Threaded Mode | Linear Mode

Massimo Gnerucci · 08-25-2015, 06:26 PM

(08-25-2015 02:06 AM)John Smitherman Wrote: The 43s should follow Thomas Okken's lead with Free 42 - run the software on hardware that others provide.

1-

**rprosperi** · 08-25-2015, 07:58 PM

(08-25-2015 02:06 AM)John Smitherman Wrote: The 43s should follow Thomas Okken's lead with Free 42 - run the software on hardware that others provide.

This is OK in addition to hardware intended for the original 43S concept, but certainly not instead of.

Many of us will hang in there...

jch · 08-26-2015, 05:35 PM

(08-25-2015 02:06 AM)John Smitherman Wrote: The 43s should follow Thomas Okken's lead with Free 42 - run the software on hardware that others provide.

1+

But since Version 1.50 Free 42 is using Intel Decimal Floating-Point Library.
And this library is a big one. Too big for the foreseen MCU for 43s ?

Jean-Christophe

Bill (Smithville NJ) · 08-26-2015, 06:48 PM

(08-25-2015 07:58 PM)rprosperi Wrote:
(08-25-2015 02:06 AM)John Smitherman Wrote: The 43s should follow Thomas Okken's lead with Free 42 - run the software on hardware that others provide.

This is OK in addition to hardware intended for the original 43S concept, but certainly not instead of.

Many of us will hang in there...

Hi Bob,

Well, it's been a long long long time of "Hanging in there". The 43S was mentioned back in a 2005 post and the feature set and keyboard layout has been hashed to death.

I will repeat what I have previously posted:

From Sept 2014:

Quote:Well, I've been waiting since Nov 2005. I'll believe in it, if and when I ever see an emulator for it. Forget waiting for the hardware. Just give me an emulator for it and I'll be happy.

Meanwhile, I'll use what I actually have and not worry about what might be someday.

From Dec 2014:

Quote:I have been following this project as well as others for some time.

It seems to me that the main reason most of these projects never materialize is that they are hung up on how to produce the proper platform - this being a piece of hardware with keys and screens.

I always wonder why the other approach was not taken. Start with an existing platform (namely the cell phone), create an emulator of the desired product, debug it on the emulator and move it to a "proper" hardware platform when it becomes feasible to do so in the future. It would have the benefit of having created something now as opposed to only dreaming about it.

I realize that this is an outrageous idea for many people.

But many hardware/software projects in the past have been bootstrapped in this way. I have read where some of the minicomputers were developed/debugged with software emulators long before there was any hardware for the software. That way the software was ready to implement when the hardware did become available.

In my view, a working cell phone version that I can use today will always beat out a vaporware, pie in the sky version that I might be able to use in xxx years, if ever.

Just one man's opinion.

From January 2015:

Quote:For goodness sake, just pick a keyboard layout and move on with the design.

And one other item - when will there be an emulator so that we can start playing around with the actual code that will execute whenever a key is pressed on this "perfect" and "optimum" keyboard? To me executable code is the most important part. An "optimum" keyboard without any code is pretty useless.

So the question is: Does the calculator software ONLY get written after the hardware is settled, or has the software already been written and is waiting on the hardware?

Sounds like the "Which came first: the Chicken or the Egg?"

I do NOT feel that having an emulated version of the 43S in any way distracts from a future hardware version. In fact, it might enhance the finished hardware project since any problem areas can be worked out before placing it in hardware.

And, for the few of us that don't mind emulations, we can be enjoying the feature set and providing feed back to the developers.

Bill
Smithville, NJ

Didier Lachieze · 08-26-2015, 07:19 PM

(08-26-2015 06:48 PM)Bill (Smithville NJ) Wrote: So the question is: Does the calculator software ONLY get written after the hardware is settled, or has the software already been written and is waiting on the hardware?

