Since the topic has been broached: Is anyone interested in giving this concept another shot? I know it will sell when finished, so this question is directed more towards technical development people. As I see it, even if the only goal was to use existing plastic parts we can give software such as WP-34s a better environment to operate within. At least one complete machine would be fantastic, but in the short term it would be best to get a development board in production (all I need are PC boards and components because I have a good friend who owns a commercial reflow oven.)

I'm not looking for a flame war, I'm sincerely looking for expressions of interest. This would be a full reboot, making use of lessons learned and taking the original concept forward. We don't have to start from scratch, and I feel much better prepared this time around personally.

Who's up for it?

I'd likely be interested again.

- Pauli

We are three. But I'm a software guy. :-)

Out of curiosity as a designer of calculator hardware myself, what were the important lessons learned from all the Open RPN effort you mentioned?

Do you really only just need a PCB to move this project forward?

If this was the case then that would presumably mean that you have a housing and key solution in place?

Because as I see it, and I've probably mentioned this before, a physical calculator design is an exercise in housing, keypad, legend, and ergonomic and manufacturing design - any PCB design is a fairly trivial aspect of it all.

BTW, reflow ovens are usually not needed for prototype PCB assembly.

Dave.

Edited: 19 Sept 2011, 7:06 a.m. after one or more responses were posted

Figure out the hardware and case first.

- Pauli

I'll go one further and say that even with okay to decent hardware, the software is by no means a given. The 20b languished for quite a while with nobody (publicly) repurposing it seriously with new firmware.

the production of a new calculator will be a combined effort of hardware designers, firmware writers, application software guys and production people. I very much doubt anyone here is capable of all of these tasks and in many/most cases folks will be limited to but one.

- Pauli

Quote:

---

I'll go one further and say that even with okay to decent hardware, the software is by no means a given. The 20b languished for quite a while with nobody (publicly) repurposing it seriously with new firmware.

the production of a new calculator will be a combined effort of hardware designers, firmware writers, application software guys and production people. I very much doubt anyone here is capable of all of these tasks and in many/most cases folks will be limited to but one.

---

For sure.

But "if you build it, software will come".

With a handheld calculator, the hardware design is essentially everything. It is the thing that costs real money, commitment, and is hard to change and tweak, it "locks you in" in many ways. It is the heart and soul of the product, if it wasn't, OpenRPN would have been released years ago as a jiffy box with a bunch of tact switches and a 16x2 LCD (and failed as result of such poor hardware implementation of course).

Firmware on the other hand only takes time, and has great flexibility, it can be tweaked until the cows come home. And at a general level, firmware is fairly hardware agnostic.

The reason people like us want to design calculator hardware is because we want a physical product form factor and "look'n'feel" that is not available elsewhere. Hardware is the basis for everything. Yes, it needs software too, but I wouldn't really consider that before you have a solid physical implementation.

For example, I did not write a line of code until I had my uCalc or uWatch projects in a usable form factor, there was simply no point. The same will go for the uWatch Mk2 as well.

The reason anyone would consider a new calc design like OpenRPN, or my uWatch or uCalc, is because the current hardware platforms available are not deemed adequate. Otherwise, we'd all think the holy grail has arrived with the 20B re-purposing project.

The only reason OpenRPN has not gotten off the ground before, is solely because of a lack of actual hardware, I'm sure everyone will agree?

Also, since the last Open RPN project was floated, there has been a boon in crowd source funding via the likes of Kick Starter. So money to get projects like this off the ground is easier to get these days, which is great, but it does come with real commitments that must be met.

Dave.

Edited: 19 Sept 2011, 7:45 a.m.

yes.

Well, I have to agree that the problem in creating a new calculator is very complex. See the problems a giant like HP and it's facilities at China are facing to.
The software, as I could see with the great WP-34S is not the biggest problem.

I just can't understand is why people that are hooked to RPN logic or RPL just not integrate to WP-34S project and try to make a bridge to HP people, in order they incorporate this project.

HP needs to be an open organization - at least in the calculators division - and pay more attemption to what their brilliant customers are doing with HP calculators.

HP 30B is an excellent starting point hardware, as I can see: it has two line display, seems to have a decent keyboard and is reprogrammable. Just HP forgoten to make add a simple connector.

Instead of trying to make each key from scratch and so the case, membranes, steel domes, case, battery compartment, face plate and all other parts I'm probably forgoting, it's so much easier to take a 40US$ still ready hardware and improve its software?

Just for bringing more clear the difficulties: remember the keyboard problems of new HP-15C and of the 50G when it was released? How many thousand dollars were and will be invested on the solution of this very simple problem? Do a group of people have the money and time to make such correction?

