32 BIT FORTH FOR P/C WITH SOUND
> > Here is a description of Phase I of my project so far: (jpegs upon
> > I have used a full size human skeleton replica for the body.
> > The body doesn't move but the head turns and the jaw moves.
> > My FORTH words can create new primative dictionry sounds when in learn
> > (currently working in a polling 8 bit mode)
> > I can concatinate the sounds to create English words.
> > (works but has poor sound quality)
> > Each word has programmed movements
> > for the stepper motors controlling the jaw and neck.
> > While the stepper controls don't require
> > much CPU power or memory, the sound engine does.
> > Windows is too bloated!
> > OS2 is too bloated!
> > Linux is too bloated!
> > I need an O/S that is small and yet can grow as it learns.
> > I need full hardware control of the motherboard and I/O devices.
> > This includes interrupt and DMA processing.
> > I've been programming for 20 years. I used
> > FORTH on my LSI-11 in 1979 .
> > It was stand-alone and gave me full control of the machine.
> > Is the PCs hardware scheme so complex that nobody has yet written a
> > alone FORTH with full control of the machine and the typical i/o cards
> > it uses ?
Yes, I believe that is the case. There is also some question of whether
there would be an adequate market for such a machine. While the prices on
general purpose PC boxes keep coming down, the demand for a
Forth-controlled PC seems pretty limited. As Paul Bennett suggests herein,
there is much to be said for avoiding the general purpose computer approach
entirely in certain applications.
> > Phase I controls the head via a desktop computers, but mobility is the
> > ultimate goal. I'd like to shink down the hardware to fit in the
That sounds like you are heading toward a general purpose robot/android.
Quite an ambitious project, but you seem to have bitten off a fairly small
piece for phase 1.
> If ever there was an application that should be done with multiple small
> processors this is it. Use one processor (with reasonable power) to run
> the speech and separate processors to control each stepper motor.
Yes, that sounds good. It so happens that New Micros produces some
suitable Single Board Computers for such an application. Our developer's
package with MaxForth and an F68HC11 SBC might be appropriate. More on the
board itself at http://www.newmicros.com/txx22p.html
We also have a servo motor driver board described at
> Put out
> a command-timing code on the comms bus and programme each motion to
> the time-code with the appropriate activity.
For real time applications, you might want to try an approach using ISO-Max
It gives some unique opportunities for fault tolerance and
multi-processing, and also permits some code to be compiled into the
hardware. By separating state information from the CPU's Program Counter
Register, it may be possible to incorporate some limited AI or expert
> All processor could run a
> native Forth implementation and you get very good control with no worries
> about an OS for which you have no source. There are plenty of simple,
> boards out there (maybe even a handfull of PIC's might do it).
It sure sounds like a good case for SBCs can be made in this application.
While DOS is notably compact compared to Windoze, it has some other
limitations that are painful to bear. Other than running your software on
a general purpose PC, what benefit does DOS provide to you?
> Transport Control Technology Ltd. <http://www.tcontec.demon.co.uk/>
> Going Forth Safely
In the words of Harold Lloyd, "Safety Last!" <grin>
Gen. New Micros