I think the answer was given above in the post #305: software for such project is written when the developers can devote a big chunk of their own free time to it.
Given this, I'm already extremely grateful to them for the 34S :-)

Marcus von Cube · 08-26-2015, 07:20 PM

(08-26-2015 06:48 PM)Bill (Smithville NJ) Wrote: So the question is: Does the calculator software ONLY get written after the hardware is settled, or has the software already been written and is waiting on the hardware?

No software development on my side except for some tinkering with the fonts Walter has designed. When I joined the WP 34S team, the hardware was readily available and Pauli had written a good part of the code base. There was enough left to do for me: It took me about a year full time.

What is the present situation? The hardware is still under (slow?) development and it is neither available to Pauli nor to me. There is no code base to start with except what we have from the 34S project. Since the new device will provide more resources (screen, memory) this code base isn't really suited to start with. Walter has written most of the manual but I need to read all this before I even can decide on the next steps to take.

Furthermore, I'm much busier in real life then I used to be back in 2011.

**rprosperi** · 08-27-2015, 12:01 AM

(08-26-2015 06:48 PM)Bill (Smithville NJ) Wrote: Well, it's been a long long long time of "Hanging in there"...

Bill - Speaking only for myself - sure I'd like to have 43S software running today, who here wouldn't? But it seems to me the soul of a new platform like this is intimately linked to the hardware it will live on, at least through it's birth. Probably a result of all previous platforms evolving this way, rather than any technical need.

Think about every great calculator you use - whether or not they are now running on the hardware they began on, they very likely all did begin on a specific hardware design. I've seen a few calculators that were conceived to live on multiple h/w platforms (IOS, Android), and some are somewhat usable, but none of them have the power or appeal of products like the 34S, V41 or Free42.

I'm not one of the designers, but I can understand that they may well feel this way too, and as a result, until an affordable h/w platform with sufficient power and flexibility to power the 43S concept is available, the rest of is will have to wait.

So, given I have no real choice anyhow, I will wait patiently and hope that the Glacial Pace of Scaled Reptiles will indeed be revealed in Nashville.

Massimo Gnerucci · 08-27-2015, 05:55 AM

(08-27-2015 12:01 AM)rprosperi Wrote: So, given I have no real choice anyhow, I will wait patiently and hope that the Glacial Pace of Scaled Reptiles will indeed be revealed in Nashville.

Amen!

Thomas Radtke · 08-27-2015, 11:22 AM

Claudios newRPL seems to be in need for new hardware.

Claudio L. · 09-02-2015, 09:42 PM

(08-27-2015 11:22 AM)Thomas Radtke Wrote: Claudios newRPL seems to be in need for new hardware.

Not really. It's working great on the 50g hardware, the only reason I'm not releasing another demo to the public is because I still can't turn the thing OFF reliably (actually, it turns OFF just fine, it's turning it back ON what's not working and I'm still trying to figure out why).
Even if they discontinue the 50g hardware at the end of the year, it will be on the shelves for a few more months, and I think the only people who care about the newRPL project already have 50g's anyway, and they will last for quite a while.

Thomas Radtke · 09-03-2015, 05:18 AM

(09-02-2015 09:42 PM)Claudio L. Wrote: [...] I think the only people who care about the newRPL project already have 50g's anyway [...]

Not true. Thus far I've been a silent lurker but follow your advances and hope, RPL survives this way. I have no 50g but a 48G.

I think a powerfull hardware needs a powerfull OS. Feature creep is a serious problem with simple UIs, and RPL has proven to be an appropriate platform for high-end calculators.

Guenter Schink · 09-03-2015, 09:20 PM

(09-02-2015 09:42 PM)Claudio L. Wrote:
(08-27-2015 11:22 AM)Thomas Radtke Wrote: Claudios newRPL seems to be in need for new hardware.

Not really. It's working great on the 50g hardware, the only reason I'm not releasing another demo to the public is because I still can't turn the thing OFF reliably (actually, it turns OFF just fine, it's turning it back ON what's not working and I'm still trying to figure out why).
Even if they discontinue the 50g hardware at the end of the year, it will be on the shelves for a few more months, and I think the only people who care about the newRPL project already have 50g's anyway, and they will last for quite a while.

