[Color Computer] [coco] Learning MW C

[Gene Heskett]

On Sunday 13 August 2006 22:03, Joel Ewy wrote:
>George,
>
>I'm fascinated by how many people on this list are into homebrew CNC.  I
>guess that's the kind of thinking the CoCo inspired in us.  I have a
>distinct memory of my first experience with Telewriter 64, and printing
>my words out on the DMP-105.  I recognized right away that the printer
>was a kind of simple machine-tool, and that the fantastic liberation I
>felt being able to compose my words in the memory of a computer before
>having it committed to paper could be experienced with the making of any
>kind of physical thing that one could construct by means of a
>machine-tool controlled by a computer.  So I can trace my interest in
>CNC directly to my experiences with the CoCo.
>
>I agree with Gene Heskett that Linux and EMC are a great way to go if
>you're really trying to get something done quickly and easily.  I plan
>on replacing my old Pentium 133/DOS/DANCAM system with EMC as soon as I
>find the time and collect the appropriate PC bits.  On the other hand, I
>completely understand your desire to run the machine with a CoCo.
>There's something truly great about a system that's so simple and
>comprehensible, and also about doing as much of it yourself as you can
>and want to do.  At this point I use my simple CNC machine just to drill
>holes with a Dremel.  What it does is so simple that a CoCo could be
>used for the task if I had appropriate software.  If it got a lot more
>sophisticated than that, then maybe a faster computer would be called
>for.  But most of the performance requirements for running EMC is really
>just overhead for the GUI, I think.
>
Not really.  In emc the display update rate is selectable, and I speed mine 
up from the default 200ms to about 50 as that smooths up the paths drawn 
by the gui as it tracks the bit motions.  The fastest loop in emc is the 
output loop, and I'm currently running mine at 25 us per loop invocation.  
Some, with really high performance motors, are running this loop on fast 
cpus at 7 or 8 us.  But that requires much faster cpu's that are otherwise 
overkill for emc.  Other loops, such as the trajectory planner and the 
individual axis control loops run at some multiple of the main loop.  This 
is all done of course with a rather neat mechanism that brings in the 
realtime manager, and actually runs the linux kernel as a subprocess when 
the realtime is loaded and enabled.  Because of the realtime requirements, 
the ACPI must be disabled, so its not possible to do a full powerdown 
shutdown from a magma enabled kernel, but it will tell you when its safe 
to powerdown the machine on screen as its last act before going to sleep.

