if HP has cancelled so much, why not...


build them "ourselves"?

I'm not well versed in electronics manufacturing, but I'm not sure why it's impossible/implausible for someone to do so.

(and I'd dearly love an update to the 42s, with rs232, a 3 line display, 128k RAM, with graphics routines and "ppc rom" type programs built in. I'd go for something the same L&W, but up to 5mm thicker. I'd pay $149 for it in place of a TI89 without an eyeblink. I'd like a mix of the best of the 32SII features in there, too. the dual shift keys and on keypad alpha's. the built in pop/sample stats.)

You can't legally copy the HP, sure. But at this point in age/abandonment, there's nothing to stop people using the concepts of the keyboard snap, shape, and feel. And a from scratch RPN system that runs 41/42 code would not be impossible to build.

Just like the PDA world is slowly moving back away from graffiti type systems and back to mini keyboards, the world might be ready for calcultors that Just Work, that can be successfully programmed for work/lab use after 2 hour's instruction, and which are small enough to fit in a pocket.

I know this is just another gripe and "what if" post- but I'm really curious. I've got a friend who does performance engineering for a Big Unix Company in the bay who claims that it is impossible to start a "garage" tech firm anymore because you won't make shareholders enough money fast enough. Maybe so, but isn't that exactly what has killed off so much of the great engineering in large companies? Is it possible to go make a good calculator and pay the rent instead of "leveraging marketshare for massive retirement income"?



I'm in for the hardware part. You may recall a thread about me doing an MLDL for the HP41 based on a FPGA/CPLD. I have found a bit more time and I have changed the design to Xilinx and modified the phase counter and other parts. The simlations are coming together now and I should have a functionally equivalent of the original MLDL (and EPROM box at the same time) within a month or so. That is, in simulation. I am working on getting the hardware organized. The whole design is done in VHDL. Onze I have the sources cleaned up I will make it available to the group.
I would love to have a 'new' HP41 clone!



sorry, address should be: meindert@kuipers.to in the previous posting


Tooling for the case alone would be over $100K. Keyboard and LCD would be similar. Figure about a megabuck to get a decent high-grade machine ready for market... plus a similar amount for marketing, etc.

I just ran my piggybank through the Coinstar machine and it coughed up $100. If 10,000 of us do the same we would be ready to take a run at it.


If the intention is to sell calcs worldwide, then only corporations would have cash - but not motivation.

If the intention is to produce on a very limited scale, then things can be very different. What if we could produce eletronics for 41, woodstock and classic series ? If case, keys, keyboard and display are available from old unrepairable calcs, replacing electronics is not that expensive - it is also a very interesting technical challenge.


getting existing units back in service, at least. That only takes a new board for the one having the separate CPU(fullnut?)

No 100K then.

BTW, 100K is LOW LOW LOW for the parts in the HP 41 case.



I don't know, I think you can do a calculator case for 100K$ these days. Tooling costs aren't what they used to be. CAD systems, electromachining, etc really help. So do various Asian countries hungry for business.


When do you start production? :-)

I vote for everything in the 41CX + Advantage Rom plus everything in the 42S. And I want the new machine to work like them, not like the 48/49 series (though I would like the ability to handle units.) A three or four line screen is sufficient. If I want to graph I will use Excel or Mathcad.

BTW, I can invest $100…maybe more after my daughter's wedding.



...Not to be a wet blanket, but I'd like to sound a note of caution. The question that you have to ask is: "Why has HP failed to maintain their leading edge calculator business, and is that something that is relevant to this discussion?"

Well, I think I know the reason, and I'll bet a lot of others suspect it, too: the PC. In order to create a successful calculator design, you must first identify some combination of requirements that CANNOT BE ADEQUATELY SATISFIED by simply producing another PC-based application. (Note that word ADEQUATE...) PC's have "enormous" (by our standards) full-color screens, lots and lots of keys, storage space and communications. Given platforms like Jornada, that's a tough combination to beat.

If you can identify the requirements, you have a chance to do something worthwhile. If you can't (or don't pay attention) you'll end up creating something that no one else cares about at all.


