Handmade calculator.


In the musical instrument would you have too distinct classes of modern instruments: mass produced and handmade. In practice, beginners buy the mass produced models, and as they advance the handmade ones. The price can easy double or triple, but in the hands of the pro, the instrument is worth it because it could have better response, mechanics, durability, features, custom ergonomics, etc...

Should there be an analogue in the calculator industry? Essentially all calculators are mass produced/common denominator tools. Why not a handcrafted piece of art like a calculator. HP didn't get everything right. Why not custom calculators...

Why should a surveyor need the exact same functions as a machinist? Couldn't a platform be designed with an upgradeable ROM and adjustable key layouts? A library of routines can be gathered. It just needs to be put together in a nice package.

...x^5 on a button?!
...arbitrary precision?
...BCD.12 - A0.1F269?

You are probably right, I am crazy...



Actaully, I have been wondering the same exasct thing---and the same analogy (music)!





Raw costs for a cheapie sci calc and more expensive calc aren't that much different. Pricing will track demand, position in marketplace, etc.

A $20 calc prob is purchased by importer for somewhere around $4-$5. Distributors, transshipment, inventory costs, etc. allow just a bit of profit to be eked out at retail.

Typ. sci calc prob has a $0.50 (50 cent) or less microcontroller on it. Depending on volume, this could drop to $0.30 (30 cents). LCD is a buck - $1.50 depending on custom layout etc. Case/keys/elastomer pad costs are prob $0.50 (fify cents). PCB maybe $0.25. Coin batt prob costs a bit too. Add assembly, packaging, paperwork, etc., translations, warranty support costs etc. and you're around the $4-$5 price. After a certain threshold of complexity in a calc, additional complexity doesn't cost much more until extra RAM and/or better LCD required, etc.

There are quite a few folks here on this Forum that can design a decent calc: hardware isn't that difficult, it's one chip an LCD and some kinda keypad, and with the math & firmware skills of various folks here, a working, reliable, accurate calc implementation on any of various microcontroller proliferations is entirely possible - even an emulation of an HP calc.

The real issue for building a calc for "us" is initial startup costs: injection molded case w/shift functions labeled (silkscreening/paint/etc), and keys w/legends.
It prob takes $15K-$50K for dies to be setup for this.

If there were a calc case that had a single wide opening for a key above the 0-9 numeric area, that'd be a nice candidate calc body to start with.

Bill Wiese

San Jose CA


As fun as it is to fiddle together some hardware and make it work, wouldn't it be easier to do this in software on some sort of touchscreen palm-top computer? Not only that, but each person could modify the software to their own needs.


but that's already there---tons of them for Palm and WinCE, The problem is, it is just not the same. THe interface of the touch screen is awful. And the Palm devices are too expensive---and crappy.


There's many touch-screen calc implementations on PalmPilot/WinCE handheld computers.

UI isn't *nearly* as good as a calc w/a good keyboard and legible display.

Calculators have had 30+ years of refinement; Palm computers are relatively new. The first 10 of these years were the glory years where most of the usability issues were worked out (key feel, spacing, large [ENTER] key ;) etc.)

HP prob did LOTS of usability/human factors studies on KB layouts, spacing, tactile feel, etc. in the 1970s. Shame to throw all that good work away.

Bill Wiese

San Jose CA


HP keyboards may be better than touchscreens, but somehow I'm skeptical that your home-made calculators will have keyboards that are near the quality of HP keyboards anyhow.


> The real issue for building a calc for "us" is initial startup costs: injection molded case w/shift functions labeled (silkscreening/paint/etc), and keys w/legends. It prob takes $15K-$50K for dies to be setup for this.

Buy why injection molding? I don't want a home built clone of an HP machine.

How about a sculpted **ceramic** case With tiny buckled spring keys? (like my favorite 15lb IBM keyboard)

Nobody would make handmade calculators for mass market profit, but for sake of the machine and pleasing the customer. In other words turn the calculator into a piece of art.

Here is an example:
Standard Trumpet - $1500

Monette - $15000?

This More realistic at $3000-5000

See the difference? Wynton Marsalis plays the fancy one. And it isn't ALL about the looks, at that level the trumpet also functions better.


That is a really cool concept... perfect for the 2004 Create a Calculator contest<http://www.hp.com/calculators/contest/>

I notice this year they are not talking about the crazy calculators they were last year (measuring the temperature on mars, painted red, white and blue... etc)

In fact, they even encourage the higher level students not to focus on the form; rather on "technical features, performance factors, and scientific and graphing abilities."

Now if only I was allowed to enter.