Doing a real thing, that you can press its keys is far more complicated than having it drawn on CAD software - that is what I believe in...

I have seen other dreams like this one in this forum: HP has just released our long time waiting 15C. The new hardware (and software) has some important bugs to fix, but people, instead off making pressure on HP to fix them, just dream for a release of another calculator model - like 71 or 42. It wouldn't be better fix this one we have on hands first??? Once solved, HP may bring to us the 16C model or 11C model.
My apologies for my poor English ....
Artur - Brazil

how about a little less ambitious starting point - a DESKTOP calculator?

i'm going to suggest (as i have before) an ATmega1284p all-in-one processor, which comes in a DIP package that most anyone can solder. it has 128k of flash, 16k ram, 4k eeprom and 32 I/O lines (including 2 serial ports).

there are several varieties of key switches that one can buy on ebay out of china for less than us$10 for a bag of 100 - some with full keytops, some with just a 1/8" stud to press on.

for an LCD, just use something standard that can be talked to down a 4-bit bus, say 24 columns x 4 lines? for batteries use 4x AA cells.

at a pinch, folks can assemble their own complete claculator on veroboard, while later on we can get some PCB kits made.

THEN, once that phase is completed, look at making it all fit inside a case the size and thickness of a credit card...

http://www.nixiebunny.com/malmbergcalc/calc.html

Edited: 19 Sept 2011, 11:38 a.m.

While I definitely agree that a hardware first approach is essential, I'm not at all sure I agree that "if you build it, software will come".

The 20b platform was available to repurpose for several years before we released the 34S. The firmware only became reality because of a labour of love from two people. Believe me it was a long and arduous process. We've had some very able assistance this year but again it was predominately one extra dedicated person.

Apart from the 34S there really hasn't been a serious attempt to produce firmware for the platform. An emulator of old firmware and a four function hex calculator are the only two I am aware of.

- Pauli

>While I definitely agree that a hardware first approach is essential, I'm not at all sure I agree that "if you build it, software will come".

I guess the thinking is the WP34S team is around and will be around... ;-)

It's harder to find $10k in funding for a HW project than $100k in SW development. That's because people will donate time more readily than money, and with HW you have nothing to show for until it's done, whereas SW can be improved incrementally.

Of course, I agree with you. SW shouldn't be taken for granted. But it seems that there's a base to built upon out there. Not the least thanks to your (WP34S team) donations of time and effort.

It is an interesting avenue. Alternatively, why not we try to lobby the current people at HP to include more keys and a better display in their next business calculator, so it can be repurposed with an evolution of the WP-34s firmware? That would give a repurposing platform that could be used for years to come. The current HP calculator people seem to have an ear to the ground, so it is a possibility at least.

$0.02 CAD

The biggest lesson learned hands down: Build the hardware first. It's fine to get software rolling early on, but people need something physically real to believe in it. That in turn will generate a great deal of confidence that a physical product will be the end result.

When I first started OpenRPN my resources were much more limited, and so were those of the world in general. While still far from trivial, making a few prototype enclosures is much more feasible than it was 5 years ago. Now I actually have equipment available to me for mold making/prototyping.

A little bit more will be required beyond just a PCB, it needs to be a working development kit with a display. Ideally, enough prototypes could be assembled for distribution to a few of the big contributing developers in the project. Showing them off as a finished product would be enough to drive pre-orders to fund production. Or will be helpful to convince a few investors to bankroll a sure thing. As my current connections go, I know someone who is a co-owner of a plastic and rubber molding company.

Yes, I understand that reflow may not be necessary. It will come in handy if any BGAs are used, though.

I will elaborate more on these topics when I have a few more spare minutes.

Simply redo the 42S and get Thomas Okken into the boat.

Alright, I'm feeling crazy. Site backups are being pulled as we speak for reference purposes. I want to keep things as simple as possible this time around with respect to the project page. Is anyone interested in taking on webmaster responsibility? If anyone has space for hosting it would be appreciated, although otherwise I'll figure something else. I think it makes sense to keep subversion on sourceforge either way.

Within the next week I will gather the conceptual designs for review. I think ground up re-evaluation of project goals is worthwhile: for example, is the *fix system still the best approach? Hardware will need to start from scratch, and that will be a good thing.

This will not be a completely blank slate, but it won't be far from it. I firmly believe the technology needed to facilitate such a project, even at low to moderate production levels, now exists and we should pursue it.

Time to have some fun!!!

Quote:

---

While I definitely agree that a hardware first approach is essential, I'm not at all sure I agree that "if you build it, software will come".

---

I agree, there is no guarantee, unless you do it yourself. But it's infinitely more likely getting software people involved in existing hardware, than it is to get hardware people involved in existing software. That's just the way the world works. The entire Open Source Hardware (OSWH) industry is built on this.

It's been proven over and over again, that all you need is the hardware, and some initial software drivers to make firmware development easy, and chances are the project will take off.

My idea with the uCalc and uWatch Mk2 is to build say half a dozen prototypes (it doesn't cost much more to make 6 than it takes to make 1) and give them away to interested software developers who are better at software than I am, hope\ing that they would be keen enough to develop for it if they find the hardware platform compelling enough. The fallback plan would be I simply write it myself.

Of course, I understand how important core routines are that handle things like the display and keyboard, so I would write those to help "bootstrap" 3rd party development. I thin that's vital.

One thing is for sure though, without real pre-production hardware, a calculator project does not exist.

Dave.

Your statements are spot on.

Quote:

---

how about a little less ambitious starting point - a DESKTOP calculator?

---

I'm not quite clear what the benefit would be as target hardware
isn't the limitation and supporting the
development host itself as a "target" in some form is a practical
necessity. If prospective participants would
benefit from having the actual target SoC handy just leverage
one of the many simulators available. For example I'd used simavr
for early development on both the AVR m328p and m1284p KINOMI
modules, but just to validate execution performance of
KEMU on these underpowered SoCs was worth continuing the
effort.

Quote:

---

i'm going to suggest (as i have before) an ATmega1284p all-in-one processor, which comes in a DIP package that most anyone can solder. it has 128k of flash, 16k ram, 4k eeprom and 32 I/O lines (including 2 serial ports).

---

May the force be with you.

But honestly you may want to consider an architecture a bit
higher up the food chain. I've laid out simple eval/break_out
boards for the at91sam7l and lpc1114 (both lqfp package versions)
for which I'm happy to provide the source/gerbers. These days
you can find quite inexpensive board houses to build more than
you're likely to need for US$10-20.

Quote:

---

for an LCD, just use something standard that can be talked to down a 4-bit bus, say 24 columns x 4 lines?

---

If you are intent on rejoining 1974 with a desktop calculator
project, and I can fully appreciate the motivation from a
novelty/retro perspective, go for broke and use one of these
for a display Futaba 20x2 VFD. Doing so at least
you'll have something of lasting conversation value for your
effort.

when i was working for trimble a few years back, we had a (just released) radio product that contained a 20x2 VFD matrix module, sitting right next to a bluetooth module in the front panel. the VFD was the appropriate technology for the application.

i'm keen on the 1284p because it is ACCESSIBLE to the average constructor. the average constructor has at their disposal a 40w soldering iron with a 2mm tip, veroboard, and whatever they can buy from digikey or farnell (or on ebay) in 1-off quantities.

the average constructor does NOT have access to PCB fabrication, SMT placement equipment, a quality soldering iron, or the money to pay for any of the above. they lack even the ability to solder 0804 SMD resistors and capacitors.

i'd like to see a project that the average constructor can build. if you can find an arm7 SoC processor in a DIP package, then i'm all ears...

Quote:

---

i'd like to see a project that the average constructor can build. if you can find an arm7 SoC processor in a DIP package, then i'm all ears...

---

Ah, I see -- my misunderstanding. I didn't realise you were
trying to remove pcb construction entirely from the effort.

Sorry I'm not aware of arm SoCs in DIP packages, however you can
find ready made eval boards which may get you to the same
starting point (or further). NXP had a deal running (perhaps
still do) where they were giving away LPC1114 arm eval boards
to anyone and everyone with a good marketing story concerning
prospective application for the board. I think it would be
worth a look.

Look at the MBED project for this. http://mbed.org

The crux with WP 34S was that Pauli and Walter were busily writing software and documentation which nobody really cared about because it did not make it to the hardware until I chimed in.

The project really took off when people started to play around with it (emulator first then real hardware). You will need something to toy around with to attract more developers, testers, and, last but not least, users and customers.

Thomas has already stated here that his software is not designed with limited hardware resources in mind. But his experience is well worth being asked for in a new project.

