Can we make'em like Bill and David? An open RPN?



#25

I had a grandeouse idea the other night. Theres at least a thousand people who would buy a calculator built like they used to. Maybe the answer is just that, like they used to.. Amongst the people here we have programmers, electrical engineers and Im sure we have some senior engineers who long ago became project managers.

Given the large catalogues of pre-produced IC's etc, is it not possible we could put together a "built like a battleship" piece of hardware, that coders like myself, and Paul etc could build on.

Start with a well thought out design which is extendable, and dont try everything all at once. First release build something with 34C feature list. It might only have trig functions etc, but each one lovelingy slaved over in 128bit precision (i learnt these algorithms in scientific computing 15 years ago, theyre in books).

Then, just grow it over time. There is still a company that bought the rights to IBM's old clickety clack keyboards, surely their are other companies that can ma,e nice buttons. There may even be an off-the-shelf generic peice of hardware thats used for remote data collection that fits the bill amd can be hijacked.

I know the wp34s software is good, but stickers on a OK keyboard dont do it justice. The hard bit is the nice package + keys, then the coders cam take over. The nut processor was reerse engineered and remade by one guy, who thought that would be possible?

Im not sure its as mad as it sounds, its how HP started, two guys in a shed, expecting only low volume.

Daniel.

Given the crazy prices on TAS, and the dema


#26

Wow! That's a novelty here...
;-)

openrpn


Edited: 18 Sept 2011, 10:17 a.m.


#27

Yeah, too bad the project appears to be dead (project home page gives a 404)...


#28

Quote:
too bad the project appears to be dead

Yeah, old news...

I'll leave the explanation of failure to someone more involved than me in this project.

Hint: we now now have 34S because only two (later on three) people were developing it.

Massimo


#29

Well, OpenRPN is dead for long. Great ideas, nice texts, promising drawings, little action hardware-wise, poor coordination. IIRC there were no more than some five regular project team members, but IMHO it eventually died due to suboptimal project and team management. Though the originator of OpenRPN was able to write English correctly, at least. Good luck!

"Those who don't learn from history are condemmned to experience it once more" (unknown author).


#30

Quote:
Those who cannot remember the past are condemned to repeat it

George Santayana

#31

What Santayana actually said is

Quote:
Those who cannot remember the past are condemned to fullfill it.

It's slight difference, though, last few digits/words are important to us :-)

e.g.

in RADian mode,
35s : Tan(1e-4) = 1.00000000330e-4
42S : Tan(1e-4) = 1.00000000333e-4
50g : Tan(1e-4) = 1.00000000333e-4

It's a trigonometric bug not related to the range deduction.

Lyuka

Edited: 18 Sept 2011, 5:38 p.m.


#32

There are actually multiple versions of that aphorism floating around, as I discovered today preparing the above post. The quote comes from "The Life of Reason," Volume I,
chapter VII. I'm not scholar enough to know if Santayana wrote in his native Spanish or not, but English translations often give "Those who cannot remember the past are condemned to repeat it." (See, for example, The Santayana Edition's FAQ.)

#33

Guilty as charged, OpenRPN was my crazy idea and there were many reasons why it never panned out. The physical manufacturing challenges are becoming much more manageable... In fact, I even have ready access to a couple of CNC machines. Team size never reached the critical mass needed to reach fruition. I also simply didn't have enough free time to manage anywhere near my potential.

In my mind, WP-34s has fulfilled many of the ideals (in a different embodiment) of the OpenRPN project. Both Walter and Pauli were instrumental to that project, and I'm not at all surprised that when opportunity presented itself they accomplished what they have.

I do have the original site fully backed up, and will talk to one of my cohorts to see if we can at least put it back up for reference.

If I ever choose to resurrect OpenRPN, it would be more hardware centered. Every time I found someone to help with electronic design, it fizzled out quickly for one reason of another. I order to avoid production and tooling costs, imagine the potential held by giving a machine such as the 30b a new set of guts. Even a modern 12C with rebuilt electronics could be a lot of fun.

One more thing: bringing back a project like OpenRPN can't come through a simple announcement. Something tangible must accompany the announcement. I think a working circuit board/development kit would be about right.

I'm at a good point in my life again to consider such an endeavor. But it would, in my eyes be more of an extension of WP-34s. There are ways existing products can be improved, and we have reasonable hardware to work with as a starting point.