That actually sounds like a really practical idea. Calcualtors are definitely on their way out, as computers become smaller and more powerful. I could see that be relevant not only to enthusiasts, but also to educators and students. A teacher could customize a class set of calculators to allow students only to be able to use some features, create a keyboard overlay, and lock the mode in. Perhaps they could mass produce keyboard overlays to be used with inkjet printers.(hey it is HP, the largest printer company).

Furthermore, if simple software was developed (for the computer probably) to allow teachers to "drag and drop" functions to keys, while still allowing the enthusiasts to program their own functions and assign them to keys (much like macros), a calcualtor like this would be widely popular. Teachers would eat it up because it would allow students a calculator to do the simple arithmetic, but not allow them to use the complex features of a calculator. Students would like it because they could place their functions exactly where they want them (i.e., not have to go through 18 menus to get to a specific function), or customize their calcualtor for a specific unit (if they are doing a unit on trig, they could have sin, cos, tan, sec, csc, cot and all the inverse functions on the keyboard). Engineers and those in the business world would like it because you could easily place your specific functions where you want them. Essentially, you would have a scientific, grpahing, engineering, and financial calculator all in one. HP could release one or two bodies, and a library of predefined key setups (graphing, scientific, financial, basic 4 function, heck... even PDA).

That is a great idea, now if only it would be more than that.

-Ben Salinas


hi ben,

heres another way to select keyboard layout.


Does this looks like the HP-Expander they didn't release?


Perhaps they could mass produce keyboard overlays to be used with inkjet printers

I use Epson stickers on my 48GX's :-)


Being that my only musical "skills" are playing back and enjoying the
music, the analogy with musical instruments would never have occurred to

Why should a surveyor need the exact same functions as a machinist?
Couldn't a platform be designed with an upgradeable ROM and adjustable
key layouts? A library of routines can be gathered. It just needs to be
put together in a nice package.

...x^5 on a button?! ...arbitrary precision? ...BCD.12 - A0.1F269?

You are probably right, I am crazy...

Not at all. Many of the HP calculators are highly customizable. In
principal, one could start from scratch with an entirely new ROM on the
49 series, which do have upgradeable flash ROMs, but it doesn't seem
practical. If the ROM source code were available, it could be modified.
Several of the HP models have user-assignable keys, and some have
faceplate overlays available. Well, at least they used to be available;
they're hard to find now. Too bad that HP seems to have dropped the
overlays. With programs, you can easily effectively add to the built-in
commands of any programmable model. On the 48 and 49 series, you have
"vectored ENTER" which can pre-process the command line and post-process
the result. On the 48 series (and I guess the 41C series too?), there
are commercial applications with libraries and often keyboard overlays
that can effectively transform the calculator into a special purpose
tool. For both the 48 and 49 series there are many user-written
applications available that add programs, directories, or libraries as
extra tools or even replacements for built-in commands. I don't know of
any commercial add-on applications for the 49 series, but there could
be. I've never tried the 38G, 39G, or 40G, but my understanding is that
they're supposed to take instructor written (or, at least, supplied)
"applets", so that the instructor can transfer the current subject to
the students' calculators.

So if HP would fix the problems with its current models, and bring back
overlays, wouldn't that pretty much fulfill your wishes? A way to cover
a legend on the keytop itself (or replace the button, like on a PC's
keyboard) with a custom one would be nice, but I'm not sure that it
would be feasible, short of supplying a model with blank keys to be
labeled by the user or whoever else does customizing.

Ok, I'm sure that many on this forum would prefer a "true RPN" model
instead of an RPL model. Surely any of the "true RPN" models could be
emulated on an RPL model, concealing as much of the underlying RPL code
as desired from the user. And on the 48 or 49 series, all four stack
registers and even the LASTX register could be displayed; wouldn't that
would be an improvement?



Bring back the blanknut; this calculator was highly customizable too, nice I/O capabilities etc... BUT they stop the production.

Is it possible to copy the HP-41CX, with little add-on like more memory and built-in cool modules in about the same /copycat package, keyboard etc as the original one and sell it for about 150$?

Edited: 5 Mar 2004, 9:00 p.m.


Is it possible to copy the HP-41CX, with little add-on like more memory and built-in cool modules in about the same /copycat package, keyboard etc as the original one and sell it for about 150$?

Yes, if you have about $1M to spend on tooling for the plastics. How many can you sell at a $150 retail price, and how much profit margin will you have on each one? Will you sell through retailers? If so, how do you get them to stock it, and how much of the margin will they absorb?

As others have pointed out, the problems in doing this aren't technical. Designing the calculator is the easy part.


This calculator was "store bought" but i believe it could be fabricated by any reasonably skilled enthusiast.


I don't know if there would be a patent problem.


