HP-25C Eumulator

[brouhaha]

(11-14-2014 02:01 PM)Chris Chung Wrote:  I would really like to do a HP-41

The -41 was originally the primary objective of the DIY series of calculators Richard Ottosen and I developed. Our first four models (DIY0 through DIY3) used PIC microcontrollers, with increasing amounts of memory. Like what you've done, we started with the Woodstock and Spice series. Our conclusion was that even the bigger, faster PIC18 part in the DIY3 wasn't fast enough to make a usable -41 without having to code most of the simulation code in tightly written and highly optimized assember, and even then the speed wouldn't have been great.

Quote: and again the LCD is the biggest problem.

Same with ours. We used a 16x2 character LCD module made by Lunex, and it only has 8 programmable characters, so it's not possible to replicate any possible use of the 41 display. The inter-character punctuation is also a problem.

It's hard to find ANY off-the-shelf display or display module that's the right width for a handheld calculator. Real commercially made calculators all use custom displays, and there's apparently very little market for a calculator-sized LCD as a standard product. It's easy to find LCDs that are either much smaller or much larger than desired for a calculator.

Quote:Using a graphic module to "pretend" the 14 segments doesn't look right to me, but may be the only solution (except to shell out $1-2k for custom LCDs).

I was perfectly willing to use a graphic display with a mostly "normal" bitmap font, rather than trying to replicate the -41 segmented display appearance.

We did find a bitmap graphic LCD module with the same outline dimensions as the Lumex 16x2 character module, but the contract is awful unless a backlight was used.

On the DIY4, we switched from PIC to Energy Micro Gecko, and later Giant Gecko, which use an ARM Cortex-M3 core. These provide more than enough CPU power for a -41, and that's the first thing I got running on them. The DIY4 originally used the Lumex 16x2 character display. We hacked a 2.7-inch diagonal graphic LCD onto it in place of the character module using an adapter board, which I refer to as the DIY4X and that was vastly superior, but quite expensive. The choice was made beacuse the 2.7" diagonal display was much taller than we wanted, but the width was exactly what we wanted. The price of that graphic LCD has come down somewhat since then. We redesigned the hardare to support the graphic display natively, as DIY5, and I've showed it at conferences in two versions, one running -41 simulation (hacked to optionally show all four stack levels), and one natively running Thomas Okken's Free42 (hacked for an 8-line text display). I polled the conference audience, and there was overwhelmingly more interest in running Free42 than the -41 simulation.

Richard and I are still planning to sell calculator hardware similar to the DIY5. It will be able to support several different open-source calculator programs, including Free42, WP-34S, and (eventually) WP-43S. The biggest challenge is that few people are willing to pay what we'll realistically have to sell it for, which is perhaps a bit less than $300. That's about the minimum we can sell it for and not lose our shirts just on the manufacturing cost; the pricing isn't an attempt to recoup any of the costs we've paid out of our own pockets for development, which I estimate to be over $10K so far, and probably at least another $20K to go, nor will it compensate us for any of our time spent, which even at minimum wage amounts to even more than the out-of-pocket expenses.

I certainly don't want to discourage anyone from working on such products, but as I've said before, making calculators is a great way to make a small fortune, only if you're starting from a large fortune. It's easy to build prototypes, and extremely difficult to build and sell a commercial-grade product that people will actually buy.