This is indeed a great proto platform, add an SPI display and a keyboard from an old HP41 or another machine and you can start working on software

Skill is definitely a factor, but you don't need expensive equipment to solder SMT parts. Take an SOIC for example. A lot of flux is helpful but not imperative, and you can do it with an iron that covers three pins at once. A 1/8" chisel tip will be fine since you won't solder just one pin at a time. To start, solder just the end pins on one side, just enough to hold the IC in place. Bridges are ok for now. Then solder down the other row of pins, using lots of solder. It may be all one huge solder bridge, but that's ok to start. Then you hold the board vertically and slowly run the soldering iron from top to bottom, re-melting all the solder on that side. The excess will stay on the iron, not the IC, and there will be no bridges. Now go back and do the first side which until now was only tacked. Clean off the flux, and it looks like it was done by machine. Perfect. It really is surprisingly easy.

Soldering 1206 chip capacitors and resistors is very easy. 0805 is ok with a tiny soldering iron but still nothing special. 0402 is getting a bit ridiculous.

VFDs are quite power-hungry and won't always be sunlight-visible. They tend to make a lot of electrical noise too. LCDs don't have any of these problems.

the problem with eval boards is that they are invariably a short-run item, so of little use in an open-source project - a year or two down the line and the hardware is no longer available, leaving the project at a dead end.

for an open-source hardware project to be successful it needs to fulfil the following requirements:

1. people need to want the item,

2. people need to be able to construct the item with relative ease and certainty of success,

3. people need to be able to get the necessary bits both now and tomorrow.

i'm not thinking of a project where a handful of developers make a few dozen units, while other folks just look on and make comments on a feedback form saying how cool it all is. i'd like to see something where 5000 or more folks decide to have a go and use the open-source plans to make their own calculator - and learn from their experiences in a positive way. it is all about mass participation, rather than just observation.

how do others feel about a primary goal of openRPN being:

"5000 hobbyists building their own calculator"?

Quote:

---

Thomas has already stated here that his software is not designed with limited hardware resources in mind. But his experience is well worth being asked for in a new project.

---

if all the graphics stuff was ripped out (it wouldn't, after all, be needed), how resource-hungry would free42 actually be? the biggest hog, i'd imagine, would on the scarce onboard RAM. this could be alleviated somewhat by shifting the user program space and registers onto external storage.

I agree completely here.

I'd have got around to the hardware port eventually. So much code to write and debug and only one me. So many changes along the way too.

Still, Marcus did this bit faster and better than I would have :-)

- Pauli

Quote:

---

Look at the MBED project for this. http://mbed.org

---

Yes that's what I was getting at. No hostility to the archaic
dip format intended. I as well find smd to dip (or more generally
100mil pitch) adapter boards occasionally
handy for SoC prototype usage
and usually just lay out whatever I happen to need. The cost
to spin a two layer PCB is amazingly cheap of late and CAD
tool choices abound. Furthermore 99.9% of the real engineering
has been accomplished on die and such SoC boards are little more
than customized low speed interconnect and bypass. Here's a few typical
examples tossed to soak up otherwise unused
flat-rate space with the first round of KINOMI boards.

That's what I meant, Marcus. Take a proven concept and a developer of a proven software.

The 15C LE apparently attracted several thousand customers in few weeks. With some clever marketing, the same can be done with a 42S remake. Places where potential customers lurk are known and easily accessible.

Doing something completely new and hoping for the big success is daydreaming. Do it after you have a selling product.

Quote:

---

Skill is definitely a factor, but you don't need expensive equipment to solder SMT parts. Take an SOIC for example. A lot of flux is helpful but not imperative, and you can do it with an iron that covers three pins at once. A 1/8" chisel tip will be fine since you won't solder just one pin at a time. To start, solder just the end pins on one side, just enough to hold the IC in place. Bridges are ok for now. Then solder down the other row of pins, using lots of solder. It may be all one huge solder bridge, but that's ok to start. Then you hold the board vertically and slowly run the soldering iron from top to bottom, re-melting all the solder on that side. The excess will stay on the iron, not the IC, and there will be no bridges. Now go back and do the first side which until now was only tacked. Clean off the flux, and it looks like it was done by machine. Perfect. It really is surprisingly easy.

Soldering 1206 chip capacitors and resistors is very easy. 0805 is ok with a tiny soldering iron but still nothing special. 0402 is getting a bit ridiculous.

VFDs are quite power-hungry and won't always be sunlight-visible. They tend to make a lot of electrical noise too. LCDs don't have any of these problems.

---

I have an SMD soldering tutorial video:

http://www.youtube.com/watch?v=b9FC9fAlfQE

It is indeed easy.

Dave.

Quote:

---

for an open-source hardware project to be successful it needs to fulfil the following requirements:

1. people need to want the item,

2. people need to be able to construct the item with relative ease and certainty of success,

3. people need to be able to get the necessary bits both now and tomorrow.

---

Whilst I don't necessarily disagree with that, #2 and #3 are not necessary for a project to be a massive success.

Take the worlds most successful OSHW product, the Arduino. Almost everyone buys the ready made version instead of buying a kit version.

I found the same thing with my uWatch project. Hardly anyone wanted a kit, they all wanted the fully assembled version. But of course they really like that it's all open source - not that most will do anything with it, but they just like the feeling that it's open source!

Quote:

---

i'm not thinking of a project where a handful of developers make a few dozen units, while other folks just look on and make comments on a feedback form saying how cool it all is. i'd like to see something where 5000 or more folks decide to have a go and use the open-source plans to make their own calculator - and learn from their experiences in a positive way. it is all about mass participation, rather than just observation.

how do others feel about a primary goal of openRPN being:

"5000 hobbyists building their own calculator"?

---

Whilst that would be nice, and is possible in theory, I don't think it's going to work in practice. The reason is most people just don't find a calculator that interesting.

Successful kits being built by the tens of thousands by the hacker/maker communities to teach kids (and big kids) soldering and basic electronics etc are usually something more exciting than a calculator. Like a TV-B-GONE, a LED POV kit, a simple quadcopter etc.

Dave.

Quote:

---

Whilst that would be nice, and is possible in theory, I don't think it's going to work in practice. The reason is most people just don't find a calculator that interesting.

---

Edited: 22 Sept 2011, 12:43 a.m.

nice

Thanks for the flowers! :-)

Not at all, they are well deserved.

You certainly did the hardware port faster than I would have. You did the power management better than I would have too I suspect.

You've also provided much useful input and implementation since then.

Then there are the 55 bytes of semi-volatile memory you uncovered that I'd completely missed and HP had too :-)

- Pauli

Quote:

---

They would probably find it more interesting if it can act as a controller too and not just a calculator [...] with data in floating-point and programming right on the device itself instead of on a host computer.

---

well the 1284p does have two serial ports, and i'm sure free42 could be easily extended to interact with these to become a useful [instrument] controller too.

harping on more about assembly methods (sorry all!), as engineers we do all tend to overestimate the complexity - and dexterity - of the general public / average hobbyist. personally, i have no trouble with placing the smallest leaded packages, even hand-constructing pretty complex prototype items with 0402 sized components, but i have had many years practice working at that level. i don't even need a fine soldering iron, just a suitable loupe, just the right flux, good solder, and LOTS of heat.

BUT, despite online tutorials being available, and assurances from the folks on here, i still maintain that even given a quality PCB, and a 100-pin QFP SoC, the vast majority of would-be constructors WILL end up with a mess that they will never be able to recover a working finished item from.

and as for buying a ready-built, why on earth would someone? no denying that the uWatch is impressively cool, but whipping it out and showing it to someone is pointless unless you can claim you built it yourself! as a built-item, an HP-11 or HP-15 is far cooler (not intending any offence, Dave. and certainly not wanting to detract from what an impressive accomplishment the uWatch is). saying "hey, look at this neat calculator watch that SOMEONE ELSE built and sold to me" doesn't ring true as a good conversation starter at parties...

Quote:

---

Then there are the 55 bytes of semi-volatile memory you uncovered that I'd completely missed and HP had too :-)

---

HP needs them for blinking the cursor. So they didn't forget them.

Quote:

---

..and as for buying a ready-built, why on earth would someone? no denying that the uWatch is impressively cool, but whipping it out and showing it to someone is pointless unless you can claim you built it yourself!

---

I suppose it is a matter of perspective. In general I find
the accomplishment of getting 48 pins of a 0.5mm pitch smd
package reliably tacked down to a board to be astronomically
dwarfed by the feat of beaming ions at a slab of silicon,
realizing a SoC design into 1e5 transistors, implementing an
economical manufacturing process, and selling it to me
at a profit for US$5.

Not to diminish the accomplishment of rolling your own
"hardware". But as stated a conservative 99.9% of the engineering
has completed before the device ever arrives in a users hands.
And given the extreme simplicity of closed bus SoCs, the
opportunity for design errors is generally limited to power rail
bypass/sequencing and perhaps clock/pll layout.

So I'd offer the actual value add in such a prospective project
begins when the reset vector is fetched. But agreed that
may be too esoteric to carry the same "bragging rights" impact.

Someone has already made a processor, has all the backwards compatability worked out etc, boards.

All is needed is a case, LCD and a keyboard, and your guaranteed a working calculator.

After this milestone, anyone can of course write/port a completely modern software set on it.

Daniel