Hi Claudio,

I acquired another HP50G some month ago for the only reason to operate on the New RPL. Since then it's sitting on the shelve waiting to be activated.

In order to be distinguishable from my original HP50G this one is blue. It's hungry :-)

If I could have something to load on this device, I could at least show it around at the HHC as a teaser. Wouldn't that be nice?

Günter

jebem · 09-18-2015, 09:07 AM

I apologize in advance to the forum members for bumping this up once again.

But swissmicros have done it.
Real case and real keys.
No FCC restrains apparently, or at least they managed to deal with it offering a product at a price point that, despite high, can be acceptable for those seeking a product like a 15C.

I believe the so called "43S" project's owners could manage to do the same if they wanted to, and offering it a similar price.

sa-penguin · 09-18-2015, 11:24 AM

(09-18-2015 09:07 AM)jebem Wrote: I apologize in advance to the forum members for bumping this up once again.

But swissmicros have done it.
Real case and real keys.

I believe the so called "43S" project's owners could manage to do the same if they wanted to...

Indeed, when it comes to physical desires, a decent keypad is a major selling point.The other big points being display, and batteries [type and lifespan].

If nothing else, I'd like to hear how they make those keys and case. Would people prefer larger buttons (with a smaller gap between them), if it meant the key font was also larger?

How easy is it to re-program a swiss micro with an experimental build of (for example) 43s? Or even 41 ? You have piqued my interest in these models...

Massimo Gnerucci · 09-18-2015, 02:24 PM

(09-18-2015 11:24 AM)sa-penguin Wrote: How easy is it to re-program a swiss micro with an experimental build of (for example) 43s? Or even 41 ? You have piqued my interest in these models...

Michael said 41 will be the next model.
I hope they will make it in portrait mode.

Dwight Sturrock · 09-18-2015, 08:46 PM

On the other hand, a 9810A style desktop 43S could be much easier to build from a hardware standpoint. Better batteries, processors and displays exist in abundance compared to the 80s & 90s. It seems that miniature keyboard for a pocketable calculator is the sticking point.

brouhaha · 09-29-2015, 07:08 PM

(08-26-2015 06:48 PM)Bill (Smithville NJ) Wrote: Well, it's been a long long long time of "Hanging in there". The 43S was mentioned back in a 2005 post and the feature set and keyboard layout has been hashed to death.

There may have been *something* mentioned with a proposed 43S designation in 2005, but AFAIK there wasn't any serious work on it until I demonstrated potentially suitable prototype hardware in Sept. 2011. For the last four years Richard Ottosen and I have been working on that. I'll describe some history of the project, then an update on the status.

Richard and I have been working since 1995 on calculator hardware intended to support open-source firmware. Progress has been very slow because this has been a spare-time project, and both of us have a lot of other demands on our time.

We've demonstrated a succession of prototypes starting around 2004, named DIYn. (There were actually two prototypes from the 1990s, which I've retconned "DIY0".) Our original design target was to provide the functionality of the HP-41CX, either by reimplementing the feature set, or by running the actual HP-41CX firmware. Our earliest hardware didn't have enough memory or a fast enough CPU to do that, so they ran firmware from Woodstock and Spice series calculators instead. We demonstrated these at HHC conferences.

DIY1, DIY2, 2003-2004:
https://www.flickr.com/photos/22368471@N...8841933989

DIY3, 3D-printed case, 2009:
https://www.flickr.com/photos/22368471@N...9238179642

From 2003-2010, microcontrollers with enough memory and fast enough CPU to achieve our objectives would drain lithium coin cells quickly, and we did NOT want to build calculators that required larger batteries or frequent recharging of batteries. In 2010, a low-power ARM microcontroller became available, and we designed new prototypes which could run the HP-41CX firmware:

DIY4 hardware with character LCD, running HP-41CX firmware, July 2011:
https://www.flickr.com/photos/22368471@N...8726860797

DIY4 with kludge to evaluate graphic LCD, running HP-41CX firmware hacked for full stack display, August 2011:
https://www.flickr.com/photos/22368471@N...8726406755

DIY4X (DIY4 using adapter board for graphic LCD), September 2011:
https://www.flickr.com/photos/22368471@N...8413263533

The ARM microcontroller chosen for the DIY4 had 128KB of flash and 32KB of RAM, which still wasn't as much as we needed for a 41CX fully configurable with user-selectable modules in all ports. A new version of the microcontroller became available that had 1MB of flash and 128KB of RAM. We were able to install that in the existing DIY4X. The new chip also added USB support, which we were able to use by redesigning the display adapter board.

DIY4X using new display adapter board that adds USB capability, January 2012:
https://www.flickr.com/photos/22368471@N...8726780827
https://www.flickr.com/photos/22368471@N...8704955497
https://www.flickr.com/photos/22368471@N...8726621113

I ported Thomas Okken's Free42 software to run on the DIY4X hardware. The DIY4 keyboard layout, having been designed for the HP-41CX, was not very suitable for Free42, so the keys had to be rearranged somewhat randomly, making it difficult to use. I modified the Free42 software to have a seven-line display with dedicated soft menu legends, rather than the two lines of the original, one of which was often used for soft menus, leaving only one line for stack or program line display.

https://www.flickr.com/photos/22368471@N...8851996198

Around this time frame, I asked the WP-34s firmware developers whether they might be interested in this hardware platform. Walter started working on a WP-43s design based on our hardware specifications.

At an HHC conference, I conducted market research by asking attendees whether they'd be more interested in a new calculator that was functionally an improved 41CX, including the four-line stack display, or a 42S. The show of hands was overwhelmingly for the 42S. I repeated this poll at the latest HHC conference, and the result was the same.

We redesigned as the DIY5, using a single PCB, eliminating the need for the display adapter, and changed the keyboard layout to be better suited to the 42S, with one additional row of keys so that we could have permanently dedicated soft menu keys, rather than menu keys shared with common scientific functions. The packaging was changed to make the calculator roughly two thirds the thickness of the DIY4X:

https://www.flickr.com/photos/22368471@N...8853138738

The DIY5 PCB had a number of minor design errors, and we decided to make some minor improvements to it including better support for IR printing. The revised hardware is called XC-40. We tried to have a near-final rev of the XC-40 hardware built in time for the recent HHC 2015 conference, and to have sent out prototypes to the WP-34s developers before the conference, but unfortunately we didn't quite pull that off. We now have the components on hand to build ten units, but due to unforeseen circumstances the PCB layout isn't quite ready.

We expect to have the new PCB layout completed soon, build one unit for hardware validation, and if all goes well build more units for firmware development. We expect to make units available to the WP-34s developers before year end.

The development of WP-43s firmware is not going to happen overnight. In fact, it might take quite a while. The firmware developers have said that they don't have as much time available as they did when developing WP-34s.

In the mean time, I'm working on porting WP-34s to the XC-40 hardware.

We hope to get the XC-40 hardware through EMC testing (FCC part 15), and to offer XC-42 firmware development prototypes (XC-40 hardware with Free42 firmware) for purchase by enthusiasts in the spring. The prototypes will have the folded polyester film case shown in the DIY5 photographs, with a photo paper keyboard overlay over the tact switches. It will be possible to flash other firmware (via USB or MicroSD card) and replace the keyboard overlay, to run WP-34s, or (someday) WP-43s.

If I'm able to get suitable CAD designs whipped into shape, the STL files will be made available so that those wanting a 3D-printed case can print their own case parts or order them from one of the online 3D printing services. The 3D case design will be somewhat similar to that seen in the DIY3 photos, but wedge shaped (thinner at the keyboard end, thicker at the display end) to use a pair of alkaline AA cells for longer battery life, and to allow at least a little bit of usage of the MicroSD card under low battery conditions. (Even when they are fresh, CR2032 lithium coin cells can barely power the MicroSD card. When they are nearly at end of life, the current drawn by the MicroSD card when active could cause a calculator reset and memory loss.)

Because the LCD display is attached by double-sided adhesive foam to the XC-40 PCB when installed in the folded polyester film case, and is damaged if the foam is removed, it might be necessary to purchase a replacement display when refitting the calculator into a 3D-printed case. The LCD display costs approximately $34 each in single quantity from Digi-Key or Mouser.

Kinma · 09-30-2015, 01:19 PM

Eric,

Let me congratulate you on the steps taken.
It seems we are headed for a very powerful calculator!

I will certainly buy one!

Kinma · 10-01-2015, 02:49 PM

And of course; also my congratulations to Richard!

I know it is a big task you two are taking upon you.
And this is just the hardware.

Quote:The ARM microcontroller chosen for the DIY4 had 128KB of flash and 32KB of RAM, which still wasn't as much as we needed for a 41CX fully configurable with user-selectable modules in all ports.

Question; how much flash and ram do you need for the 41CX and how much do you need for the free42 running on the Gecko microcontroller?

brouhaha · 10-01-2015, 07:40 PM

(10-01-2015 02:49 PM)Kinma Wrote: Question; how much flash and ram do you need for the 41CX

RAM:
Native firmware: 7KB
For simulated RAM registers (status, main registers, extended memory), 8KB
Stack: probably under 100 bytes

Flash:
Bootloader w/ LCD, keyboard, USB, MicroSD support, FAT filesystem, rudimentary diagnostics: 45KB out of 64KB reserved
Native firmware: 75KB
Base 41CX ROM: 48KB (six ROM pages, 8KB per page)
Port ROMs: typically 24KB per port to support HP 12K word ROMs (32KB for HEPAX), plus 8KB per page for hard-addressed ROMs

Note: ROM images could be packed to 5KB per page, with minor performance hit

Since flash memory has limited endurance, in order to allow user to change ROM selection, it may be advantageous to put some port ROMs in RAM instead, but RAM size tends to be the limiting factor.

Quote: and how much do you need for the free42 running on the Gecko microcontroller?

RAM:
Native firmware and Free42: 16KB
Stack: unknown
User memory: as much as possible

Note: Free42 is unlikely to work well with less than 32KB of user RAM. Giant Gecko provides 128KB total RAM.

Flash:
Bootloader w/ LCD, keyboard, USB, MicroSD support, FAT filesystem, rudimentary diagnostics: 45KB out of 64KB reserved
Frmware, native plus Free42 (has MicroSD, FAT fs, but no USB support): 257KB

Since this code is under active development, the flash requirements could grow a fair bit before release, expecially if the USB stack is added.

I have not attempted to optimize RAM usage any more than was necessary to get the firmware to be generally functional.

For both 41CX and Free42, the bootloader when running uses 14KB of RAM. In an ideal world, that would be RAM reserved for the bootloader only, so that it would be possible to run the bootloader without losing the user memory. Due to the general scarcity of RAM, the memory is not presently dedicated in that manner. With some effort, a portion of the bootloader RAM, including the exception vector table and LCD buffer, could be shared with the calculator RAM.

I have not attempted to optimize RAM usage any more than was necessary to get the firmware to be generally functional.

Free42 was in general not written for use in low-memory environments. The original 42S had 8KB of RAM, but Free42 will not work well with that little. Free42 depends heavily on dynamic memory allocation from the standard C library, even for things like floating point numbers in the stack. As a result, it is quite possible for Free42 to fragment the limited memory of a microcontroller. For this reason, I consider Free42 on the XC-42 to be experimental. To avoid fragmentation problems, it is advisable to periodically write the full state of the calculator to an SD card, and reload it. Unfortunately when running from CR2032 coin cells, use of the SD card under low battery conditions could cause the calculator to lose all memory, because SD cards can draw 100mA, which is well beyond the spec of even a fresh CR2032. This is one reason why the evolution of the case design is moving to a wedge shape with a pair of alkaline AA cells instead of the CR2032 coin cells.