RTLinux lessons, and EDM's

Yesterday I succeeded in getting RTLinux to run the Frank Zappa demo. I
have previously had EMC running with my nonreal time Kernel, so that
shouldn't be much of a problem to finish up. I had two major (to me)
stumbling blocks that hopefully others can learn from.
The first one came when I was updating the Kernel headers and source
files from the website. After applying them, when I rebooted to continue
my endeavor, I was unable to read files from the CD! I got error
messages saying that the Kernel was not comples to support ISO9960 (or
whatever) file systems! Needles to say, losing the CD drive hurt! It
turns out, best I can figure, that since the Kernel was now updated to
xxx.0.36.3, instead of the 0.36.07 that was in lilo.conf, something
would not complete the boot properly. With lilo.conf edited to reflect
the new name of the Kernel, the CD drive was again accessible.
The second problem I faced was that the zImage file would never get
generated in my attempt to compile the RT Kernel. After figuring out how
to scroll up in the shell window, I was able to read the error messages,
and noticed that they all were related to the pci directories. I looked
in the xconfig settings, and there was a note about problems with
certain systems, so I turned off pci support, and was then able to
compile the Kernel successfully. I'm using an old Dell pc with an Intel
82430FX PCIset Pentium 75 chip. Ian, this might have been what you
faced, when you ended up replacing the motherboard..? I'll look for a
patch for this problem later, but for now I've got a functionin RT
Kernel. All told, I've probably got 12 hours into the Linux setup so
far, and I did not set up a modem or network, just sneaker-net between
machines. That said, many of my problems were related to the fact that I
have zero unix experience, so the learnig curve was steep.
*******************************
On to EDM: I recently reread the R. Langolis HSM article on building
an EDM.
Someone here said that they thought another design from ?Model Eng was
better. Could we elaborate on that concept a bit. The HSM book has the
basic design, which is in two parts, the spark circuit, and the stepper
control. Near the end of the book there is a circuit modification for
the spark circuit that modulated the spark intervals. I'll crudely
describe it as taking the straight DC output, and making it a square
wave output. THis was claimed to cut faster and smoother. Is the other
design superior in the spark arena, or in the stepper control area, or
both?
Jon Elson mentioned having EMC monitor the spark current and advance
the electrode into the work. THis is essentially what the stepper
control in the HSM design does. Maybe we could have the actual current
monitoring done as was shown there, and have emc accept motion inputs
off of a card or calculatd in a small code generating program.? (I
really don't know where this was going, but it does seem like if the
board designed by R.Langolis can move a stepper based off the current, a
part of that board could be made to talk to EMC to drive the Z axis,
allowing the idea Jon put forward to work.
Hope somebody can benefit from my lessons learned!

Brad Heuver
Mechanical Sub-Systems fax 734-523-6052
734-52-33480 bheuver@...