This sounds like an HP-41 or 71 with keyboard overlays!

HP did this ages ago....

I have a copy of an HP catalog from March 84 offering full customization on the 41 and 71, I can scan it for anyone interested, but it will be a big file.




This thread got me to spend Saturday morning following up on an old idea.

I've got a 41CV keyboard hooked to some of the keys (39 out of 50) on a TI-83+SE.

It works fine, as far as it goes. The key assignments are all mixed up, but I can re-map those in software. I'm going to restore the rudimentary RPN application (four-level stack, etc.) and change its key assignments to make it behave appropriately, then start playing with Z-80 assembler.

Early indications are the TI's PCB may just be made to fit inside a modifed 41 case. The big LCD is dicey -- maybe just a hair too wide, but certainly too tall. However, I've found a non-functioning card reader case, so if I have to extend the 41 to mega-sized, it might work. (I've also toyed with simply grafting the HP-41 front onto the 83's case in place if its keys -- but that's a less desirable option.)

It seems to me a TI-83+ titanium (or whatever) has been hinted at -- it seems to be an 83+SE in a fancier case. If the technology has been simplified, however, that might make both its PCB and its LCD better candidates for insertion into a 41 shell.

Why bother? Well, there's the obvious 41 keyboard quality, with its full-size Enter key. It's set up for overlays (doeas anyone have a spare available?), and has alpha legends on the key fronts for alpha capabilities. The 41 case is deep enough that it should accomodate more modern circuitry and batteries.

Also -- the keys from my disassembled HP-28s seem very close in shape and size to those on the 41, and might serve as double-shot substitutes. I haven't disassembled the 41's keyboard -- can anyone tell me whether the clamshell keys are similar enough that they might be used? (THere's the 19B's keys as well -- a whole host of quality, alternate keytops might be available should one wish to go this route.)

The TI-83+SE, as a graphical calculator, is maybe not the ideal candidate. But it's Flash programmable, it's got plenty of memory, it's got a PC link (USB) and it's got an emulator and SDK available for free.

If there's any interest, I'll add a couple of pictures tonight.


The clamshell keys are incompatible with 41 and Voyager keys. The clamshells are of totally different dimensions, most notably the height. The small (non-numeric) 41 keys are the same physical size as the Voyager keys, except for the ENTER key.

Are you looking for a blank overlay or the standard cream colored with legends? Email me.

Edited: 6 Mar 2004, 7:58 p.m.


It would be interesting to take one of those handsome old HP desktop RPN calculators and rebuild it with updated functionality. It would be a much easier than a do-it-yourself pocket calculator, too.



Yes indeed it would!

I have a dead 97 just waiting for that ;) - and some time.
(Though I really like the HP91 KB layout, but my 91 still works ;)

Bill Wiese

San Jose CA


If you look at the internals of a fullnut HP41, it should be very easy to implement a new motherboard to replace the 20 year old HP design. No fancy connectors (just a series of pads), no ICs with zillions of tiny wires bonded to the keyboard, etc etc. The only problem would be that you would have to use the existing keyboard layout (at least the labels on the top of keys) and the display (more of a problem really as you'd have to live with the single line display and its cranky interface -- it presents itself as an HP-41 peripheral).

The problem is that the existing HP41s are not getting any younger and your "new" calculator may develop age related problems almost immediately (failing keys, inoperative displays, flaky battery connections, etc). Moreover, even broken 41s command high prices, so getting your raw materials will be expensive, and the moment you release your new design, prices will go even higher (advice, better corner the market of used 41s before releasing your product :-).

Apart from that, you can use one of the cheaper microcontrolers to emulate the entire 41CX and most modules in a compact and very low power package.

You don't even need to develop an emulator from scratch, just use one of the existing ones that are available in source form.



Vassilis, nice comment in the current issue of IEEE Spectrum.

As far as replacing the motherboard of a fullnut CPU, that is a project that I am currently working on. However, rather than emulating the nut CPU with a microcontroller, (with the attendant speed and power issues), I am nearly done with the design of a clone of the nut CPU suitable for loading into an FPGA. Once I complete the basic design, I'll add a direct interface to standard Flash and SRAM (basically an internal MLDL) plus a dynamic speed select mechanism (aiming for up to 50x). The details of when and how to switch speed automatically I am still working out. Of course any access of a real port peripheral, as well as the keyboard scanner, will need to run at regular speed.

I still haven't decided whether or not to include the Timer module functionality or not, and am always searching for undocumented features of the nut CPU to make the clone as accurate as possible. The behavior of certain instructions when executed immediately after the Carry flag is set is sufficiently bizarre that I probably will not copy it. Right now my design has no such restrictions.


Forum Jump: