calculations are 1 5 3 times faster on the 33S These are the new functions and operations e In addition to the engineering display format the programmable functions and ENG cause the exponent of x the contents of the X register to change in multiples of three e W and GZ move to the bottom top of the equation or program list e AG and GY RPN programmable logic switches e The real number functions WG EN 2 and FA SGN sign of x e F UNTG returns the greatest integer that is less than or equal to the argument Same as FLOOR on the 48 49 e MN same as INTG y x It returns IP y x on early units E Rmdr same as the 48 49 MOD command i e y x INTG y x e 0 fs SEED stores a random value as the seed for Wea RAND The SCRL function has been replaced with cursor keys The E for busy and HYP annunciators replace the blinking PRGM and EQN and the HYP prefix operator respectively Flag 4 now has its own annunciator And there is now a RESET hole on the back the second edition of the manual page A 4 states that stored data usually remain intact when pressed but it always wipes out the memory in my units Besides the changes derived from the two extra digits in the display there are several differences between the 33S and the 32SII The length of an equation is no longer limited only by the available memory but to 255 characters When entering eq
11. s Nevertheless users can now choose from three different modes since the equation list has its own entry system Remaining bugs in my production 33S are the 0 SEED bug described in V23N2 page 4 nCr returns bogus results for large arguments e g 2000nCr1000 the bug reported by Ben after exiting a self test LC L or LC U4 the RAD GRAD 0 4 and HEX OCT BIN annunciators are always cleared as well as the stack level one but the corresponding modes and flags are not actually altered i e if radians mode was active before starting the self test it will keep operating in radians mode after the self test even though the RAD annunciator is not on Speaking of LC key combinations and not Le is now the still undocumented Oops key which cancels the CC W C L4 and C l L operations so long as is pressed before releasing LC like ON B on the 48 49 The forty two learning modules and training guides for the 33S that Gene and W odek have written are evidence less of the need for outsourcing than of HP s resolution to recover the good name they once had The modules remind me of the instructional materials produced by the HP Educator Programme see V20N2 pages 3 and 30 One of the activities for the 30S prepared by Colin Croft in Investigating Mathematics through Patterns Volume 2 dealt with Dubois formula for estimating body surface area given a person s height and weight The same problem is tackled in the 33S
12. ticle for RCL 20 will put us in the picture It s hard not to think about design while waiting for a flight at the Copenhagen airport Posters of Scandinavian designed chairs and lamps hang overhead as compact quiet luggage carts glide across beautiful teak floors These elegant yet practical items are not only nice to use they are also delightful Many people feel this way about the traditional keys on HP calculators they provide just the right feedback with a satisfying click Despite a somewhat loud click keys have a good if rigid tactile feel There s no sign of double shot injection moulding yet or is that now a red herring All keys on my units register flawlessly even when hit on a side Whether ENG and deserve to be primary keys or not is certainly debatable but the keyboard layout is a definite improvement on that of the first prototypes where W and were all primary keys see the cover picture on V22N5 If they were mistaken it is well to remember that ACO enjoyed no better success the ON key on the 10BII does nothing once the calculator has been turned on Thinking back the first time I saw the keyboard I was reminded of the paintings of Salvador Dali Not that it s a work of art but it s hard for a Catalan not to think about Dali s designs this year when we are celebrating the centenary of his birth Still some functional design might well lie behind it all some suggest that the chevron keyboard makes the 33
13. uations trailing parentheses of two argument functions are now required on the other hand and can be used interchangeably when entering the second argument of such functions Also in equations expressions like EXP X 2 now mean EXP X 2 fixing certain inconsistency of the 32SII on which 25 52 evaluates to 0 while 5 2 25 evaluates to 50 A colon LJ and not a space is used for separating arguments of functions in equations both the colon and the space can be used in messages Calculations with TOO BIG numbers out of range numbers displayed in non decimal modes are now possible provided the result is not outside the range e g RPN 4 E 10 GW BASE 4 i e binary mode 1010 BASE 1 restores decimal base returns 4E9 while the 32SII returns 3 435 973 836 the largest 36 bit positive integer divided by 10 On the 33S if a Summing up the HP 33S Page 2 program is interrupted when calculating an integral such calculation cannot be resumed Lastly certain calculations do not produce the same results on both machines e g gamma of 9 29 which suggests that the CPU a Sunplus SPLB31A based on the 6502 is running native software rather than emulating the 32SI In fact MEM and SHOW readings have changed as well as the checksum for equations and programs The following table shows the allocation of user memory The 32SII values have been taken from pages 12 22 and B 2 of the 32SII owner s manua
Download Pdf Manuals
Related Search