I still don't think arc vels are right
if I do line->small arc->line at 1% FO (which is about 1ipm), I still get a slow arc
this is on a relatively high accel config
strange, the task issued velocity looks believable though
I also get accel violations on my small arcs :-/
not much, maybe 10%
wonder if it's a different/old problem
but, the 1% FO thing still seems wrong
it's 1% of requested velocity
the thing I thought was a simple fix isn't
say my machine's max speed is F100
I program F1000 in my gcode
I set FO to 1%
should I get 1 or 10?
I know I'm talking to myself. I'm going to bed now.
I lied, I didn't go yet
in the above case, in my tree now, you get F1 for lines and F10 for arcs...
what should FO% be applied to?
the "feed", I know
sure, but emc has never worked that way
before limiting for machine accel/vel
ok, that's my opinion though
since forever, if you set it to 1%, you get no more than 1% of the machine's feed
1% of the maxvel?
1% of min(machine constraint, F word)
in the example, I think arcs are right at F10, and lines are wrong at F1
ok. I think it should be min(machine limit, 1% of F word)
I wonder how anything else does it
so you could take something written for a fast machine, and run it on a slow machine at 25%
I'm tempted to say I agree with you
I'm nervous about such a huge change though
unles it's stupid for some reason I don't know :)
[03:54:39] <SWPadnos> http://cgi.ebay.com/New-Haas-TL-1-CNC-Tool-Room-Lathe_W0QQitemZ280166419203QQihZ018QQcategoryZ97230QQssPageNameZWDVWQQrdZ1QQcmdZViewItem#ebayphotohosting
I wonder what the reserve is
damn! gotta bid at least $510
that's a few miles from you...
a few dozen hundred
looking for a lathe?
I don't recall why I started browsing tonight
when I should be (a) finishing the waveform generation component and (b) packaging an RT OS on an SSD drive
sure is easy to buy stuff with ebay+paypal - too easy maybe
yeah, that may be true
I used to go to the post office, get a money order and envelope/stamp, take it home, type a note with the auction number, and put it in the mailbox
too much of a PITA that way
click click - done and charged on a credit card
if I planned ahead (typed the note and took it along), I could mail it from the post office
then at the end of the month - "oh shit" + click click, and the credit card is paid :)
that's all automatic...
it's funny - I haven't signed up for any automatic payments
numbers in, numbers out
it seems to me the only recourse you have in many cases is to withhold payment
which is harder if things are paid automatically
I only have a few bills that can't be paid online, and online is fast enough that it isn't much of a burden
I use an actual credit card for larger stuff where I think I might have trouble.
heh - I charge as much as I can to get the miles
I even pay the phone bills (landline and cellular) with a CC
it's because of you the CC companies have to screw the other guy so bad
I have one that i use and pay monthly, then there's the $20k or so (just went under 20, actually) that I have on the *other* cards ...
those are low interest though - 3.99 and 4.99% I think
so you ARE the other guy
yeah - down from $65k a year or two ago
I am both guys
now I really should go to bed... goodnight
let's argue about FO tomorrow when everyone can pop their forehead veins
I mean, chip in
excellent - I won't be here :)
hmmm. halmeter never calls hal_ready()
(at least, they all show as initializing)
SWPadnos: that's not too grave a problem, since I don't think either one creates pins or parameters. but it should be fixed..
.. I'll look at it
useful to know what files have hal_init but not hal_ready: grep -l hal_init *.c | xargs grep -L hal_ready
oh ick, I'd forgotten about joystick.c
steves_logging is now known as steve_stallings
regarding feedrate override issue - I think it should be handled the same as other machine dynamics limitations, like blending. The machine should do the best it can to follow what is requested in the G code. If the machine paramaters for max vel or accel change the results will change.
Yes, I know that will possibly make significant changes in how the slow paths are cut in relation to the fast paths, but does feedrate override slow down a blend move that is already limited by accel limits?
Is there any way to get a backplot that color codes the segments that are speed limited by machine dynamics so the operator can visualize the problem?
steve_stallings: no, not really
didn't think so, but I have learned to always ask, sometimes you get a nice surprise
cradek when he was testing the tp had the plot change colors during a blending move. It was pretty cool. Don't think it was kept though.
I think you can see it on the pictures here http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TrajectoryControl
interesting... cradek mentions a "simulated" machine, not the same as a backplot, but potentially useful still
steve_stallings: there's servo-sim which has a simple model of motor inertia using a low-pass filter, and check_constraints.hal which continuously tests that acceleration and velocity are within specific limits (making it easy to capture traces of constraint violation in halscope)
I'm not sure exactly which of those chris was referring to
ok, later, got to run now, thanks
steve_stallings is now known as steves_logging
well - I could have done that ;)
* skunkworks_ likes to use the 'summon' command though
that reminds me of the wraith
roll playing game?
skunkworks_: stargate atlantis
but yeah, it's usually in rpgs
11:18:22 up 3:32, 5 users, load average: 66.18, 93.11, 125.36
good night all