This should be more than sufficient to start some flaming, but if anyone is interested in doing something crazy, let me know.


#34

OpenRPN will get a mention in my WP 34S history presentation at HHC. It was a pity not to have had access to the original designs when I prepared it but Walter has helped me out with one of the proposed keyboards. In fact, WP 34S has inherited some code from OpenRPN.

I would love to see and help with another project in line with WP 34S.


#35

Quote:
In fact, WP 34S has inherited some code from OpenRPN.

More code than the presentation says :-)


- Pauli

#36

Quote:
If I ever choose to resurrect OpenRPN, it would be more hardware centered. Every time I found someone to help with electronic design, it fizzled out quickly for one reason of another. I order to avoid production and tooling costs, imagine the potential held by giving a machine such as the 30b a new set of guts. Even a modern 12C with rebuilt electronics could be a lot of fun.

Sorry, I don't see how changing the electronics in the existing already open products would be of much benefit?
Don't they already have decent amounts of processing power and memory for their given user interface?

(I'm not up on the 30B repurposing project, so I'm only guessing they have plenty of resources, but it's likely I'm wrong...)

Quote:
One more thing: bringing back a project like OpenRPN can't come through a simple announcement. Something tangible must accompany the announcement. I think a working circuit board/development kit would be about right.

Sorry, but I have to disagree strongly on that one, a prototype board would not accomplish or prove anything. A PCB with a processor, a bunch of switches, and a display hacked up to work is bordering on trivial. It is in fact a common practical project in microcontroller classes.

I really think people want/need to see attempts at actual housings and keyswitch designs.

I know that's the "hard part", but that's kinda the point...

Dave.


#37

Quote:
I'm not up on the 30B repurposing project, so I'm only guessing they have plenty of resources, but it's likely I'm wrong...)

Yes and no.

  • The CPU is fast. Very fast. Pretty much any space/time trade off is made in favour of space.
  • There is a lot of flash (128kb). We've filled this several times but that is due to cramming extra functions in.
  • There isn't a lot of persistent RAM (2kb). The is 100% full. One bit left currently.
  • There isn't a lot of volatile RAM (4kb) but this seems adequate. We've only run out a few times and each time it could be coded around.
  • The keyboard is adequate.
  • The display is very marginal. The CPU LCD driver can only handle 400 segments. This means the dot matrix portion is tiny.

The most pressing limitation is the display. The next is the shortage of non-volatile RAM.


- Pauli

#38

I recognize the Hugh of this proposal.

#39

Is English your first language?


#40

That's not bad English—it’s a bad pun!

#41

You could start a project at www.kickstarter.com - but for that you already need to formalize your project a bit and put a price-tag there.

I guess the "platform" needs to be quite open, as key-to-function mapping might change, so perhaps using exchangeable keys. Or allow to place plastic labels inside crystal-clear keys?

I also would like to see a calculator with an e-ink or other hires display (aka retina-display on the iPhone), it does not need to be large, but those low-res LCD dot-matrix displays as found on the HPs are quite outdated technology. E.g. one could then place three lines (smaller-font) or two lines (larger-font) text on them, etc.

I would buy such an calculator instantly :-)

#42

I really like this idea.

Someone mentioned E-ink displays. One can order, from E-Ink, development samples here: http://store.nexternal.com/eink/8-x-7-p19.aspx

8x7 segmented display:

Nice thing is, power is only used when refreshing the E-Ink.

If this really gets going I'm willing to donate for development.

Edited: 18 Sept 2011, 1:28 p.m.

#43

Quote:
I had a grandeouse idea the other night.

Welcome to the club.

Quote:
Theres at least a thousand people who would buy a calculator built like they used to.

Which is a woefully insufficient number to merely break even
given the engineering cost to "make'em like Bill and David".
Don't you think if modern HP could do so given their engineering
and manufacturing resources and economy of scale they might be
tempted to take a stab?

Quote:
Given the large catalogues of pre-produced IC's etc, is it not possible we could put together a "built like a battleship" piece of hardware, that coders like myself, and Paul etc could build on.

You perhaps are missing the point.
The issues aren't programming nor finding an SoC platform able
to support it and this application. Rather the big rocks are
in estimate of decreasing cost/complexity:

1. Locating or manufacturing a suitable display
2. Enclosure manufacturing
3. Key cap manufacturing
4. Creation of any legend plates (secondary key legends, etc..)

The above issues have substantial up-front tooling/engineering
costs and don't lend themselves to an "eventual progress with
available resources" development model as do firmware
and electrical/SoC/other_mechanical tasks.

If you've found some other way to recover the above up-front
expenses which don't rely on volume sales, you certainly have
my ear.


#44

1. Off-the-shelf LCDs are in the $5-$10 range even in singles. Custom LCDs, when we looked into them in the late 1980's, were a couple thousand dollars to tool up for.

2. We just this year looked into a custom enclosure requiring molds to be made, with a lot of complex curves-- not a simple design at all. The quote was $50,000.

3. Along with #1 above, we looked into a set of double-shot keys similar to the HP-71's. NRE was a few thousand-- I don't remember exactly, but it was under $5,000.

4. We made a lexan panel with dead-front annunciators in 1994 and the tool-up was $2,000.

All together, it's well under $100,000, and the custom case takes most of it. For low volumes, if you can be satisfied with an off-the-shelf case from a company like PacTec and get them custom machined, the per-piece cost will be much higher but there will be almost no up-front cost.

I've designed and brought many electronic products to market myself (but not calculators). See my post at message #10 at http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv015.cgi?read=79592 to get an idea of how cheaply it can be done. If the calc is developed by a bunch of enthusiasts who don't expect to get paid, it gets even less expensive. This is not that difficult! It seemed pretty clear to me that the reason OpenRPN failed was because of all the nay-saying and discouragement that came from those who insisted that it couldn't be done and kept expecting prototypes to come out as quickly as some of the consumer-market products have to like cell phones and iPods which last only a short time and are out of date before they're introduced if you don't do a blitzkrieg development schedule. This is different. We're hanging onto favorites that are decades old. What's the problem if it takes two or three years for an OpenRPN calc to go from concept to first hardware? If someone takes up the project again, let's give them our encouragement (and help if we have something to offer). Those taking up the project will probably be working on it in the evenings and when they don't have to mow the lawn. It's not their livelihood, so we can't complain.


#45

Quote:
1. Off-the-shelf LCDs are in the $5-$10 range even in singles.

You can certainly find lcd modules in surplus for that
price -- mostly COB text modules. But they aren't typically
suitable for this application due to footprint bulk,
excessive current consumption, incompatible voltage rail,
and being of less than preferable transreflective design.
Text only modules IMHO are somewhat of an unenthusiastic
compromise as, well you're limited to that rigid format.

The last point may be workable as the lower
polarizer could be retrofitted. Even the current consumption
could be managed if due to a brute force resistive voltage divider
by removing the chain and substituting a high
impedance equivalent feeding ultra-low current voltage
followers. But the form factor bulk of typical modules
tends to be the impasse.

In any case if substantial resources are expended
on an enclosure design, the display supply needs to be
reliable.

Quote:
Custom LCDs, when we looked into them in the late 1980's, were a couple thousand dollars to tool up for.

I wouldn't assume too much from that data point given the
intervening time and changes in technology. Also
price varies with topology, form factor, controller,
etc.. so that figure by itself doesn't offer much guidance.

I sent a quote request out last week for custom/semi-custom
32x160 and 32x192 STN reflective formats in COG, COF, and bare glass. Unfortunately
a bundled glass + controller unit may be
a practical non-negotiable
here as contemporary controllers able to support these
densities are not really feasible to provide in reflowable
packages.

Quote:
2. We just this year looked into a custom enclosure requiring molds to be made, with a lot of complex curves-- not a simple design at all. The quote was $50,000.

3. Along with #1 above, we looked into a set of double-shot keys similar to the HP-71's. NRE was a few thousand-- I don't remember exactly, but it was under $5,000.


Was the 55K up-front simply NRE or was some production volume included in this cost?

Quote:
4. We made a lexan panel with dead-front annunciators in 1994 and the tool-up was $2,000.

I investigated UV resistant polyester recently and IIRC the tooling
cost was around US$500. Lexan/polycarbonate besides being more
expensive was actually claimed to be less durable in this service.
But I do like the durable under-printed matte-upper surface design
of these panels for legend use.

Quote:
All together, it's well under $100,000, and the custom case takes most of it. For low volumes, if you can be satisfied with an off-the-shelf case from a company like PacTec and get them custom machined, the per-piece cost will be much higher but there will be almost no up-front cost.

I wasn't even considering use of such utility enclosures.
For an initial prototype nearly anything will do. But I
doubt a pocket calculator sporting pactec aesthetics would
attract much of a following. I'd even speculate a professionally
3D printed enclosure would do better in this respect.

CNC milling may also be feasible for limited production
and opens up the possibility of using a cast aluminum
blank.

Quote:
If someone takes up the project again, let's give them our encouragement (and help if we have something to offer).

To be clear, I'm certainly not discouraging the effort.
Far from it -- I'd encourage it and would be interested
to contribute assuming the direction appeals to me.

But if we're looking at a design which can compare functionally
and aesthetically with say, a legacy HP product,
there will be substantial NRE involved. Leveraging funding and
patience of interested parties needs to happen in the
context of some type of business model and have a realistic
execution strategy and schedule.

I'm not saying it can't be done, but rather substantial
engineering and logistical challenges exist on the
path to getting there. No one would be happier than I for
this assumption to be proven wrong.

[edited for clarity]


Edited: 19 Sept 2011, 8:54 a.m. after one or more responses were posted


#46

Quote:

You can certainly find lcd modules in surplus for that price -- mostly COB text modules. But they aren't typically suitable for this application due to footprint bulk, excessive current consumption, incompatible voltage rail, and being of less than preferable transreflective design. Text only modules IMHO are somewhat of an unenthusiastic compromise as, well you're limited to that rigid format.


Try the Sharp Memory LCD's:

http://www.sharpmemorylcd.com/index.html

Quote:
I'm not saying it can't be done, but rather substantial
engineering and logistical challenges exist on the
path to getting there.

They do, but if you are clever about the design and physical implementation, the hurdles can be quite small indeed.

My uWatch Mk2 calc watch project I'm working on looks to be do-able for under $1000 in development costs, complete with fully custom case. And for that price you get half a dozen of them as a bonus.

Dave.


#47

Quote:
Try the Sharp Memory LCD's:
http://www.sharpmemorylcd.com/index.html

None of their standard product offerings appear to be of
high aspect ratio normally used for a compact calculator display.
Also at horizontal graphic resolutions of interest the active
area sizes run about half of that optimum for this application.
So at this point it is likely a custom glass proposition.
Although given the pricing for standard product thus far it
certainly is promising.

Note AFAICT there appears to be a mix of reflective and
transreflective product in that overview.

#48

Feel free to contact me if you want to talk about such madness, because I've been down this road. If you are interested in resurrecting OpenRPN, I'm not opposed to it in any way.

The last crazy thought I had on the topic was to build a case that would enclose an iPhone and use physical keys with snapdomes that could interact with the touchscreen. This is one of this less crazy concepts I have. Rapid prototyping is a much more viable option than 6 or 7 years ago, and I have access to CNC equipment.

Shoot me an email and let's talk.


Possibly Related Threads...
Thread Author Replies Views Last Post
  [PRIME] RPN: another attempt at returning more than one value to the RPN stack Marcus von Cube, Germany 5 357 11-05-2013, 02:44 AM
Last Post: Marcus von Cube, Germany
  How to open 82162A Thermal Printer Chassis? Ron Fredericks 1 183 07-17-2013, 03:58 AM
Last Post: Tony Duell
  any open source HP 10BII emulators? John 15 658 06-12-2013, 09:58 AM
Last Post: Kimberly Thompson
  Still a new colour calc ! When you'll make it HP ? Mic 10 450 01-13-2013, 09:17 AM
Last Post: Marcus von Cube, Germany
  Bill Smithville NJ please contact me Don Shepherd 1 114 09-23-2012, 03:36 PM
Last Post: Bill (Smithville, NJ)
  Make PIL-Box ILPer program run on OS X Juergen Keller 6 313 09-07-2012, 04:38 PM
Last Post: Hans Holzach
  Programming cable for hp-30b -> wp-34s Open letter to Gene Nigel Rowe 37 1,267 08-02-2012, 12:30 AM
Last Post: Guy Dufour
  Never make a dime of profit... Kees Bouw 19 643 01-04-2012, 05:29 PM
Last Post: bill platt
  HP48 open/repair guides back online Han 2 176 11-04-2011, 11:15 PM
Last Post: hpnut
  An "Open" letter to "DIY" and "Repurposed" calculator projects Ethan Conner 40 1,259 10-21-2011, 04:41 PM
Last Post: Ethan Conner

Forum Jump: