cradek: I just tried scoping motion.program-line, but it seems to be zero all the time
ok, I lied
halmeter shows non-zero, gotta figure out what scope is doing
odd - 2nd run and scope captured it just fine
hmm, I think thats a halscope bug
I added the new channel, hit force to trigger it, and the new trace didn't appear (no label or anything), then I hit force again, and it appeared on the second run
ah crap - program-line doesn't help with this program - all the g33's come from the same line (they're in a loop)
gotta unroll that loop
and now you're wanting tomp's #variables in hal
#vars are in the interpreter
it would show you what loop iteration you're on
increment a #var each loop, output that value to hal
for short programs, the interp is done and taking a nap before the spindle reaches full speed - everything is in the motion queue
i guess you could do that with M66 rigt now
m65 i mean
unrolling the loop worked
there are definitely some "misbehaviors" with G33 - they may be a result of shortish segments, but for whatever reason, I'm getting the wrong pitch on some segments, as well as the bobble
the bobble happens on sim-lathe, but the wrong pitch doesn't
unless I change the axis accel values to 5 (X) and 8 (Z) to match my machine
then sim-lathe shows the same errors
ok, we are seeing two different things here
the bobble is NOT a "short move" effect - I lengthened both moves, and it still happens
some of the other weirdness is a short move effect
ok, a series of identical segments (same pitch, same X, Z decrementing by 1/4" each segment) only shows the bobble between the first and second segments, the other segment transitions are smooth
it is independent of the length of the segments, and doesn't care whether the non-G33 move that preceeds the first one is a G0, a fast G1, or a slow G1
can you try doing a series of G33 moves, then a non-G33, then another sequence of G33 (in the same program)?
just curious as to whether the problem is from "startup" or from "synch-up"
I'm about 99.94% sure its sync up
when I was running the real part, I got the bobble on every threading pass
oh, that's true
I wish I had some small clue about the internals of the main TP
* jmkasunich lives by comments, and they are few and far between in that code
dammit, where is that cradek fellow
in his Hot-Rod Lincoln :)
gonna drive me to drinkin'
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-03-11.txt
Good morning alex
How is it going today?
pondering on a mortgage
Good time to buy in the US ;)
so I heared
but commute costs might be much
Heh - doesn't the owner need a usa branch
they already got one
in chicago iirc
[12:02:11] <alex_joni> http://www.cloos.de/english/unternehmensprofil/standorte/North_America.php
Schaumburg, Illinois 60193
That would be in the same state as the cnc workshop...
Galesburg, Illinois, USA
194 miles according to yahoo directions
[12:25:02] <skunkworks_> http://maps.yahoo.com/index.php#mvt=m&lat=41.492945&lon=-89.19409&mag=10&q1=Schaumburg%2C%20IL%2060193&q2=Galesburg%2C%20Illinois%2C%20
sigh, too bad gene didn't mention in his *first* post that this was using stepconf, or I'd have looked into the problem sooner.
hm, the pwmgen manpage doesn't document .offset!
fenn_ is now known as fenn