Re: HPGL ...

--- In CAD_CAM_EDM_DRO@yahoogroups.com, Hal Eckhart <hal@c...> wrote:

> Please elaborate on "proper grounding, isolation and modern
> electronics". I've tried lots of schemes which all turned out to
> be useless when I have copper going straight from the motors to
> the bus. The table is earth grounded, but nothing fancy.
>

For noisy environments I urge people to isolate the table/torch from
the controll electronics. That means not giving the units a common
ground connection for noise to travel on. I don't use Xylotex
hardware myself since I run servo's (Gecko 340's) and even on the Z
axis control I use a Gecko 201. All of the gecko's have opto
isolation so you can keep the step and direction (PC ground) separate
from the table ground. Actually the term "common" might better fit
here but would be confusing since Common is used on the Gecko to
indicate the Opto reference point and it can be +5 ! If you have
limit switches on your table then their wires are a rich source of
conducted noise back into the controller.

Probably the best money you will spend is to invest in a breakout
board that has both buffering on the step and direction signals AND
opto isolation on all inputs back into the PC parallel port. By
separating the PC ground and the power grounds it will go a long ways
in getting rid of noise problems. In severe cases you may still have
to add a little filtering to the step and direction lines. Another
good thing to have around is some ferrite cores you can wrap you
wires through. Twisted pairs help on signal lines. Any shielded
wire needs to be ground at one end only (once again to keep conducted
noise between the systems down.


A lot of commercial systems use a common grounding scheme where
everything is tied to one point (star ground) but they have the
advantage of hooking all of that up as a finished unit and to deal
with any noise problems at the design stage.

> -----------------------
>
> It's really good to hear that you're running without trouble on
> the Xylotex hardware. I've asked numerous vendors about plasma
> noise, and I didn't get any promising feedback. I'm tired of
> pulling my hair out that there isn't enough left anyway.

See comment above. I mentioned the Xylotex because it is a cost
effective design. I does suffer from not having opto isolators on
the inputs so all of the voltages end up grounded to the same
common. Depending on your wiring the logic levels can get
overwhelmed by the PWM motor noise or worse by the plasma start noise.
One thing that would help is to use a buffered breakout card and put
a lower value resistor (330 ohm) close to the step and direction pins
to a close circuit ground. That will lower the impedence and make it
harder for noise to effect the signals. You can't do this with a
raw parallel port since it won't drive the added current.

>
> -----------------------
>
> You mention a fast computer. Is this because the steps are
> generated from software? I haven't tried anything speedier than
> a souped-up 486, partly because it wasn't necessary for DOS, and
> partly because I was worried about the plasma noise issues and
> more modern hardware.

The new Windows based software (i.e. MACH2) does generate the pulses
and Windows doesn't like to have it's cycles taken away. You have to
have some more overhead for the basic timer (CPU frequency). There
are some solutions (DeskCNC; FlashCut) that use an external pulse
generator and will run on slower computers but you still need
something that will run Windows 98. The only problem with using
external hardware to generate the pulses is that you have to match
the software to the hardware since there is no "standard" so you are
committed to that vendor's solution. Of course you can stay with good
ole DOS and try TurboCNC but it lacks one of the features I find is
important for good plasma cutting and that is CV (Constant Velocity)
cutting.

>
> Are there timing issues to worry about under Windows? There's
> also the issue that being a blacksmith, there are lots of big
> hammers in my shop. Big hammers, pricy computer, Windows. Could
> get expensive if I don't watch myself. ;)
>
> I've dabbled with EMC, and it looks promising. But I'm not sure
> I'm enough of an ubergeek to run it day to day.

EMC has some good things to think about. It is extremely flexible
and will run on slower computers but you need to be comfortable with
LINUX as your OS.

>
> -----------------------
>
> I'm very interested to see your THC circuit. Not having one
> frequently causes trouble with uneven sheet. If I can solve the
> noise issue, I might play with the spark gap and see if I can
> get a little more length out of my machine (an ESAB 875). I saw
> a demonstrated CNC machine that used a Hypertherm (I think)
> torch which they fired at about 3/8" to save on nozzle wear and
> then dropped down to keep cutting. It would be really sweet to
> combine that idea with a floating torch.

Join the Master5 group and go to the Files section and look for THC.
You will find the THC schematic. There is also an improved assembled
and tested design over at www.campbelldesigns.com We have gone a
long ways in solving a lot of the plasma cut issues by integrating
the THC with MACH2. Since the tip is now under total control of the
software you can easily do things like pierce at one height and cut
at another. The latest version of SheetCam even has that choice on
the plasma tool so it gets integrated into the code. By using a
floating head (one that won't damage the tip if it hits the work
surface and the feedback to MACH2 you can totally automate the Z axis
reference and zero steps. Once the control software "knows" where
zero is any moves of the THC during the cut are tracked and cancelled
out after each cut so you are starting from the same point each
time. I pierce at about .2 " and cut at about .14 inches. Its all
done in my post on the CAM side.

>
> Is there any other way to see the design? M5 hasn't approved my
> membership yet. Maybe I should have been a little less flip
> about the "why do you want to join" question. It does seem
> pretty odd though.

I don't recall a question like that on Master5. Are you sure you are
joining the right group? Also join the MACH1MACH2CNC group which is
the follow-on group to Master5.

>
> -----------------------
>
> You mentioned Rhino in passing. I'll admit that it's not as easy
> as 2d, but it's about as easy as it gets. It has a nicely
> configurable DXF export and opens almost anything. And a great
> free demo. Even if you've run out the demo, you can always print
> to file and import the result. But the most interesting thing is
> that it can extract the 2d faces of a faceted object; even faces
> that are curved (in one direction). I tried writing up one
> project a year or so back at:
> www.casaforge.com/gallery/the_big_roof/construction_story/
>
> I managed to draw an object in Rhino in 2 hours (never having
> used it before) that I couldn't draw in AutoCad in a week. And
> extracting the faces was another 2K plugin.
>
> -----------------------
>
> And to drag this back to the starting point, sort of... I've
> never really had any trouble with getting HPGL into the
> Microkinetics software (aside from using a floppy disk in a
> filthy environment). I like the extremely basic HPGL that Corel
> spits out. You can just look at the file and figure out what's
> going on. And if you use a decent text editor, it's simple
> enough to edit. Even doing things like duplicating objects and
> generating code automatically is about as straighforward as it
> could be, in almost any scripting language. You can keep
> accuracy to about .001", and even more if you cheat.

Hopefully as you hang around here and some of the other CNC related
lists (don't forget DIY-CNC list) then you will see that the
Electronics, CAM and control end of this hobby have evolved a lot.

>
> Hal Eckhart - Casa Forge - Minneapolis MN - http://www.casaforge.com