WP43S, may I contribute?

[Georg Bisseling]

Dear brouhaha, mvcube, paulidale, wbonin,

This is my first post, so I may introduce myself.
I am a software developer living in Germany and programmable calculators like the TI-57 and the HP-11C led me to my current profession.

This summer I started to play around with the Arduino platform and after blinking LEDs, using a super sonic sound transceiver to measure distances, controlling displays and even connecting to the ethernet I came up with the idea to do a small 4-level RPN calculator inspired by the HP-11C.

After some research I used the very same decNumber library that is used in the WP34S project and shortly after that I wanted to have trigonometric functions etc. and I came along the WP34S project: Seeing that kind of frustrated me! Such a cute case, such a beautiful keyboard compared to my self-soldered monster...

I got over my frustration and completed a decent 4 level stack calculator that displays the whole stack in a 16x4 display running on an Arduino Mega 2560 using your routines from decn.c/h.
Still not content with my keyboard though, to say the least.

Now I see that the diminishing availability of HP 20b and HP 30b has driven you to build your own WP43S hardware. When are you ready to share your considerations and decisions? Wouldn't it be nice to be Arduino compatible? One could develop on a cheap Arduino Due clone (84MHz ARM 96KB SRAM, 512KB Flash at 15€!) and later on switch to the integrated calculator hardware.
Making the calculator hardware software compatible to the Arduino platform could raise the demand and therefor lower the price considerably.

Can I contribute in any way?

Greetings from Bonn
Georg Bisseling

[brouhaha]

(11-30-2014 12:28 PM)Georg Bisseling Wrote:  [...] cheap Arduino Due clone (84MHz ARM 96KB SRAM, 512KB Flash at 15€!) and later on switch to the integrated calculator hardware.
Making the calculator hardware software compatible to the Arduino platform could raise the demand and therefor lower the price considerably.

We're paying significantly less than 15€ for the part we're using (EFM32GG380F1024), which has 1MB of flash and 128KB of RAM. It only runs at 48 MHz max, and the calculator will probably default to running at 14 MHz. It draws far less power than most other ARM microcontrollers even at comparable clock frequencies.

As far as I know, there's no reason the Arduino Due code couldn't be ported to the EFM32 parts, but I haven't heard of anyone doing it. It's an interesting idea, but I don't really think that if I did it, it would sell more calculators.

The big per-unit expenses are the LCD, PCB fab, assembly & test, and regulatory compliance (only amortized over a relatively small number of units).

Quote:Can I contribute in any way?

The majority of the calculator firmware will be entirely independent of the hardware. There's just a small library to interface to the keyboard, display, and IR LED. I think suitable calculator firmware could be developed on anything (e.g., a desktop or laptop), and as long as it is designed for the I/O to be replaceable, it won't be hard to port. It didn't take me long at all to get Free42 running on it.

[Georg Bisseling]

brouhaha,

Thanks for your kind answer. It seems that the hardware design for the WP43S is already quite stable and I am pretty sure that your choice is very well suited for a battery driven device. Frankly, I expect your project to deliver a result that can easily compete with commercial offers or even outperform them.

Maybe the idea of an arduino based calculator platform is viable elsewhere, as soon as the problem of assigning different functions to keys can be solved with something else than stickers. Using a touchscreen is the most flexible and cheap solution, but then you could just write an app for some smartphone, couldn't you?

I am really curious to see what you come up with!

Regards
Georg

P.S.: 15 EUR is the price for the whole populated PCB including VAT, Shipping and an USB-Cable afair.

[brouhaha]

(12-01-2014 08:09 PM)Georg Bisseling Wrote:  P.S.: 15 EUR is the price for the whole populated PCB including VAT, Shipping and an USB-Cable afair.

Except that it would still need another PCB for the keyboard etc., so I think the cost still ends up higher than our putting the EFM32 on our own PCB.

I think using an Arduino is an interesting idea, but not really where we're going with our hardware. However, the whole purpose of developing our hardware is to support open-source firmware, so the firmware could possibly be portable to or from an Arduino-based platform.

[BarryMead]

One consideration that no one has brought up in this thread, is that perhaps by selecting a pre-built high-frequency micro-controller assembly such as this one http://www.silabs.com/products/mcu/lowpo...k3700.aspx , or this one http://www.ti.com/ww/en/launchpad/launch....html#tabs might be able to get around the FCC or other international RF certification requirements. If the kit included only the preassembled Keyboard/Display/Case assemblies, the Keyboard/Display assemblies may not require FCC / Internnational RF certification as they are not operating at high frequencies. This might eliminate the problem of US ONLY markets and make the calculator available to a worldwide sales base again.

You could sell ONLY the "Keyboard/Display/Case" assembly, and ask users to supply their own pre-certified micro-controller and downloaded firmware, to complete the kit. This would shift the cost of Certification off to those who make
thousands of boards not hundreds.

One would have to investigate the "Battery Life" of such an assembly. If a Pre-Certified microcontroller assembly exists that satisfies the computational speed, I/O, sleep/wake, Sleep Persistent calculator state SRAM, and Battery Life requirements, it would be wise to consider using it instead of designing a custom unit if only to shift the certification costs to someone who can better amortize the costs over thousands of units.

Any thoughts?

[brouhaha]

(12-10-2014 05:36 PM)BarryMead Wrote:  and ask users to supply their own pre-certified micro-controller

Note that those modules aren't pre-certified, they are uncertified. However, as you say, if the "calculator" vendor didn't supply that module, at least under US FCC regulations I don't think the vendor would be responsible for certification of the final calculator assembled by the end user.

The main drawbacks of that approach are that the physical packaging will be larger and more cumbersome, and the total end-user cost will be significantly higher.

[BarryMead]

(12-11-2014 09:39 PM)brouhaha Wrote:  The main drawbacks of that approach are that the physical packaging will be larger and more cumbersome, and the total end-user cost will be significantly higher.

I completely agree with your assessment of how it would adversely affect the completed calculator.
If the certification laws in other countries work like they do in the U.S. would it be worth the extra
international support base to use such a module. Then there is always the risk that the manufacture
of the selected module could "STOP MAKING THEM" at an inopportune time.