EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/Master_User.lyx: fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/common/machining_center.lyx: fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/gcode/ (main.lyx mill_canned.lyx tool_compensation.lyx): fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/gui/ (axis.lyx mini.lyx tkemc.lyx): fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/Master_User.lyx: fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/ (main.lyx mill_canned.lyx tool_compensation.lyx): fix a whole bunch of typos and other things piasdom found
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gui/ (axis.lyx mini.lyx tkemc.lyx): fix a whole bunch of typos and other things piasdom found
cradek: when you get a chance, the dapper repo should be ready for signing
* jepler is building the 64-bit packages, will announce after they're pushed to the repo
ok, trying to remember how to do it now...
hi cradek, thanks
(I was having the same feeling when doing stuff like tagging the release .. hadn't done it in awhile
cradek: yay thanks
announce about to go out
welcome, thanks for all your work
I hope it's right!
I have been using trunk so much I don't even know what's in it.
there's a list :-P
EMC: 03jepler 07v2_2_branch * 10emc2/VERSION: bump version after release
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: bump version after release
EMC: 03jepler 07v2_2_branch * 10emc2/src/configure: bump version after release
I won't be gone altogether for the next few days, but I expect to be busier than usual...
(I was worried for a moment because my machine didn't home right .. it tried to move past the switches. but I'd commented out an 'addf debounce' when debugging some bug introduced in 2.3, and left it that way.. oops!)
wow did I get anything about the hm2 changes right in that e-mail? sheesh
it was mostly right
bit of a chaotic release
good evening all. what was all that in the emails about double floats in the PID loop?
KimK: are you from colorado?
I did a trunk update sunday and my system is not the same.
can you describe 'not the same'?
Roguish: jeff switched the hal_float_t type from single-precision float to double-precision, there were some bugs but they should all be fixed by now
had to competely re-tune. all the setting were screwey
when did he do that?
week or so ago
bizarre - I didn't have that problem
you were using trunk before that too?
Roguish: are you using m5i20 or hm2_5i20?
cradek: almost totally untunable.
the regular m5i20.
i've modified my axes configurations several times to get different configurations, but have not had to re-tune the axes, until now.
i had to go into the yaskawa drives and change their tune a bit too.
unfortunately i did not get finished in the time i had to work, so i can't give a wholly difinitive explanation.
what day did you update?
sorry does that mean 2 days ago?
I don't have any guesses. pid was broken very briefly but that was some time before the 9th
i'll work on it again in a few days.
I'd like to see your entire hal configuration
maybe you are using a component that I'm not, and that component is broken
i'll go back to the stock 2.6 and see how it behaves, and then try the trunk version again.
any debugging you can do would be helpful. I wouldn't suggest retuning anything because you'll just have to undo that work when the actual problem is fixed.
also, you can use cvs up -D... to back up along the history of trunk. maybe you can isolate the change that broke it.
got a real job for a while: www.seatel.com
i thought for a bit it might have been in the m5i20 driver, but i did not look for any changes there. then i saw the emails concerning the double floats in the PID.
I know PID is not broken because I'm using it.
but the fact remains that something is wrong with jon's system, and also with yours.
his may be the ppmc driver making bad assumptions about word sizes.
but for yours I don't have a guess
ok, well, i'll work on it.......
and let you know if i find anything.
wait, you had to re-tune the Yaskawa drives also?
ok, close enough :)
it's a bit tricky tuning the double loop. yaskawa and emc
has the machine been sitting idle for a while?
yeah, i haven't worked on it for a couple of months.
you're using PWM converted to analog
then I'll bet something has changed mechanically
unless of course a bug is found or something :)
haven't touched it for a few months.
has the temperature changed much in the shop?
gotten chilly, way down into the 50's
i gotta move wherever you are
east bay, san francisco. walnut creek.
we'll get a week or 2 in the upper 20's in the winter and a few short spells of low 100's in the summer, with low humidity. but that's it.
seb_kuzminsky: Hi seb, good to talk to you. No, from Omaha. I've been meaning to talk to you anyway. I've been planning to order some cards from Mesa for awhile now, hope I get to it soon. Anyway, I was going to ask you if you needed any beta testers, and what cards you wanted to test. But now that you've turned in your hostmot2 for 2.2.7, maybe you're done testing for awhile?
no snow, no tornadoes, no wicked thunder storms, no floods, just the occasional earthquake.
and lots of fog
but you do have the House of Prime Rib (in SF anyway), which can overcome a lot of negatives
a little fog here. i'm just on the boundary between the bay fog and the valley fog.
oh, lucky you :)
Roguish, did you try using the machine before updating EMC2?
no. always been on 2
no, updating EMC2, not updating to EMC2 :)
ie, before the TRUNK update
no. just did my regular update.
KimK: huh, your IP says you're connected via Level3 in Denver, strange, oh well
KimK: i would surely like more testers!
ok. if you know roughly when you last updated, you can check out that version of TRUNK and see if it acts the same as you remember
i'll try some more this weekend. earlier if i can.
what's the deal with 8.1 ubuntu?
KimK: any card you want to test would be helpful... the 5i20 and 7i43 seem to be the most-used ones, so one of the other boards would improve the test coverate
we have no packages for it, and I'm not sure if there's an RT kernel built for it yet (I think there isn't)
no rtai build ?
that's what i thought. i have a machine at a client's that i have to remind not to touch the updates.
you may be able to get it to work, but (1) we have no official build for it and (2) at least one user has tried to get EMC2/RTAI to work and couldn't
you shouldn't be prompted to update
not a programmer, so no worry here.
since 8.04 is an LTS release and 8.10 is not
SWPadnos_ is now known as SWPadnos
seb_kuzminsky: OK, I'll keep that in mind. I'll do what little I can, we'll see. EMC2 is weak on large amounts of DC/ladder I/O, so I'd like to test Mesa's new SPI hi-density I/O cards, but prob can't afford to right now. And I'd like to see a way to connect multiple PLC racks to EMC2 by ethernet (RT?) or RS-422/RS-485? (maybe automation direct PLCs due to cost?) We'll see.
Those assorted PLC modules are just so convenient
SWPadons: what's the status of the m5i20 driver? any changes? is it frozen? any known problems?
no changes since June of last year to the "old-style" driver
any known problems still exist?
hmmm. that's what my email commit list tells me
I don't think so
ok, i don't really have any desire to move to the new drives unless there are compelling reasons.
faster, whiter, brighter, ect.....
one nice one is that you can use the pins from unused encoders/PWMs as I/O
it can also do different quantities of encoders/PWMs (and stepgens of course)
so you can do 4 encoders but only 3 PWMs, if you have 3 motors but want a handwheel also
i use the hal encoder for my handwheel.
well, you can do that too
do you have a spindle encoder?
but freeing up io's would always help.
yeah - I thought you needed two boards because of an I/O shortage
no spindle control yet. very high on the list though.
i still have a strong potential to retro a gantry. that's 4 motors.
sure, plus spindle and handwheel
they don't need the spindle now. just milling tons of foam.
another nice thing with the new driver is that is supports the 5i22, which has 4 connectors
that 5i22 is way more $
here's one for you. i am now at www.seatel.com
stabalized marine antenna systems.
inertial stabilized with motor assists.
the controls are still in assembler on an 8 bit processer.
they build there own stepper and bldc drives.
I saw the link earlier, and immediately thought of wireless internet access on cruise ships
(annoyingly slow and annoyingly expsensive)
all over every where.
i'm on a DirectTV project for offshore platforms.
sure - oil, research, even large fishing vessels
and a railroad unit!
heh. mostly useful in Europe
from 12" dishes to 15' dishes.
very busy right now.
ah well. off for now. thanks for the info. 'night all.
see you later
I still expect the bogus to be in ppmc write_DACs but I'm not seeing it
well, the PID output bounding like that was fairly indicative of a problem on the output side
I wish he would have plotted pid feedback and command too
I'm just happy that the vertical scale is now printed with the trace names
my lathe runs -- it seems nearly impossible that pid is broken
see you Seb
well, that's true
that's why I suggested that he try servo_sim
hm, I have pid with debug=1 on hnc
(well, not your lathe, but ...)
I wonder if it's still busted from the param to pin changes
see you KimK
I'm going to go try it right now.
oh, were you saying good night to seb? :)
sure enough - that's it
but ... I fixed that
I'm so inept at fixing others' ineptitudes
oh I forgot - alex wrongly fixed my wrong fix to his wrong changes
wrong, wrong, wrong - that's one too many or one too few
addr->error_i = (hal_float_t *) hal_malloc(sizeof(hal_float_t *));
oops. one too many * there
and there are no warnings?
why would there be a warning?
oh, good point ;)
SWPadnos: No, I'm not leaving yet
SWPadnos: I wasn't trying to rush you off either, lol
it's OK, I just wander off to bed at some point but leave the IRC client connected :)
EMC: 03cradek 07TRUNK * 10emc2/src/hal/components/pid.c: fix alex's fix to my fix to his param-to-pin changes
I'm assuming those hal_mallocs get cleaned up when the component exits ...
* cradek shrugs
looks like it comes out of the hal shmem region
yeah, it's a hal_malloc rather than kmalloc
cradek: the old saying says "it's impossible to make something foolproof, because fools are so ingenious"
hey, I resemble that remark
I wonder if that was affecting roguish also
I thought it was fine all along, and it was for me, because of debug=1. that bites.
I can't imagine a bit of retuning making it work though
he may be using that too, who knows
I doubt he succeeded retuning it.
I recall him saying 'virtually untunable'
which I bet means 'I never got it to work right'
heh, probably so
I hope he knows his old settings.
I guess that didn't register with me. "virtually untunable" doesn't lead me to think about the machine being at fault at all
having to tweak things might though
when I enabled the amps I got an instant following error, with no noticable motion at all
just a soft 'thunk' and then the amps disabled
what's your MIN_FERROR?
a sign reversal problem?
no, a "this space not large enough for your double" problem
ok, so it gets whacked with a full scale command, and dies .001 later (or so)
yeah I like that. when something is wrong, STOP!
when something "looks" wrong, STOP! :)
not bad for servo amps as old as I am
did you end up having to replace caps in one of them?
and I rebuilt (or screwed with, take your pick) one of the tachs
I found it interesting that the Yaskawa maintenance manual says that you should replace all the electrolytics every 5 years
that's very surprising
for only a few hundred $$ per unit
surely they aren't so bad anymore
I think they still are actually
electrolytics in daily use don't go south that fast
maybe the newer solid ones aren't as bad
there was some recent discussion on CCED about tantalums going bad too, when they're not usebeing d
uh, not being used
how the heck did I get that massive a typo
tantalums going shorted?
my 1939 zenith has what appear to be all original tubes (13 of them?) - all new caps of course
they may not be original, but they are old, and marked zenith
(followed by going up like a match head, usually?)
KimK, I'm not sure of the failure mode they're talking about (though I do know that tants usually die shorted)
maybe we should have developed better cap technology instead of transistors :-)
trancapacitors of course
I suppose the driving force that keeps caps shitty is price
ah, here we go: http://www.bibliotecapleyades.net/ciencia/ciencia_extraterrestrialtech04f.htm
and if you sell a cap that just works and works, for years and years, guess how many more you get to sell? OK, OK, </cynicism>
cradek: Is your Zenith a cabinet or a tabletop?
shutter-dial console - beautiful radio
can't find a picture of it on google images.
This may not be cr's radio, but it's accompanied by an explanation of what the shutter-dial is. Apparently also known as robot-dial (30's ad campaign?) http://www.tuberadioland.com/zenith9-S-367_main.html
Maybe we need to add a "miracle switch" to EMC2? But what should we have it do...
a miracle, of course
Get rid of cutter comp errors by making a smaller radius than you have a tool for?
Just press the "easy" button!
correct the most recent operator error
that would be a miracle
"undo error" is more miraculous
and then there's the "prevent errors" button
Boss: How did you make that big divot in the table? Operator: Well, I was trying to make a pocket two inches *above* the table...
(they should be next to each other)
jepler: finished all other announcements (hope I didn't miss anything..)
alex_joni: wiki ?
micges: hmm, yeah.. might be :)
thanks for pointing it out
my hardy machine wants to remove emc2-dev as part of a partial upgrade... wtf???
emc-dev is in 2.2.6 version still I see
BigJohnT: do not install 2.2.7 untill emc-dev will be in 2.2.7 also
on my 6.06 machine it did not try and remove emc2-dev... only the 8.04 machine
on my 6.06 it disallow installing emc-dev
maybe just some packets are not updated after 2.2.7 release
BigJohnT: hmm.. crap, seems I might have fogotten to upload 2.2.7-dev
BigJohnT: in 3-4h it should be up
looks like that intermediate version of pid would have worked OK until sizeof(hal_float_t) > sizeof(hal_float_t*) which wasn't until the doubles changes
jepler: lol, that's reassuring :)
EMC: 03jepler 07TRUNK * 10emc2/docs/man/man3/hal_type_t.3hal: explain newly-introduced types
huh, the url logger gives me for today's bookmark displays as japanese or something
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-11-12.txt
the file looks fine, if I change the encoding away from utf, it displays ok
it's probably CIA's coloring that makes it autodetect wrong.
oh, so I'm lucky to be using encoding-ignorant Windows programs then (all I get is little blocks where the color codes are)
"by disconnecting the network cable, a local attacker could gain access to locked sessions."
I Had to open it in wordpad.
EMC: 03seb 07TRUNK * 10emc2/scripts/runtests: not that tests aren't fun and all...
EMC: 03seb 07TRUNK * 10emc2/src/emc/sai/driver.cc: fix a small memory leak
EMC: 03seb 07TRUNK * 10emc2/src/emc/usr_intf/emcrsh.cc: fix a rare segfault bug
I was wondering what was all over my desk after running TRUNK it was memeory
EMC: 03seb 07TRUNK * 10emc2/src/hal/user_comps/vcp/vcp_main.c: plug a small memory leak
EMC: 03seb 07TRUNK * 10emc2/src/hal/utils/halcmd_commands.c: fix a small memory leak
seb: did you run a code analysis tool on emc, or did you just take a big fanfold printout of all the emc source and read it from start to finish?
EMC: 03seb 07TRUNK * 10emc2/tests/README: both test.hal and test.sh indicate a valid test
BigJohnT__ is now known as BigJohnT
I have this line of code: pin io float servo.#.scale[SERVO_COUNT];
is scale will be writable and/or readable ?
jepler: rsync: send_files failed to open "/home/emcboard/www.linuxcnc.org/hardy/dists/hardy/emc2.2/binary-amd64/.emc2-dev_2.2.7_amd64.deb.ASMPUt": Permission denied (13)
BigJohnT: emc2-dev pushed into the repo.. should be fine now
BigJohnT: emc2-dev pushed into the repo.. should be fine now
ok, thanks alex_joni I'll do the upgrade when I get home :)
Lerman_____ is now known as Lerman
reversing the path in emc would be a big undertaking - correct?
like - not going to happen anytime soon.
yes, that seems accurate
[Global Notice] Hi all. That split has taken services and 10k with it. We're currently investigating the cause and hope to have it resolved soon. Further updates via wallops (/mode <nick> +w). Thanks!
[Global Notice] Hi all! Terribly sorry for that (fairly substantial) split! Unfortunately one of our sponsors experienced routing issues affecting our secondary US hub, Services and one of our largest MRS servers.. The issues are hopefully fixed for now, but if you have some spare connectivity and hardware, perhaps you should consider helping us not keep all our eggs in one basket! Again, sorry and thanks for using freenode!
alex_joni: weird -- should be ok to rm -f it