Re:G code diffferences?

From: Ron Ginger <ginger@...>

>
>Darrell wrote:
>>
>> Just about every controller out there has its own variations on G Code. The
>> current closest thing to "Standard" is the FANUC flavor of G Code. ....
>
>
>Could you explain some of the things are are often different? Surely
>they must hold the basic G codes the same, do they alter the use of the
>various M codes for things like coolant on/off, etc?

Most seem to agree on:
g0 as get there fast
g1 get there at feedrate
g2 make a clockwise arc
g3 make a counterclockwise arc

Much beyond that there is a lot of proprietary mix. RS274x is a lot like
RS232 -- everyone claims to have it but a lot of the systems can't talk to
each other.

Some use:
g41 for tool offset left
g42 for tool offset right
g49 for cancel tool offset l/r

others do it with g21,22,?

Some use m words for things like start of flame table burn while others use
g70,71. The fact is that most any g-code system will run a machine but the
actual code to produce the same part may be very different from control to
control. And that is why there is so much effort put into post processors.

I think that most of the system builders thought that if they got into a
shop with one machine, they were more likely to get more sales there if the
g code didn't work like other brands -- but hindsight's 20/20. The net
effect was conversational programming with no standards or interoperability.

A good model for thinking about this might be the translators that take
jpg, gif, tiff, ??? and can make any other out of any starting file.
Another might be to think about the many different written and spoken
versions of English. (I once asked directions at a filling station in South
Lancaster, MA. Might as well have been listening to Spanish! The hand
signals helped, though.)

What we need to think in terms of here is that there are dialects of
g-code. Then we need a kind of translator from AB to Fanuc to EMC to ??.
The really valuable test of the translator would be to have a 3d visual
model generator that could compare the product before cutting.

I know a little bit about the cam stuff in vector. It doesn't use a real
post processor, rather a config file. Fred has written a config for EMC
that works pretty well at least for the 2 1/2 d stuff that I do. There are
also config files for a number of commercial controls and some more
esoteric machines like 4 axis wire/flame/plasma/laser cutters.

You can look over most of the EMC RS274NGC code in the linuxcnc.org
handbook. You can find links to the real documents on the pages of several
of the NIST folk.

Ray