>If you really want to do it with the CoCo, just to be truly intimate
>with the machine, to prove it can be done, for pure hack value, or for
>whatever reason, here's what I would suggest:
>
>1.  It sounds like you're using (Nitr)OS-9 already.  If so, get the
>source code, and find a driver for the hardware that is as similar to
>the hardware you're using to drive your steppers as possible.  I imagine
>this would be a parallel port of some sort.  The Disto printer port
>might be a good place to start, or something that drives a 6821.  I
>don't remember if you've mentioned how you've interfaced the steppers to
>the CoCo.
>2.  Use the existing code as a base for cooking up an assembly language
>driver for your hardware.  Don't try to start from scratch.  That would
>be an agonizing learning curve.  But look at the commented NitrOS-9
>disassemblies and modify them stepwise until they work with your
>hardware.  It's much easier to understand and modify existing assembly
>code (if it's well commented) than it is to write entire device drivers
>from scratch.  (At least for me...)
>3.  Use BASIC-09 or C to write the user application that interfaces with
>your driver.  BASIC-09 should be fast enough to read your data files and
>feed them to the driver.  If you need interrupts, put all that in the
>driver.  Steal an interrupt service routine from an existing driver that
>uses one and modify it to suit your needs.  The whole question of how to
>do the equivalent of a POKE in C kind of misses the point of the
>C/UNIX/OS-9 philosophy.  The device driver really should be the only
>piece of code that accesses the hardware.  The driver is there to
>provide an abstracted interface to the hardware for the application
>program.  It sounds like you have your hardware working by POKEing
>directly from BASIC-09.  That's cool.  You've got a quick and dirty
>solution just to make the thing go, and you know it works.  But now if
>you separate the direct hardware manipulation off into a real assembly
>language driver, you can play with the logic of the rest of the program
>in isolation from the hardware component.  And if you do decide to make
>it a closed-loop system, or put in limit switches, or otherwise make the
>hardware more complex, you can segregate that complexity from the
>user-interface and data file manipulation functions by keeping the
>hardware access in a proper driver.
>4.  If you still aren't happy with the speed after putting it in a
>proper driver, start putting more of the smarts in hardware.  Use your
>machine to make PC boards for your next generation motor controller
>interface.  Instead of making sine waves in software, use a stepper
>motor controller IC and just send it step and direction signals from the
>CoCo.
>5.  Try to make your driver as generic as possible.  Comment your source
>well, and contribute it back to the NitrOS-9 project.  Somebody else
>might put it to use to run a plotter or another CNC machine, or who
>knows what.
>
>That's how I would approach it anyway.  Good luck, and keep us (or at
>least me) informed on your progress.  It will inspire me to try more fun
>projects with the CoCo, and to get back to work on my CNC machine.
>
>JCE
>
>George's Coco Address wrote:
>> There are too many replies with too many variants to respond to them
>> all. Lots of good stuff, though.
>>
>>  Since I'm ignorant about C, and only intend to use it on a Coco, I
>> suppose MW C is good enough unless I run into problems.
>>
>>  I have been developing a system to machine some small parts for my
>> steam engine projects. I'm using stepping motors on what I call my
>> "Tiny CNC" 3 or 4 axis machine.It uses three or four stepping
>> motors(depending on the application)
>>  It uses a dremel for the cutting.  The machine is built, the interface
>> is built and it works.
>>
>>  A stepping motor requires a lot of steps to make one revolution, both
>> positive and negative on each of the three phase coils to make it step
>> properly.
>> In B09, I can step ONE motor at about two hundred steps/sec using only
>> positive and negative voltages (square wave). Stepping more than one
>> motor takes more time and therefore, slows the process. These stepping
>> motors step at 1.5 degrees/step.
>>  The stepping motors directly drive a screw that is 36 pitch. This
>> requires 8640 steps/inch or about one inch per millennium.  For the
>> most part, when actually cutting, the step rate is just perfect as the
>> dremel can only cut so fast and it works fine. However, at times
>> (especially when moving to a new location to start a new cut), it takes
>> a long time to move it at this slow
>> stepping rate. Should I wish to step these motors faster, I need to
>> also use a modified sine wave. I figure four bits per coil could do
>> this. Of course, this takes more time for the computer to do and speed
>> is getting more serious.
>>
>>  The reason for my desire to learn "C" is so I can build a subroutine
>> in C to do the stepping of the motors. This is why I want to discover a
>> way to use the equivalent of Peek and Poke in C.
>>
>>  It won't be a large subroutine by any means. My B09 subroutine is,
>> maybe fifty lines of code. I figure C would be a lot faster once it's
>> compiled. If it's fast enough, I intend to add some feedback from the
>> steppers. The usual way is with optics. This would require some more
>> peeks, would slow it down and make the process more time consuming.
>> After that, I can begin learning to do the math in C and therefore,
>> start migrating from B09 to C completely.
>>
>>  If I could wrap my head around assembly, I would work with that. But
>> I've tried ASM and keep failing. If I knew what I was doing wrong I
>> could probably ask a question but I don't have a clue how to work it. I
>> figure maybe C would be easier for me.
>>
>>  Can a C routine be called upon from an interrupt? That would be great,
>> as the optical position sensor could call on the routine. I can solder
>> that one up too.
>>
>>  It seems if it isn't B09, I'm lost
>>
>>  I NEED ....."AN IDIOT'S GUIDE TO 6809/OS9 ASSEMBLY( HOW TO WORK THE
>> THING)"
>>
>>  Then I could begin to learn the microprocessor.
>>
>> George
>>
>>
>>
>> Brought to you by the 6809, the 6803 and their cousins!
>> Yahoo! Groups Links
>>
>> <*> To visit your group on the web, go to:
>>     http://groups.yahoo.com/group/ColorComputer/
>>
>> <*> To unsubscribe from this group, send an email to:
>>     ColorComputer-unsubscribe at yahoogroups.com
>>
>> <*> Your use of Yahoo! Groups is subject to:
>>     http://docs.yahoo.com/info/terms/

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2006 by Maurice Eugene Heskett, all rights reserved.