Robotic Programming for 35s?

[David Hayden]

So I have this cute little 35s with lots of memory and no I/O. It occurred to me that something like a Lego Mindstorm could be programmed to press the keys on the calculator. Add something to parse a program listing and you have a robot that will enter your code.

Another possibility might be an old X-Y plotter, modified with a soft tip instead of a pen, and with a holder for the calculator.

Has anyone done this?

Dave

[Katie Wasserman]

We discuss this here every once in a long while, here's one thread (start reading about half-way down). David Smith mentions that he made one once, but I don't remember anyone saying that they have or know of such a machine recently -- creating one would be a fun project!

-Katie

Edited: 13 Nov 2010, 11:15 a.m.

[Don Shepherd]

Katie, trying to do something like this for program entry on the 30b would probably be impossible because of the "shift + hold" functions. You would, I suppose, need two actuators to depress two keys at the same time.

[uhmgawa]

Quote:

---

Katie, trying to do something like this for program entry on the 30b would probably be impossible because of the "shift + hold" functions. You would, I suppose, need two actuators to depress two keys at the same time.

---

Good point. Or at bare minimum a dedicated solenoid for keys
involved in simultaneous-down sequences. This even impacts the
use of an analog multiplexer as it will be necessary to select
more than one column/row in such scenarios.

Anyway the input mechanism is fairly straightforward. Getting
output from the multiplexed voltage levels of an lcd will be
much more interesting.

[Bill (Smithville, NJ)]

And another link where I had pondered doing this for a HP-42s.

Robot Programming HP-42S??

I never got very far with it, but still interested if anyone can pull it off.

Bill

[Thomas Chrapkiewicz]

Hello David:

[You and I must have some type of Extra-Sensory Connection.
(Good hearing from you - again - enjoyed your talk in Sept.)]

I have both Lego Mindstorms systems and a number of xy plotters 'laying around' and I've been considering exactly this for a number of I/O deficient calculators (35s, 42s, 15c, etc).

Here are the current issues as I see it:
[in no particular order)

1. As the pen plotters would have to move the calculator (in the y direction), the mass of the calculator may be a problem. However, as I have velocity and acceleration control with the 7550 plotter, this may not be an issue.

2. The linear 'throw' of the PenDown for the plotter may not be sufficient. Force could also be an issue - but the actuator for the calculator will require a bit more thought.

3. Mechanical modifications of the plotter to accept the thickness of the calculator(s).

4. I/O limitations of the Mindstorms (I use the older RCX as opposed to the newer NXT systems) to sense actual position.

This is not (yet) at the 'top of my list', but it is a most interesting prospect.

TomC

[David Hayden]

Quote:

---

Hello David:

[You and I must have some type of Extra-Sensory Connection.
(Good hearing from you - again - enjoyed your talk in Sept.)]

---

Thanks Tom. I enjoyed meeting you too (and everyone else at HHC2010). I even have an idea for a presentation at HHC2011 :)

Dave

[uhmgawa]

Quote:

---

Another possibility might be an old X-Y plotter, modified with a soft tip instead of a pen, and with a holder for the calculator.

Has anyone done this?

---

I am envious of your free time. :)

Yet for as entertaining as a mechanical/robotic version may be
I'd probably resort to a pair of analog multiplexers, one of each
sitting on the row and column scan matrix, commons tied together.
Yea grafting this into the calc isn't great. But it is more
reliable and would eliminate otherwise undetectable contact
bounce.

Unsure of the key matrix topology of the 35s but I'd hazard it
isn't over 8 lines wide for either row or column. Never bothered
to poke at it as the 35s is pretty much a black box without
any reasonable means to probe its firmware outside of a nitric
acid bath. AFAICT HP isn't sharing any useful details either.

While not the same effect as an x/y articulating probe, I'd
consider a solenoid matrix if you're intent on a mechanical
approach. The actuation force of the the 35s keys is ~150gm
which should be workable with cast off surplus solenoids.
To deal with interfacing the likely large solenoid body size to
a relative tighter key pitch, flexible plastic rods could be used
aimed at the key locations via a perforated alignment plate.
Setup of the solenoid drive in a matrix similar to the key scan
topology will minimize circuitry.

In any case please share any pictures with the class!

[David Hayden]

Quote:

---

I am envious of your free time. :)

---

For what it's worth, there may be real money to be made here. For example, this software for sale is $100 for the software alone and $300 for the software preloaded on a 35s. Subtract $56 for the cost of the calculator and they're charging $144 to key in the program.

[Thomas Chrapkiewicz]

...with a bit more thought:

I would like to use somewhat commonly available hardware-i.e a plotter or Lego Mindstorms.

1. My pen plotter idea is out (for now) as moving the calculator around,AND transferring the mechanical movement of the pen dropping is very difficult (for me). The plotters that are commonly available (HP7550, 7475) are not true table top plotters, but move the paper within pinch rollers (in the y direction).

2. The Lego RCX does have three motor outputs (X, Y and Pen) and three inputs - certainly sufficient for this task.

I will pursue #2 for now - as time allows!!!

Of course, we will still have the mechanical problems:
1. Aligning each particular machine (35s, 42s, 15c etc) will require unique 'tooling' to hold the calculator properly - both in X / Y and vertical distance.

2. Key pressure - I believe that the appropriate type of damping material along with limiting the vertical 'drop' of the pen will take care of this.

3. Certainly feedback will be an issue, as the robot will not have vision to ensure that the correct key was hit.

4. Of course, multiple/simultaneous keystrokes will not be possible - although perhaps the user could work in synchrony with the contolling software for the (potentially) few keysteps that would require this.

TomC