FPGA Based Open Calculator Platform

[M0R33z]

Hello all,

I am currently following Christophe's project (DB48X) regarding his RPL implementation on the DM42. I think the project is absolutely great and I look forward to each new version. Thanks Christophe.

Christophe and all those who develop things for HP calculators, especially the development of emulators are doing a great job. Also what Michael from SwissMicros is doing is absolutely fantastic.

Here's what I've been thinking for years and I've been discussing this with Christophe for the last few days. What if there is an open hardware calculator platform ? It should be similar in design to the HP48GX.

OCP - Open Calculator Platform

I imagine the following specifications:

- Metal case
- High-end keyboard (HP/SwissMicros)
- Low Power FPGA (Cores for HP 48 Saturn CPU, HP41, ...)
- IR / RS 232
- BT / WIFI
- MicroSD
- Flash Memory (a lot)
- Colour Disply (like HP Prime)
- sound
- and so on

I can imagine that the financing can be realised via crowdfunding (indiegogo, Kickstarter). And maybe produced by SwissMicros or even by HP itself.

With this hardware platform, we would then be able to implement any calculator (even TI :-) ), assuming the community supports the development of CORES.

Licences for ROM, here there might be support from HP for little money. But that is another topic for now.

Is this idea realistic?

Maurice

[EdS2]

I think it would be necessary to outline what the advantages are going to be, of an FPGA story as opposed to emulation on a microcontroller. There might be advantages, but I think it's a tradeoff. I'd suggest that if any calculator is well-enough understood to be implemented on FPGA, then it's well-enough understood to be emulated in C.

If you're thinking of doing this for fun, of course, then the advantages are not the important thing - it's that you want to do it and expect it to be fun. That's always fine!

[M0R33z]

An FPGA has many advantages compared to software emulation. One of the most important is of course that you are not dependent on the operating system of the respective platform.

The guys from the following projects have also thought of something. There are emulators for all the projects.

https://mega65.org/
https://www.specnext.com/
https://mister-devel.github.io/MkDocs_MiSTer/

In the FPGA it is possible to model a large part of the logic. Yes, this can definitely also be done in C, but I see much more flexibility here and the community is given the opportunity to develop more than just pocket calculators on a low-cost hardware platform. You are no longer dependent on the operating system like Apple or Microsoft or even Linux. And you can put it in your pocket ;-) I rather see the fun here.

[blackjetrock]

i wouldn't rule out a microcontroller either. The development tools tend to be more open and with a powerful enough processor you can emulate a lot of different things.

Here's a Casio CQ-1 replacement PCB:

https://youtu.be/XzfspKhKC18

and an FX502/602 PCB:
https://youtu.be/5w6tmykqrYM

Both of those use the STM32F103 family of microcontrollers. I now use the RP2040, which is considerably more powerful. It can emulate the 6303 in the Psion organiser:

https://youtu.be/v2YWStDrQIs

The emulation runs at about the same speed or maybe a little faster when emulating the LZ ROM.

Andrew

[David Hayden]

I think we should move away from emulating or re-implementing the HP processors. The problem is that they're just too limited by today's standards. The Saturn has a 20-bit address space addressing ~1 million nibbles, or 512Kbytes. You just can't do much in that space and the HP calculators themselves jump through hoops to get around the limitation.

I think we're better off writing calculator software with modern languages for modern processors, like free-42, DM32, and newRPL. These are all smoking fast.

As for a hardware platform, we seem to be nearly there with the swiss micros platform. Face plate overlays are already possible, so the only real impediment I see is the keys themselves. I don't think If there were a way to make reasonably priced keys then we could use that platform for whatever calculator we wanted. Of course it would have to work with the number and layout of the existing keys, but that seems like a small price to pay.

Just my 2 cents worth.

Dave

[Jeff_Birt]

(10-24-2023 02:10 PM)David Hayden Wrote:  I think we should move away from emulating or re-implementing the HP processors. The problem is that they're just too limited by today's standards.

The advantage of emulating the existing HW is that you don't have to write all new firmware w.r.t. the calculator functionality. If you start from scratch, you have to also add every bit of functionality that would have been in an original calculators ROM.

I think DB48X is the sort of project you are thinking of. As I recall he found it was not possible to get the functionality desired working with the SM emulator platform so the code base is new and will run on multiple platforms (including SM).

[John Keith]

(10-24-2023 02:10 PM)David Hayden Wrote:  I think we should move away from emulating or re-implementing the HP processors. The problem is that they're just too limited by today's standards. The Saturn has a 20-bit address space addressing ~1 million nibbles, or 512Kbytes. You just can't do much in that space and the HP calculators themselves jump through hoops to get around the limitation.

I think we're better off writing calculator software with modern languages for modern processors, like free-42, DM32, and newRPL. These are all smoking fast.

I strongly agree, with one caveat. I believe that any "modern" HP-48 or 50 should also include a true emulator such as EMU48. This would allow the running of the existing RPL code base, including libraries and programs written in System RPL or assembly. Also, I don't think that the current Swiss Micros platforms are adequate for a modern RPL calculator. I would consider the equivalent of the prime G2 to be a minimum level of acceptable CPU/Ram/Flash, otherwise it would be a toy, suitable for collectors but not for serious users.

[Craig Bladow]

(10-22-2023 05:05 PM)M0R33z Wrote:  An FPGA has many advantages compared to software emulation. One of the most important is of course that you are not dependent on the operating system of the respective platform.

The guys from the following projects have also thought of something. There are emulators for all the projects.

https://mega65.org/
https://www.specnext.com/
https://mister-devel.github.io/MkDocs_MiSTer/

In the FPGA it is possible to model a large part of the logic. Yes, this can definitely also be done in C, but I see much more flexibility here and the community is given the opportunity to develop more than just pocket calculators on a low-cost hardware platform. You are no longer dependent on the operating system like Apple or Microsoft or even Linux. And you can put it in your pocket ;-) I rather see the fun here.

I completely agree. The point is to have fun and learn something in the process. I have a Mega65 and the FPGA design has opened up many possibilities as well as better Commodore 64 emulation than can be done on a PC.

Cheers,
Craig

[EdS2]

(10-24-2023 11:13 PM)Craig Bladow Wrote:  

(10-22-2023 05:05 PM)M0R33z Wrote:  An FPGA has many advantages...
I rather see the fun here.

I completely agree. The point is to have fun and learn something in the process.

I'd agree too, if this were a personal project. But the head post says

Quote:I can imagine that the financing can be realised via crowdfunding (indiegogo, Kickstarter). And maybe produced by SwissMicros or even by HP itself.
...
Is this idea realistic?

so it's in that scenario that I'm making my negative comments.

[MeindertKuipers]

My personal opinion, based on my FPGA and microcontroller experience, is that with an FPGA you still depend heavily on the operating system to develop and specific tools that are typically vendor bound and closed source.
Most microcontroller tools are supported on many OS'es and allow others to develop and contribute as well. In addition, when planning to use SD cards, BT or Wifi, you will have to use some level of CPU support. Microcontroller libraries are often open source.
I was planning myself to start with an FPGA project to interface with the HP41, but when discovering the features of the RP2040 microcontroller I decided that that would be a much more flexible and open path towards something real and working.