Looking for the name on a concept AKA Do "reverse data loggers" exist?

I don't know that there is a specific name for this type of device other
than a generic "data playback," but it's generally used in such things
as animatronics, machine training, and QC testing. You might try
companies that specialize in these applications.

If you want to roll your own, a basic microcontroller would be all
that's necessary. You might find software already written for the
microcontroller of your choice, or you could write your own. You
basically need to run a clock -- all microcontrollers have at least one
-- and compare the clock times with the recorded changes in EEPROM, RAM,
ROM, or Flash.

Most low-cost microcontrollers do not have analog outputs, but making
one isn't difficult. Probably the easiest is to use a separate DAC chip.
There are several that are multi-channel -- 11 or 12 lines of your
microcontroller will provide (as an example) 8 bits (256 steps) for the
voltage, and 3 or 4 bits for addressing which analog output you want to
apply the voltage to. There are 10, 12, and 16 bit DACs available as
well.

If you create your own solution I'd say the parts would be in the $20-50
range, depending on the number of output lines and other features.

-- Gordon

Aren't there serial DACs. So you could serial out to the DAC instead
of using a bunch of I/O lines in parallel.
I have an ADC that has a serial out.

Joe McKibben

Sure, all kinds. But MCUs with a several 8-bit wide ports are common and
inexpensive these days, so that shouldn't be the deciding factor. It's a
bit more coding and overhead to bit-bang out or use SPI or I2C or
whatever, but it's a good approach as well. I'd probably use serial
myself because I don't like wires!

-- Gordon

I Googled on "animatronics" and ran across this site:
http://www.animatronics.org/
Its worth a look for the Black Magic Woman animatronic video if nothing else
!
The other links might provide some info to give you some other ideas ...

Good luck !
JCD

It's pretty good. The lips are the interesting thing.

Of course, Carlon Santana doesn't sing the song, and he's right-handed.
<g> The following video does the group, and Santana justice:



-- Gordon

Yeah - I agree - the lips are in perfect sync.
And I doubt any animatronic whatsitz will EVER be able to tease a guitar
like Carlos Santana!
Dang! Now I want to go to a concert!

It's the lips that really make this animatronic, yet the site is
virtually devoid of any discussion how the designer/programmer achieved
the sync and lip shapes. Might be a trade secret. Oh, well.

I guess the fact that it is one of the most famous guitar riffs in
modern rock, coupled with a very rudimentary animation with the skeletal
hands, that made me look for more. I might have had a different reaction
if the animatron was strumming a ukulele and singing kumbayyah!

-- Gordon

That would be something involving a D/A (Digital to Analog) converter(s).
Your data logger would have used an A/D (Analog to Digital) converter, along
with some other electronics to store the data as a file.


Does the robot have a specific purpose ? Or are you simply trying to get
started in the hobby by building one based on concepts  & devices that you
are familiar with ? I ask this because the statement "any change is command
steps will be rare" sounds like you have a very specific purpose in mind.

Without knowing more, I would *guess* that you need to look into something
like a small programmable microcontroller that offers one or more D/A
converters onboard, such as the Basic Stamp from Parallax, the Oopic, PIC
AVR, etc. Prices range from around $50 on up depending on what you need,
what gear you already have, etc.

Tell us more and I'm sure we can provide more specific suggestions for you.
JCD

I am working on an open source project

 www.reprap.org

I have been primarily working as a chemistry specialist, but had the
idea of replacing the computer with a calibrated command set stored in
a small amount of memory. I estimated that would allow you to put $100
or so into each machine and then run them off of a single computer.
Since the current build rates are very slow there really isnt much
need to change the command set more often than once a day, or even
less if you dedicate specific machines to the construction of a small
range of parts. I was also thinking that would simplify
troubleshooting of the system since you could standarize a set of
"shake-down" tests and graph signal vs response.

Ghostwriter

Hmmm ... This puts a different spin on it - at least for me - than what I
was originally thinking you needed. But the project is certainly cool!

Not the answer to your questions, I realize, but my first concern from
browsing the reprap catalog site at http://reprap.org/bin/view/Main/HDPE  is
how well does the HDPE plastic stick to itself during the deposition of the
various layers ? I know for a fact that a couple of people that frequent
this group can make some very learned & intelligent comments on that
subject, for sure.

Getting back to your original question, if the build cycle is really fairly
slow, and depending on how fast the plastic (or other extrusion material)
comes out - you may be able to get away with just serially commanding each
motor/actuator from the main computer. Another approach would be to use the
main computer to download an entire program (small, that is) complete with
coordinates stored as code data, and then trigger each motor controller at
the appropriate time from the main computer (using the main computer as an
executive) .

Hope that makes sense.
jcd

That really depends on how hot you can get the material before it hits
the air and also on hot heavy of a bead of material you are
extruding.  The problem is that the hotter/larger beads tend to sag so
you have balance the interlayer adhesion versus part deformation.
Currently some of the machines have produced watertight vessels using
HDPE. Its traditional for the first "real" test of an extruder be the
creation of a shotglass to celebrate with.


At least some of the work is revolving around adapting the technology
for communities that have very limited access to computers and
computer power. A lot of my fellow repraps are seriously interested in
the one laptop per child product, so I had been wondering how many
robots could be controlled by a single computer. Thats where the idea
of a small amount of memory installed as an intergrated part of the
robot struck me as potentially useful.

Ghostwriter

I'm not quite sure how a digital-to-analog conversion would be useful
here, or maybe I'm missing the application. I think you'd want to avoid
any AD/DA conversion because it is inherently "lossy." Fabrication
machines are really CNC devices, and should be controlled and monitored
digitally. Controllers can be quite simple. You can operate most any CNC
machine with a low-cost microcontroller, with the G-code (or other
programming) steps stored in a few K of Flash. However, most people use
a PC because the interaction with it is easier.

What, specifically, were you intending to control with the analog
voltages that have been stored?

-- Gordon

A lot of it is simply trying to take my understanding of electronic
control circuit and expand it to understand what the simplest method
of control for something like this would be. My thinking is that when
we attempt to simplify the hardware, dead recogning rather than
complex feedback loops would be simplest.   That may be impossible to
"dial in" to sufficent tolerance to make it worth the effort.

Even if its not high enough tolerance to be acceptable to Americans,
Africa is potentially a completely different story.

Ghostwriter

You lost me.

In any case, for these types of systems tolerance is really more than
exactness of sizing or measure. It's the difference between working and
not working. Also, digital is usually cheaper, oddly enough.

-- Gordon