I can start by saying that I have yet to see someone actually whip out a PDA to do math.

And I *certainly* don't see students booting PCs to solve for the limiting reactant in lab.

The hand sized calculator fills a very specific need, portable math. The *efficiently* (I have to come down against the 48 on this one)programmable calculator isa life saver in the field. Even if "the field" is a datacenter with $5,000,000 in Sun computing hardware. (My last job before going back to school)

most of the students I know carry calculators.The vast majority of them can't find a factorial button,with hand holding- but that's "market share". I'm not sure I'm after large amounts of that.

I'm - personally- after the HP42SX with a better graphics system (okay, not for plotting, but for games, flashcards, and such -and try finding a decent chem/psych/stats flashcard app for my jornada 720!)

Yeah, I'd *love* to be able to export data over a serial port into SASand SPSS. But that's what a single CF slot for "roms" and a rs232 port are for.

Overly ambitious?

I don't know.

1 Million Dollars seems a bit much to get together, and at a $10 net, youhave to sell 100,000 to break even.

Maybe reycling the 42s/17bii components is a good idea- maybe finding the leftovers or tooling fromindonesia is a good idea.

I'd be willing to do some stone age mass production work to get them going.and I can surely pitch in my $200 worth.

(hell, start the deal out here in davis and I will work for free to get something going)



I just cannot leave Brazil easily, but I'll do what I can from here.

I can see (premonition?) a site created with this very purpose in a few months, as it grows completely out of control...

Count on me.




let's get back to when the HP41 was introduced. Available PC's (Apple, Commodore, Sinclair) were too expensive, had a lack of graphical resources, shown no compatibility with each other and could not be SW/HW upgraded. What do we have today? High evolved desktop/laptop/palmtop, high graphic resolution, fully upgradable, a lot faster than the first ones AND prices are about the same.

Once, Bill Gates compared computer evolution with automobile evolution. If automobiles had evolved as computers, they would cost a lot less than they cost and would run thousands of miles with a drop of fuel. What if we extend these comparisons to calculators?

We had extremely advanced computing devices at that time. I remember that the HP41 was unbeatable. Many people (me included) would never have a computer, but having an HP41 was mandatory.

The fact is: we still have our automobiles, being them perfect or not. Also, it's a fact that automobiles' "users" do not have the second option; calculators' users have. They can buy a palm, an organizer, and deal with them as calculators.

What we must offer is something computers don't. I'm figuring it out, but a good starting point is: "build your own portable computing device at home and get rid of ready-to-use stuff". Some of us would love building our own calculators, it's a matter of fact. And youngers are trying to "taste" something different. Maybe it's not happening everywhere, but as Linux shown itself a second option, now it is an option. I believe that this is mostly because it can be controlled by the user, not the opposite.

And it was brought to life by one guy only (that's what is written in the books). We are a bunch of guys. What do we have to loose?

Yours are wise words, and very well needed in a moment like this, because they demand reflection.




You've gotten to the deepest heart of part of why I want this.

I am always trying to find a "reasonable" development system for my PDAs- I want the same UI that the rest of the machine has, some simple language standard and functioning that will let me pop together what I need for a chem lab or a psych research project inside of a half hour.

My friends constantly tell me the same thing- over and over:

"It's a PDA, it isn't *supposed* to be a programming platform"

And that is *exactly* why you won't see a lot of people using them in the field, using them for onsite geology, surveying, or arch. digs. This is exactly why you don't often see a Jornada strapped ot the thigh of a cessna pilot.

It's made to run apps, not be user modifiable.

and that is, perhaps, the one key element that a pocket calculator needs that simply doesn't exist anywhere else. The ability to make elegant, useful, complex programs on the spot, in the field, in your hand.



Low cost linux handheld is an option.. that's the reason I started Scientific Digital Assistant project:

Bets regards,


Hi, J. Manrique.

As we are in here, this is a new, under development project. I think this is a good starting point.

I'm gonna read everything that exists in the site and get back as soon as I have all info I think I need.

Best regards.



I believe everything posted here must be seriously balanced. Christof, Meindert, Renato, David and many, many others are interested on an HP "clone". This is a brief post, just a warm-up for others.

Renato is completely right about case, keyboard and (rebuilt?) display/LCD. Many Hewlett-Packard calculators may (easily?) be reversed-engineered, at least the old, fixed 4-level stack models. Expandability is another story. And why not to try other design, other lay-out? Something like the HP97, say, portability + desk-top operation. I like having my calculators in hand, but I prefer using them in a more friendly, less hostile environment.

Using existing case + new electronics is a must to keep existing calculators with not-repairable electronics. Bringing them back to life with new "guts" would ensure prosper, long live to them.

This is just a brief starting...

(if any of you want to see a previous thread about this, have a look here. There are others about the same ubject, I'm just digging the Articles to find them.)


For low quantity production, there are manufacturers of ready-made standard electronic enclosures for small handheld devices. For example:



These are just a couple, and this not meant as an endorsement, as there are many other manufacturers out there.

For keypads, this Google search turned up quite a few:


The cheapest way to go would be an off-the-shelf peel-and-stick membrane switch array. Obviously it's nowhere near as good as the old HP quality keyboard, but it would be cheap & quick.

I suspect suitable LCD displays could be obtained from supplier's such as Digikey. (Lots of online, surplus electronics dealers could also supply those displays cheaper, but their availability isn't as reliable.)

So, there's the calculator case, the keypad, and the display. All we need is to come up with the brains.....


Whatever reality takes us to, of course.

but is that *snap* impossible to attain?

The important bits are that feel. it's not just HP- I have some other old lab hardware that has that feel. Like a tadpole made unix laptop. Or my 42s

I definitely- personally again, this is just me- want a 3 line display,maybe a few pixels widerthan the 42. I definitely want to makethe size compatible with my old bdu shirt pocket. (pioneer size) Thicker is fine.

And while I'm droolin over the thought of something like a full megabyte of storage space for pacs/formulae, I'm NOT in ANY way drooling for RPL. The coding model of the 41/42/32 is exactly what I'm after. (though an onboard assembler would be useful for adding to it without having to cart around a PC)

Extravagant, expensive, etc. I know. But the capability to have more RAM and serial interfacing are, honestly, the only things I *really* want to add to a 42S. (sure, I like the 32SII keyboard layout better)

I can dream- I can also say that I willdo my relatively meager best to make an actual project go.


... then we need electronics ... then we need software, but I guess there would be people willing to do that.
Just being able to recreate guts for classic, woodstock and coconut would be great. Should we care about selling calcs to students, or to 48 users, or to ti users ?



about the brains: mine is in running, second-plane mode.

I'm posting as I have news.

BTW, is there any good source of information about how to perform Floating Point operations in a binary-based system? If there is a concise, downloadable source of info about this, I'd like to read. My docs about this subject come from the Z-80 era, and I believe better algorithms exist today. If the docs are related to any specific processor (Microchip, ATMel, or any others), this processor should be the starting point.

As I told, I'm running in second-plane mode...

Thank you.

(have you noticed that it's Sunday, where I am in Brazil it's 9:20 PM and there is a lot of activity on this thread? Wow!)


Try http://www.radiks.net/~rhuebner/mathlib.html


I wrote a big text first, but I lost it and I think I also lost the primary focus; but let me try again...

Thanks, Renato; I'm downloading the files and I'm gonna read them soon.

Sometime ago, I read some articles about PIC compilers (I still have them) found in sources from Y2K. I am sure there are new improvements, also alternative HW sources (ATMel, among others). I remember that Microchip offered a C-like language for the compiler, easing the SW development.

I sometimes feel worried about using compilers to develop such low-level applications (let's face it: starting an O.S. form the bottom is easy with nowadays tools, but it is still a huge work to be accomplished...) because final executable is always bigger than the final one obtained by using plain assembler resources. In this case, time spent with the new design grows exponentially faster. Two variables to be balanced... Also, porting the existing ROM-dump from an HP42S or HP41CX would demand some time; otherwise, nothing else would need to be developed. I know this is not legal, and HP lawyers include a clause about using internal programs as being as being illegal practice. That's why I asked about floating-point routines.

If we find the most suitable nicrocontroller for the job and start the design from scratch, is there anyone that would guess how long would it take for us to offer a first operating assy? Just a guess, based on subject (instead valid) feeling. We are a spread-around-the-word team, and a lot of things will go against us, but we can simply deny this facts and go on as if we do not know it's impossible... or hard enough to give up. I believe at the first moment someone says "Hey! I have a prototype that allows arithmetic in a four-level stack register.", the others will ask for the "blue-prints" and O.S. listings, so we can go on at home (I would!); later, someone else would say "Hey! I add I/O capabilities to your prototype" and another one would say "Mine has registers!", and another would say "Mine handles ALPHA characters!'' and suddenly we would see a new (GNU Licensed?) RPN-based machine available to be homebuilt and homeimproved.

Shall we? Who else will dare this dream?

I need specs for microcontrollers, mostly for their development tools/compilers. Anyone would, please, easy my search and point some sites out? (I still do not know it's impossible...)


Real hardware is not necessary at first.

www.microchip.com has PIC simulator for download.

sdcc.sourceforge.net has C compiler and simulator for z80 and 8051

Also, AVR from atmel would be interesting.

http://kit.microcontrolador.com/index.html has a AVR board for US$110.00 (in Brazil).

http://www.mosaico-eng.com.br/ has PIC boards and information

http://www.symphony.com.br/pic.htm also has PIC boards, with development support


Any reason not to make a preliminary design- some hand hacked harware- with a flashable rom?

that would allow for binary updates through a CVS system of some sort. First stack, then sine, then....



the most flexible the development system is, the faster we can change/correct it. By flashable ROM you mean Flash technology, right? And implementation would, I believe, follow this guideline, too: stack manipulation, simple math, transcendental math, and so on...

That's my point, too.



Hello, Renato;

I agree with you that the running programs will not need the HW to be tested, but we need to define which HW to use before going any further. It seems to me that available microcontrolers can handle almost any level of implementation, so SW will not be a defining parameter.

Do you agree with me or I failed in my analysis? I want to know if I am aiming the wrong way before shooting... O.K.?


(this is getting a lot better!!!)


Luiz wrote:"we need to define which HW to use before going any further."
Renato comments: No, we don´t. We just need to compile nsim for a simulated microcontroller, then try it on a PC. Nsim is at http://www.brouhaha.com/~eric/hpcalc/nsim/ .
After we do that, we will understand what capabilities we need from the hw. Also, PIC simulator provides a stopwatch, so we will get performance data.

Luiz wrote:"It seems to me that available microcontrolers can handle almost any level of implementation, so SW will not be a defining parameter."
Renato comments: PIC family is huge, we can also choose 8051 or AVR, or Z80...

Also, using nsim and a 41rom we can have a 41OS running very soon on a microcontroller. Then we need NV memory, display, keyboard. Then we need case, batteries...


Hello, Renato;

I read what you post and I'm reasoning about it. I'm not posting in a hurry, so I'll not have to erase my answer later. You get a point here, let me just get it into my mind.




I am an engineering student very interested in electronics. During the 2001/2002 term I took lessons in digital electronics in which I studied PICs among other things. So I would like very much to add my point of view to what has been written here.

PICs are microcontrollers oriented to inputs/outputs, this mean that they can easily handle keyboards, LCDs, digital signals, etc. but they do not have a really huge calculus power. This makes them, I think, appropiate for a simple calculator, but I do not know if they can easily handle advanced calculus operations. If my knowledge is up to date, what I can not assure, their assembler language does not even include multiplication or division of integers. So, the required complexity of software is really big.

There are C compilers for PICs, but they produce too slow and large code to be really useful if you want to push the limits a little. This information comes directly from class lessons so it should be verified. ¿Anybody has experience programming in C for PICs?.

On the other hand, assembler programming for PICs is really easy, as the instruction set is really small and the Hardvard architecture makes programming really easy, together with the free programming tools given by Microchip at http://www.microchip.com.

Hardware requirements are really small, at least in prototype phase. I think that you could have, say, a
2 line LCD with a 32 key matrix keyboard and a PIC for less than euro 100, and probably if you manage to do it with a 16F84, less than euro 60. Aprox, euro 1 = USD 1. And no special tools are needed for assembling the thing.

Just my point of view, and without hardly studying the requirements.

Excuse my poor english.

César García.
CdU de la EPSIG www.etsiig.uniovi.es/asociaciones/clubusu


The cpus used in HP calcs are also very simple. What makes them powerful calculators is the software. You can find more info about this in the museum (see processors and simulations). There is software that could be ported to run on a PIC. See the simulators by Eric Smith at http://www.brouhaha.com/~eric/hpcalc/


Actually, it is a fairly trivial design to build a circuit board that could replace the guts of the HP35/45/70/80. Slice off a row of connector holes and drop it into a '55 chassis. Flip it over and it fits a '65. With a little fudging around it could even go in a 67. A similar board would drop into any Woodstock (finding those pesky fork pins is the main problem, pogo pins could be made to work)

A single PIC chip could easily handle the task. Progam storage in serial EEPROM could go to 64K bytes easily. Even a time/calendar could be handled.

I've thought of doing this many times, but have not seen enough dead machines that would be worth rebuilding. Those that can be repaired have been by scrapping corroded/cracked/carved machines that one really would not want to repair. It is still a likely project if I ever have a little free time.


HP is highly sought after because of the high quality (read high cost) of it's calculators.

If HP cannot keep the calculator line making money, a would-be startup company will not either. No one is going to buy a cheap quality HP look-alike. If they wanted cheap, they'd buy TI.

It is the high quality that makes them non-competitive, in the high volumn calculator business.


I agree with parts of that.

And I don't want a "cheap clone"- though I'm willing to accept that development may require some of that flavor.

I- again, this is Just Me, I can't speak for anyone else- I don't want a large share of the market. I don't wnat to "beat TI" or something.

But selling a few thousand calcs isn't an impossibility. If we can acheive the usability, portability, and quality of the 41/42/32sii styles of programming and use, with the rs232 and extra ram/rom capacity.... well, I honeslty think that, evne beyond a group of hobbyists making personal machines, we could sell a few here and there. enough to make carly want to acquire us? I *hope* not!

I recognize that the world has changed in the past 40 years, and that many people think there is no way a small group can do a thing- that all research, development, and production requires a multinational corporation or government. But I think that, if you aren't concerned about Making Lotsa Cash Today For Short-term Stockholders, it's still possible.

That having been said- it's still at the "group of hobbyists" level now. Why should we not try to build something?


I've said this before, so please forgive the "broken record" repetition . . .

Also, the fact that I haven't actually DONE anything yet (except buy the unit and download the software) must be taken into account . . .

The TI-83+/SE line offers a screen big enough for a four-line RPN interface, a decent (NOT rubber!) keyboard (with, incidentally, an "Enter" key), 32K RAM, over 1 Mb of Flash-able ROM (!), a fast Z-80 processor, a library of floating-point routines & transcendental functions, and a PC connection, all in a hefty package bigger than a -42s, but thinner than a -48G. Also available are a free emulator and assembler, tons of .pdf documentation, very helpful members of an interest group, and an internet discussion board.

(Meanwhile, the TI-89 is M68000-based, with similar resources available -- it offers much greater resolution in the display.)

It looks to me (after cursory investigation) that the TI-83's method of flashing and using "APPS" wouldn't provide a truly "native" feel to an RPN interface..) I suspect that one would want to disassemble and tweak the calc's operating system itself, not simply develop another add-on. Easier said than done, but it can't be impossible, eh? (There are, I think, two RPN Apps already available -- I plan to try them out before starting out on my own.)

But the process should be fun -- transform the interface to four-line RPN, revamp the storage scheme, enable keystroke programming as in the -41/-42S and voila! The next in line of the last (best?) generation of programmable RPN calculators!

I've always had an interest in portable, programmable computing devices, and HP provided the best available. That will no longer be so, with the passing of the HP-48/49. So, go to the next best thing!

Do take a look. It could reduce the project to a software development task only!


Hello, Paul;

your thoughts sound interesting, too. I'd like to know how much do these beauties cost. Is there any link you could point us to? If the specs (PDF files) are there, I'm also downloading them.



I got a (barely) used TI-83+ Silver Edition on eBay for $70. (Right now they're $129.00 retail and slowly coming down.) It included a Graph-Link cable and a "Resource CD" with manual, link software, etc.

The emulator, upgraded operating system, and developer documentation are available at the TI web site, including
Discussion Groups: Technology.

I looked at TI-89's and they seem to be going for about $90.00 on eBay -- they're retailing for something like $160.00 or so.

If there are any interested participants, we should agree among us on the platform (TI-89 or TI-83) and cooperate on investigating the full range of possibilities . . .


I'd posted this price I saw in another Forum message. It looked like that had a half-a-dozen of these.


The TI-83 Plus SDK

Somewhere at that site is a similar page for the TI-89/92, but I had enough trouble finding this one!


A couple more . . .

TI-83 developer's site

Ti-89/92 (Motorola 68000-based devices) developer's site


If anyone is interested . . .

I've been playing with "TI-Basic" on my TI-83+ Silver Edition. I figure I'll play with the calculator at that level before diving into Assembly language programming.

I decided to create an RPN user interface as well as I could using Basic, and use that as the UI model for an ASM effort (if it ever gets that far).

All in all, it's not a half-bad calculator with which to work. If Hewlett-Packard hadn't made anything since the HP-35, I might even say it's my favorite. (Of course, if they hadn't made the -6S, -30S and -49G, I might not be looking elsewhere . . . )

The Basic is a little quirky, and doesn't seem to have error trapping (I'm still pretty new to it), but I have been able to implement a simultaneously-visible "five-level stack" (X, Y, Z & T, as well as Last X) interface and RPN behavior for all the numeric functions available (directly and shifted) on the keyboard. It's quick enough, and is actually useful, though incomplete.

It operates as a program that must be started (with one or three keystrokes) after the unit is turned ON, and must be stopped in order to turn the unit OFF. (The calculator's automatic shutdown doesn't seem to work when a program is running.)

All keystrokes but one may be intercepted and put to use by such a program. I suppose menus would be possible too, but I'm not going to get carried away.

As of now, I can enter positive & negative real numbers (but not yet exponents), perform stack operations (ENTER, SWAP, ROLL UP, ROLL DOWN, DROP & LAST X), and calculate arithmetic, logarithmic and trigonometric functions. (This in something like 3K of the ~24K RAM available.)

No exponents (just yet), imaginary numbers, programming, or anything else, except outside of the RPN interface, via the calculator's native features. There's no storage arithmetic or base conversions either way! I'm next going to experiment with an idea for a hex/decimal mode that displays both integer and hexadecimal representations of the stack entries side-by-side, and supports input in either format.

So, it seems to be a good testbed for UI questions, despite the already-decided-upon keyboard legends. If nothing else, going to a foreign environment and implementing a simple RPN interface has given a fair appreciation of certain complexities, such as stack drop and Last X updating.

The Silver Edition is TRULY ugly, but I chose it for its expanded Flash ROM. Apparently, a regular TI-83+ gives you otherwise equivalent functionality. The manual is actually very complete, and the language can be picked up real quickly.

After a bit more UI experimentation, I'll be moving into Z-80 Assembly programming, and will update this with my impressions when appropriate.


I forogot to mention, take a look at the Calculator Robot -- Now THAT's cool!



Thank you, Paul; this is a "Must See" in many senses.

Although "we" are an "HP-guys club", this is almost a warning for the days to come...


Forum Jump: