were does "[TRAJ]NO_FORCE_HOMING=1" go to let an Axis sim setup work without homing? Is it the Axis.ini file?
I did a recent cvs (this last week) and don't see that one in the ini
DanielFalck, I think it goes in whatever machine .ini file you are using
DanielFalck: in this case if it's not listed, it's assumed to be 0
yes, if you want it, put it in yourmachine.ini
Hi Jeff.. whats cookin?
LawrenceG: how many miles are you from the US?
about 100 as the crow flies
LawrenceG: what city?
Parksville BC is the closest bus depot
Nanaimo is about 1/2 hour away
LawrenceG: Well, try the US side instead - closest city?
Bellingham then Seattle
re no home stuff- it worked
[TRAJ] on a separate line
thanks- I'm using sim at my desktop here
the other (easier?) think you can do is set homing velocity to zero, so home doesn't cause any motion or search for switches
then you can still home
for sim it doesn't matter, but for a machine with no switches there's a huge benefit to homing (so you get soft limits)
I'm using axis to backplot gcode and I launch it from gedit- a lot, so not homing is convenient.
LawrenceG: $135USD to Bellington WA
I figured out how to alter tool 0 to look a bit different- that was fun
I guess I could put a flag in the tool table to try to show different types of tools
JymmmEMC, thanks, but will have to pass.... how new are the batteries.... maybe ebay them without batteries to get shipping reasonable... but new batteries are kind of pricey... I just bought a single for my alarm panel
LawrenceG: I have batteries, but was including the existing ones in the price.
LawrenceG: Still, the shipping is price of $130 was two 80lb pkgs
LawrenceG: gimme your zip?
[01:04:34] <LawrenceG> http://imagebin.ca/view/M6s1ai.html
picture of this morning
JymmmEMC, dont worry about it, I have a 12v UPS which runs nicely off a large battery
SWPadnos: are you there?
SWPadnos: i thought that I had finally rid myself of the rabbit uc forever when i discovered the pic32 line of ucs. turns out that my new project requires 256K of sram (the pic32 only has 32k of ram and can't interface with external memory very well), and the rabbit's pretty much my only choice. the project also requires i2c, so I thought I'd check out the rabbit's i2c lib (i2c.lib): http://www.sonoma.edu/users/b/bloomal/Labs
(sorry i have to vent) fucking rabbit. FPOS. whoever designed this garbage should be shot in the gut.
press fit - 0.376 dia hole on a 0.375 shaft
got it half way on, and now it won't budge
the hole was .001 larger than the shaft?
0.375 hole, 0.376 shaft
the sleeve can be re-made rather easily, the shaft not so
famous saying(?) "Take a Dutchman as he means, not as he says", LOL
I'm about to start turning the sleeve down till I can get it off
is it a through hole?
can you apply heat (lots! i.e. oxy/acet) to swell the hub while cooling the shaft (boiling water, maybe?)
but the sleeve is about 1/4" from a much larger shoulder, so I can't really grab it to pull off
KimK: I already tried propane torch, got it blue (about 500 degrees F)
can't really do much to keep the shaft from heating up
if I had heated it before I put it on, that would have worked great
maybe propane & ice?
can't cool the shaft
the sleeve is 3/4" long, only 1/4" to go
OK. Guess it's turning then.
so I have 1/4" of exposed shaft, and the shaft end is 1/4" down the hole
no way to cool a 1/2" length of shaft from one end and the 1/4" that is exposed
drip liquid propane on it?
turn the propane tank upside down?
How about don't cool it and heat hub till orange (soft)?
that will heat the shaft to orange too
so tap gently
they are simply too well joined to heat only one (or cool only one) at this point
my roomie in college was a grad student in Chemistry -- we tried to get a bike stem out by freezing in liquid nitrogen
turn it off
didn't work, but it was a lotta fun
You might be interested to know that that is how they keep railroad wheels on railroad axles. The hole (approx 5.0", 5.5", 6.0, 6.5", others?) is 1/64" (~0.016", 0.4mm) too small for the shaft. Cold press too, no heating or cooling.
there is a lot of that on aircraft engines too, bearings are a interference fit on the shafts
Oops, those are the bearing journal sizes, the wheel lands are a little bigger (inch or two?)
well, turning worked
there is no way anything else was gonna work
sorry you had to give up your hub
I turned it till about 0.005 of sleeve was left, then slit the sleeve with the corner of a triangle file, expecting it to pop off, or at worst peel off
there was an area about 1/8" square that was cold-welded on
even with the rest peeled off, I had to turn that last piece off - no detectable joint between it and the shaft
* jmkasunich <-- dumbass - I tried to press it without oil
how much torque does thing handle? what does it do?
very little torque at all, the fit just needs to be enough to keep it from falling off
a couple tenths of interference would do, but trying to hit it that close is really tough
OK. So I guess .001" is too tight then.
[02:23:27] <jmkasunich> http://jmkasunich.com/pics/worm03.jpg
the part in question is the top one in the pic
the sleeve to the left of the gear provides a surface for a needle bearing to run on to handle radial load
the screw is threaded into the gear and pinned, then turned down to accept the sleeve
Yes, I can see why you didn't want to botch up the shaft. Wise choice. Sorry about the gear. How long did it take to make the gear?
haven't made the gear itself yet - just a gear blank so far
(the pic is from one of these contraptions I made several years ago)
but still - there is a fair amount of time in the gear blank and screw, and I don't have enough bronze to make another blank
Sorry, that must predate me here. Contraptions?
its a test jig I make as a side job
used in labs to test "crack isolation membranes"
if you want to lay ceramic or stone tile on a concrete floor, you put the membrane down first, so when the floor cracks and moves, it won't crack the tile
Interesting, I didn't know they did that. Thanks.
Say, with all the manufacturing capacity available in-channel, does anyone know of any buyers that are looking to have things made? Especially by CNC? A friend of mine has a small CNC shop and is looking for work. Maybe a common complaint these days?
small cnc shops, from the chatter at my shop, are the ones having problems right now
who isn't having problems?
bigger shops in my area
our shop couldn't keep one guy on because we're slow, and he was under 90 days
you're in a small shop?
dude got a job right down the street no sweat to one of the places we occasionally outlet work to
no, but we do our product and no one else's
so when sales is slow, even though we have a decent number of machines, we get thin on work
so people aren't buying air bearings
generally not in the first quarter, no, but it picks up in 2nd and 3rd
then we crush to make it out by the end of the year, then it slows down again
but with the economy, it's just a little slower this time of year than normal
isn't that always the way
most of our customers are overseas and in industries that haven't been hit, but are slowing production just in case
I wish I lived in York PA, there are houses selling for $14K
I could buy a shop
there are probably cheap houses around here, but they are too far away
the locals are a batch of theives
Well, just got upgraded!
JymmmEMC - now new and improved!
Upgraded to 8.04? Everything seem OK so far? What's your machine and interface?
LOL, sorry I meant my internet
Ah, well that's good too
Right now I have ADSL 3M down, 512K up. In January moving to AT&T U-Verse and will be 18M down, 1.5M up
This should be a simple question but I sure can't find the answer; How do I configure the default location where emc2 opens files? I don't want to use the default ~/emc2/nc_files.
in the ini file
PROGRAM_PREFIX = ../../nc_files
Is that the machine config .ini file or the axisrc?
Got it. Thanks cradek.
[05:15:48] <eric_u> http://cnczone.com/forums/showthread.php?t=70620
How do you change an x driver so that it doesn't stick across reboots?
actually want it to stick across reboots, I'm just wondering how Heinz did it
Look into the /etc/X11/xorg.conf file for the 'Device' section. There may be a 'Driver' parameter. The newer (8.04) and later generic config doesn't always specify the driver but you could try adding it to force it.
Also check the /etc/modules and make sure the old driver isn't getting loaded there.
JymmmEMC, you still awake?
leave a message at the beep........
LawrenceG: whats up?
hey.... I am cursing dvd's.... I want to back up my home directory, but file structure seems to be limited to 6 levels deep.... which mine fails
LawrenceG: nix or win?
are you burning ISO or UDF ?
I could tar it up, but that makes for horrible backups
LawrenceG: are you burning ISO or UDF ?
or shouldI say 'creating'
gnomebaker dvd data disk is all I have found so far
I doubt it's a level limitation as much as /home/lawernce/cnc stuff/emc/source code/configuration/machine1/really long file name goes here.txt
ISO is limited by length of full path name size
UDF not as much
yea.. it says it fails on level 7 and that only 6 are allowed
I'll get the error message....
LawrenceG: what proggy?
I: -input-charset not specified, using utf-8 (detected in locale settings)
genisoimage: Directories too deep for 'lg/.macromedia/Flash_Player/macromedia.com/support/flashplayer/sys/#bin.clearspring.com' (7) max is 6.
I've only burned under win, using nero. but I think you can use mkisofs to create an iso file that you can burn
you mgight see if there is a mkudffs that cna do the same thing.
I even nuked flash and emc as they failed depth tests.... next one to fail was one of the .gconf directories.... I will see what alternatives I can find
[06:24:58] <JymmmEMC> http://ubuntuguide.org/wiki/Ubuntu:Feisty/CDDVDBurning#How_to_burn_files.2Ffolders_into_CD.2FDVD
How to create Image (ISO) files from folders Read #General Notes mkisofs -r -o file.iso /location_of_folder/
-r is recursive I belive
the CD/DVD creator seems to go into an infinite loop.... I dont think it handles 3.4gb of data properly (2gb limit)
take a look at that page I linke you too, some good info in there (in general)
bbiab... freezing out here - hard to type, need to warm hands up
JymmmEMC, thanks.... something is cooking :}
LawrenceG: ok, im back for a few... how ya doin?
JymmmEMC, , I think we gotter
k3b seems to work and also burnt one using cmd line stuff building an iso then burning the image
LawrenceG: cool beans. ISO's are great things
did you create the iso using mkisofs by chance?
yes... k3b had some interesting questions about following links and also whether to use long file names that violate jolliet (windows wondt like the dvd)
Yeah, as M$ isn't case sensative
anyway... I feel happier with a backup... I hate depending on a single piece of hardware... I have had 1 experience with a bad p/s that took out everything including the disk
LawrenceG: check out rdiff-backup
Like rsync with version control
or just use rsync if you don't need the version control
on my last box, I actually had a spare box that I would only turn on to do backups using rsync
dvds are nice for offsite backup, but they dont get done as often
I really want a NAS box, but awefully expensive for what I want.
Well, use rsync + mkosifs to automat it and nag you to burn the ISO's
external sata drive is nice as well... faster than network
true, I have a 1TB here ext
its time for bed... thanks for your help... catch ya tomorrow or next year!
G'Night, tomorrow perhapse. glad I was able to help (for once) =)
bob999__ is now known as KimK
[08:54:23] <DanielFalck> http://imagebin.org/34596
time for bed
Are there any GPL / LGPL experts here?
If I write a LGPL program, can it call a GPL library (in a dynamic lib)?
Also would it then be legal fro the LGPL app to call a closed source library with the GPL lib also loaded?
The two libs wouldn't be aware of each other's existence but theoretically they would be in the same address space.
that's a mighty fine question, to which I do not have an answer :/
My brain is melting...
The reason for this is that I am writing a GUI that can be connected to pretty much any motion control system.
For instance it could run under Windows and talk to Mach or EMC (via emcrsh)
Or even both at the same time :-0
Personally, I'd say GO FOR IT, :/
It is a fully customizable GUI so you can place your own controls, have multiple pages etc.
Comms with the motion control is done through plugins.
Currently I have basic control over EMC using emcrsh.
Hopefully in the next few days, I'll have it actually controlling hardware, rather than just simulation.
Thanks. I am a bit worried about the licensing issues though...
dno, perhaps ask in #gnu ? :p
I might do that. Thanks.
Looking at gnu.org, I have found something slightly worrying:
If the program is written by US federal government employees in the course of their employment, it is in the public domain, which means it is not copyrighted. Since the GNU GPL is based on copyright, such a program cannot be released under the GNU GPL
The original EMC was released by NIST as public domain.
Was it legal for the GPL to be applied to the original emc code?
my view is original code has the original license, new code has whatever the writer/s wishes
tbh, if you leave the licenses like they are, include them with the libs they belong to, and your own app,... Can't really go wrong
if you don't claim you've written them, i don't really see a lawsuit pending,.. especially if yours will be freeware too
I'm just trying to figure out the worst-case scenario. I expect to release my code under LGPL.
From what I understand so far, if I distribute my app with both closed and LGPL plugins it will be in violation.
However if someone else distributes the closed plugins that would be legal.
isnt gpl3 supposed to sort the sillyness out
Most EMC code is GPL2. You cannot mix GPL2 and GPL3 in one app.
maddash is now known as jubjub
LesNewell: as i suggested earlier, join #gnu and ask.
it's why that channel is arround
They aren't all that sure either :-)
seems a hotbed of techies in the Devon Cornwall area judging by ebay toys
LOL. yup, MarchantDice are just down the road - great for ballscrews, slides etc.
I currently looking for electronics test gear and find a few sellers from the area
You can never have enough test gear :-)
I has a habit
I can kick the habit any time.
whenever you run out of money?
Something like that :-)
There seems to be a fair number of UK guys here.
a few, they dont all lurk in channel full time though
nah, only during the day
some at night, but most need sleep
I pulled out an old motherboard - slot 1 500mhz. ran the latency test and glxgears all night. this morning latency was 13000 :)
took 3 tries to find working memory to atleast get me to 320mb
I installed dapper. the old drive that I found to use with it actually had bdi on it.. (didn't boot though)
actually I don't know that.. I gave up after it was at the splash screen for a few minutes.
If you really want to beat on that motherboard, try glblur.
If you haven't got hardware opengl your mouse may get a bit erratic...
heh - on this machine glxgears is really slow..
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-12-30.txt
Maybe not such a good idea. Makes my 1.5GHz Dell cough.
My latency's worse than yours :-(
what kind of numbers are we talking?
17500 on my 800 meg athlon for glblur
dammit- XChat for Windows keeps crashing on me.
Just tried running EMC in a virtual box under Windows - latency's a bit high :-)
windows!!!! get linux installed over the top of windows that fixes it
12300 on a P4 with glblur
lol. Much as I hate to admit it, I find it quicker to develop code with VS2005
I love the debugger.
I will have to try glblur
VirtualBox latency 1161419446 and counting
Do you think it will run my steppers at 100kHz pulse rate? :-)
you should probably go back to mach... ;)
Er - not if I can help it. I need something reliable.
archivist: there is a guy on cnczone looking at doing gear hobbing..
[15:11:42] <skunkworks_> http://www.cnczone.com/forums/showthread.php?t=70544
using 2 rotories?
I don't really understand what he is trying to do.. (as you can see from my uninformed post)
bevel with flycutter is not hobbing
dunno why he want to ignore Z thats silly
unless he is too tight to get another drive
but he is attempting to generate, that will be fun! and not sure of maths and correctness with a bevel
LesNewell_: that's the proper attitude ;)
It was the last straw when it stuffed a cutter into my chuck on my lathe fro no apparent reason.
did you run winamp?
10HP spindle motor + 1HP serbo = big mess :-(
I hear that messes things really good up
so far all my crashes have been from me ;)
er - servo
Rutex servo drives - they just keep trying in a crash.
LesNewell_: you was using mach?
how are they driven?
I was using Mach.
step/dir or some servo int?
I had heard lot of similar stories
the must be step/dir
e.g. analog, serial, eth, ...
step/dir. They have a HUGE following error limit.
you might want to try a hardware board sometimes..
I hear it makes a huge difference :D
you basicly have lots of advantages (besides the extra cost ;)
Got EMC on the lathe now. Love the MPG response. Not so keen on the available GUIs
well.. that's the fun part in emc2 :)
you can always write your own
wrong window sorry
Funny you should say that - look in the logs fro an earlier discussion :-)
my connection is crappy atm.. I can barely type
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-12-30.txt
Been looking on the Mesa site - Drool...
I have an unused 5i20 + a couple breakouts from them
nice stuff ;)
Their 2kw drive loks tasty as well.
was the discussion today?
Yup. A couple of hours ago.
ok, so only the GPL2 vs. GPL3 vs LGPL part?
Yes. It is a problem for me 'cos the front end is pretty universal. It could even be used as a front end for Mach.
I understand that
LesNewell_: we had a guy from FSF contact us a while ago, he was (if my memory serves me right) a licensing engineer
I'll try to look up his email addy, and send it to you if you want
Thanks. I am in the process of ripping my plugin system apart to use sockets instead.
That way I'll be completely covered.
I just found an email from you from 20 dec. 2005 ;)
I didn't remember that we had contact that far back :)
What was that about?
the emc1 interpreter
Oh yes. That was fro the GRex project.
This new GUI fro emc is based on work for the GRex
got a screenshot?
Coming up if I can get GIMP to play ball.
[15:46:38] <LesNewell_> http://www.sheetcam.com/emc/shot.png
This is only a testing layout.
It has a built-in layout editor.
* alex_joni watches bytes dribble in
The screenshot shows it running on my Windows office comp and talking to emcrsh running on my house comp.
I always wondered what the attraction for coloured buttons was
If you don't like the bling, it is easy to change the graphics through config files.
I was just testing the theme code.
good to know, but it still makes me wonder :D
Das blinkenlights - impresses the yokels.
I only like towel.blinkenlights.nl ;)
alex_joni: Broken link
I like http://blinkenlights.net/
Defintiely thinking big :-)
If your're into electro http://www.youtube.com/watch?v=vWL1QOXAcgM
LesNewell_: you need to connect using telnet
btw, I finally found the email I was after:
Licensing Compliance Engineer, Free Software Foundation
seems it's from 2006 ;) so not sure he still works tehre, but I can send you his email addy
Thanks. It would be handy.
towel.blinkenlights.nl is good :-)
is your fastmail addy still valid?
back,(with firewood even)
'Speaking of Mach3: The price is going to increase on January 1st. You might want to order it before then.'
* archivist doesnt bother
emc2's price will double on Jan 1
0 * 2 =0
* archivist buys more
alex_jon Just curious, do you have anything like AQMD/EPA out there?
my google skills are rather limited
alex_jon1: EPA is the Environmental Protection Agency
they bust you for pollution crime
sure there is
alex_jon what seb_kuzminsky said
and I'm quite confident the restrictions are more harsh than in the US
alex_jon do they restrict fireplace usage on certain days?
There are air quality laws in some towns here. Depends where you live.
epa is federal, local laws may be stricter
JymmmEMC: home use fireplaces? nope
alex_jon K. Here, on bad days they do. 1st notice is a warning. 2nd is a $500 USD fine, 3rd is $1000
and they have ppl driving around checking chimneys
there are houses here that only have wood heating
I'd say the majority rural population
Yeah, and wood is MUCH cheaper than gas/electric
hey - turbocnc v5 will run in windows.... http://www.dakeng.com/turbocnc5.html
the "here" that JymmmEMC is talking about is california
jmkasunich: figured as much :)
chimney checkers - I know of places in the US where chimney checkers might get shot
jmkasunich I wish that was here... Some ppl especially elderly can't afford to heat their place otherwise
alex_joni: um yah.. Sounds like it is just 4.01 with a gui.
skunkworks_: hmm.. no coloured buttons, but they couldn't help with simple text displays
had to use those "nifty" LED displays
Well, they have a 3D view too it seems
I don't think I would even bother with TCNC5
But I actually like MS-DOS too =)
I'm not sure it's 3D
well, a visual then =)
my new motherboard has a com port, but it is just a header, no db9 connector. kinda sad
Hi, I added some minor ticks to the meter widget http://imagebin.org/34611
someone who wants to commit it?
put a patch on pastebin, I'll review and commit
[17:52:30] <JanVanGilsen> http://pastebin.com/m32c6e8e0
what file does that go in?
line 14 -> 20 make the minor ticks
you mean lib/python/pyvcp_widgets.py?
this would be easier if you posted an actual patch
if i'd knew how ...
is your code supposed to replace the "draw_ticks(self,cpr)" at line 271?
oh, I bet it is the draw_ticks(self) at line 345
are you working on a cvs checkout?
cvs diff -u will show the differences between your modified checkout and the original
I'm not sure if that is exactly a patch, but it is certainly close - if you can pastebin that it would help
* jmkasunich wiats
jmkasunich: vacation been productive?
jmkasunich: the thing from cvs diff -u is exactly a patch, which is nice
cradek: thanks - I wasn't sure if any additional diff options were needed
there are some long expressions in that function, I don't feel like going crosseyed comparing original vs. modified, so I'm waiting for the patch
I think the only changes are the code tacked onto the end tho
[18:07:33] <JanVanGilsen> http://pastebin.com/m383ed75e
I only get the first 14 lines - it is cut off
hmm, true - there should be a couple lines of unmodified context at the end
interesting that the original pastebin cut off at the exact same place
JanVanGilsen: this is what my cvs diff looks like: http://pastebin.com/d19c3f69a
why doesn't your diff show the last couple lines? did you not paste them?
[18:16:08] <JanVanGilsen> http://pastebin.com/m4b4fae6b
heh - when you jog turbocnc5 the dro says 'moving' instead of displaying the position.
i didnn't paste them
* skunkworks_ hugs emc2
thanks JanVanGilsen - changes committed
skunkworks_: you actually tried that? ewww
heh - it didn't trip the virus software...
hehe thank you for committing :)
might be a manual virus :)
a CNC virus - it machines copies of itself
nah its a dos attack
ah, now I know what DOS stands for ;)
By the way, what was the final conclusion with my lathe diameter patch?
I wasn't following that closely
I thought there was some uncertainty about things like canned cycles?
I sorta lost track myself.
As far as I can remember we couldn't get any agreement on what happened with the threading canned cycle.
hello all, Jerry here, I have a test system on my laptop, just setting it up (not hooked to anything), and I can't start the code, I keep getting a joint 1 on limit. I have the limits switches as inverted so as to not have them
the laptop is being used to test axis mods, not to run a machine.
just realized there is no parallel port on this machine so maybe I don't have to invert the joint 1 limit?
wyoudn't you be best off using one of the sim configs then?
I guess I could, but had just copied the config from my running machine over since it had the filters, etc mods in the ini. I am going to try to not invert joint 1, didn't try that
Guest289: use halcmd or halmeter to look at the limit signals
if your computer has no parport, but you are running the parport driver, you are going to be reading garbage
is there a way to tell emc to just ignore the limit switches? I don't have them on my mill either
and you are going to be writing "step pulses" to some I/O address where the is (hopefully) no hardware
if you really don't have them, then don't connect them in your hal file
working now with joint 1 in non-invert
(if your hal file was generated by stepconf, then I don't know - doesn't it let you choose that?)
you can choose to invert or not invert but I didn't see a place for ignore or don't have...
I haven't used stepconf enough to know what all its options are
don't you get to decide what is on each parport pin?
also, does the mdi command when run in a terminal window work? I am getting an emcstatusbuffer invalid error
in stepconf you can make the limits 'unused' I guess I should have seen that
dunno what you are saying - if you are doing something in a terminal shell, copy the exact command and the exact response and paste it here (or at pastebin.ca if it is more than a few lines long)
the exact command is -> mdi /path/to/emc.nml
I then get the MDI> prompt and just press enter to get the status
with EMC already running?
yes, emc is running
I'm afraid someone else will have to help, I have no experience with the command line mdi (didn't even know it existed to be honest)
pressing enter gives me a traceback with the final line being -> emc.error: emcStatusBuffer invalid
you are supposed to get status
another question, i can't get this java based app to start in firefox under ubuntu, it keeps asking for a plug-in, I've downloaded and installed them all...
[18:59:21] <alex_joni> http://dsplabs.utt.ro/~juve/blog/index.cgi/photography/01230658670
alex_jon1, I should have been out taking pictures today, got a new to me Canon
huh - I thought the busybox prompt was the pollirq issue. I think I might be wrong.
skunkworks_: be my guest and suggest that ;)
skunkworks_: what Canon?
alex_joni a 300D I got given a dead one, tried to repair (not very hard) but went and got one on ebay yesterday,
sweet.. guess it's not very pricey these days..
that's about 200 eur?
at xmas the price was higer had to wait till just after :)
did you get a kit or only the body?
about dunno what the rates are atm
includes lens and box of bits etc
cool. wonder what the current price on a 1000D is
I had film EOS so have a long zoom to fit as well
dunno wasnt looking at that model
it's the last from the rebel series
hmm.. about double than that
seem to be 350-400 £ here
good night all
good night alex_joni
cradek: when I'm running this config http://www.pastebin.ca/1292221
, I have different axis x velocity than commanded
when I'm F500 I have 417 on ddt
can you take a look ?
micges: that ini has MAX_VELOCITY=80 for all axes... how are you getting 417 i wonder?
I've tried for 3h with alex fix it but with no result
how can you get a vel 5x higher than your specified max vel? something smells fishy
look at http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-12-23.txt
417 is mm/m, not mm/s like the ini
it happens to be a ratio of 50:60 (417 is 5/6 of 500)
(dunno if that matters)
SWPadnos: I see that you look at the conversation , any idea why changing servo-period changes fb velocity ?
could be something weird with ddt
shouldn't be, but could be
I have 417 on axis.0 hal, and ddt
your PID parameters look far different from any I'v eseen
yeah it must be due to european 50Hz AC power
I of 35000 is insanely high
his gains are in metric
yes metric only
micges: what's your hardware? are you stepping out the parport?
hw is 3x baldor servo driven by analog vel signal
not with that base period
what's making the analog vel signal?
io card simmilar to mesa or so
but not a mesa?
connected with pc thru LPT
ok, so it's got encoder inputs and analog output?
have you scoped the analog vel-control signals coming out of your controller going to the baldor amps?
you should get a following error if the hardware refuses to keep up, so that shouldn't be the problem
processor + io pins + enc counters + some stuff
well, I don't know what the problem is, and I have to think about moving the lathe now. good luck :)
good luck with your new lathe SWPadnos :-)
you should reduce the configuration to one that can be run as a simulator; that would help us actually look at the behavior on our own systems
still on the trailer?
current vel from status is 500
should get it lifted off today
pure X move (no Y/Z)?
ok (just checking :) )
on sim it is also
micges: and what voltage is your board sending to the baldor?
sorry +- 10 v
see, that's why you should reduce it to showing the problem is on sim -- that proves that it's not due to your hardware interface or to you drives, none of which we have
then we can skip this whole discussion
no micges i was asking: when you command a 500 mm/m move and the ddt of x-pos-feedback says 417 mm/m, what voltage is your board actually outputting?
when I saw same problem on sim I have not measured it
oh so you *have* a sim config that shows the problem?
on that config http://www.pastebin.ca/1292221
when you run it with sim configuration will se the problem
no, we can't run that file because it refers to hal files that we don't have (core.hal)
jepler: copy only params from traj, emcmot, axis_n to sim and run
micges: replace the traj, emcmot, and axis_? parts in sim/axis_mm.ini with the ones from your pastebin?
ok, so what part is *different* from axis_mm.ini? I don't see this behavior you describe with axis_mm.ini, so knowing that you should concentrate on the differences
you know which parts, when moved around, cause the problem to happen
now narrow it down to as small a part as you can
base period to 150000
all traj section and all params from axis_0
noob alert here... i started emc2 with the axis_mm config with micges changes, it wont let me move because it's not homed...
when i try to home it it just takes off
set search latch vel to 0
seb_kuzminsky: remove all the HOME_ lines from [AXIS] to get "homing does nothing" behavior
jepler: didn't know that default behavoiur is home does nothing, thanks
ah in fact it is the relationship between BASE_PERIOD and SERVO_PERIOD
I have base to 150000- it does nothing
I have servo to 1000000
I have pid-thread additionally set to 300000 (2x base thread)
micges: i did what you said, and "g0 x100 f500" gives me a reported vel of 4800 (80 * 60)
so i dont see it
that's with axis_mm with those sections from your .ini
change BASE_PERIOD in sim_mm.ini to 150000 and you'll see this behavior that micges reports
I get 476.19048 = 500 * 1000000 / 1050000
because the motion planner actually gets a period of 1.05ms, a multiple of 150us
i'm using axis_mm, the [EMCMOT]BASE_PERIOD is 150000
SERVO_PERIOD = 1000000
jepler: on my config:
13:38:35 <micgesEMC> ok I have clue: when I change servo-period from 1000.000 to 500.000 I have : f500 fb-vel=379
13:39:40 <alex_joni> * alex_joni looks in cradek's direction :)
13:40:27 <micgesEMC> when I change servo-period from 1000.000 to 2.000.000 I have : f500 fb-vel=477
TRAJ_PERIOD = 10,000,000 (without ,'s)
if you don't run anything in BASE_PERIOD, set it the same as SERVO_PERIOD, or set it to 0
I must have pid-thread at 3,3 khz
in a "real" system you'll still see some error but not quite as large because the requested period and the actual period will not quite be the same
float function thread
ok, then make SERVO_PERIOD an exact multiple of BASE_PERIOD
I misunderstood "I have base to 150000- it does nothing"
that was only way to run it
OK, what I'm saying is that the fix to your inifile is to make sure SERVO_PERIOD is an exact integer multiple of BASE_PERIOD, and TRAJ_PERIOD is an exact integer multiple of SERVO_PERIOD
you'll still have some small error due to not getting the exact BASE_PERIOD you requested, but it will be much smaller than the 5% error you get now
i've got BASE_PERIOD=0, SERVO_PERIOD=666,000, TRAJ_PERIOD=10,000,000 and no problem visible
are you guys seeing the problem by tweaking configs/sim/axis_mm.ini? or something else?
seb_kuzminsky: yes. I saw a problem when I changed BASE_PERIOD to 150000 in axis_mm.ini and observed Xvel in halmeter
ah, i was looking at the "Vel:" display in axis
running this program: G21 / G0X0Y0Z0 / G4P1 / G1X200F[10*60] / G4P1 / G0X0Y0Z0 / M2
I expected to see 10 for Xvel in halmeter but saw 9.5xx
i ran this: "g0 x100 f500", axis' Vel said 4800 (expected), and halmeter Xvel said 80 (expected, i think)
seb: g1 not g0
use g1 so it goes at the requested feed rate
can someone recommend me a small cnc machine for jewelry/rings/goldsmith work? Thanks!
ok with jeff's program Xvel is 10 and axis vel is 600
seb_kuzminsky: if you were seeing the bug with your G0 Xvel would be about 76.19 instead of 80
seb_kuzminsky, do you currently have SERVO_PERIOD = (not an integer multiple of BASE_PERIOD)?
-BASE_PERIOD = 1000000
+BASE_PERIOD = 150000
jepler: I can see solution for it, but can't see reason ?
this is the change I am suggesting to see the buggy behavior
SWPadnos: yes 0
do you get a base thread with that? (it's obviously not possible to have one with 0 period)
micges: because emc/motion/motion.c does calculations about servo_base_ratio based on the inifile PERIODs, not the actual periods of the threads
SWPadnos: i have only a servo thread
jeff suggested it since base doesnt do anything in this config i think
yes. he was trying to point out a way of seeing the problem
which is to have base and servo periods which are not integer-related :)
but i'm not seeing the problem
base = 150000 and servo 1000000 and you will see
BASE_PERIOD=0 was a workaround I was suggesting because I thought (incorrectly) that micges had no useful work in his base thread
now i've set BASE_PERIOD to 150,000 and SERVO_PERIOD to 1,000,000, but i still dont see the problem
actual servo thread period is 1,050,000
(from halcmd show thread)
only thread is servo-thread
uh this is trunk, from the past day or so
right now - latency is 5534
I've verified that I see the problem with no other config file changes than the one I showed above
running for 20 minutes or so...
I have some local changes to axis.tcl and axis.py, but that's it
jepler: i just tried sim/axis_mm.ini with your one-line change, and i *still* dont see the problem here
i'm running in a VM, maybe that's relevant?
axis' Vel: says 600, but halmeter Xvel says -9.5 here too
ok that's confusing
where does axis' "Vel:" come from? Not from XYZvel, that shows the same as Xvel
seb_kuzminsky: the motion controller calculates it
so it's commanded rather than actual velocity?
jepler: seting servo to integer multiply of base solve the problem
the motion controller thinks it's working at a rate of 1ms (and sets the stat buffer velocity according to that) but it's really running at 1.05ms
ddt doesn't have that "bug"; it finds the correct velocity based on the actual period not the requested period
this is very dangerous (very hide) aspect of emc
jepler: ah ok
is there a reason the motion controller can
can't sense the period?
seb_kuzminsky: it looks like old emc1 code that was insufficiently modernized
gotcha, that makes sense
micges: most people wouldn't notice a 5% difference in speed from requested, but I accept without argument that in your case it is important.
I have 417 with 500 commanded
Its more than 5%
but you can easily get the diff down to the 5% range, right? by tweaking your thread periods?
seb_kuzminsky: It can be fixed by correcting base/servo multiply
a reasonable workaround until the underlying problem gets fixed
when you add to hal file thils line : loadrt threads name1=pid-thread period1=300000 you'll see more difference
jepler, seb_kuzminsky, SWPadnos: thanks for show direction to bypass(workaround) problem, this was very importand bug I have
heh, all i did was get in the way ;-)
but i learned something too so thanks guys :-)
ah yes -- a 300us thread makes the next interval not below 1ms actually 1.2ms, giving 500/1.2=416.67 speed measured by ddt
isn't that a feature by design? (I mean - it can be made to work without much of an issue doing the right multibles?)
skunkworks_: it should either give a sane configuration or error
5773 latency - those goal boards are pretty nice.
I don't know the issue with this one.. I plugged a pci-e card in and now it seems to be running stable. (windows and ubuntu would just crash)
chester88: are you ChrisM?
heh - just puting a name with a name ;)
I am samco on cnczone.
oh ok...lol we all have so many names!
I try to stick to just one name
less bullets to dodge..lol
I put 2 and 2 togather when you mentioned you worked on classic ladder
though occasionally i'm forced to go as JeffEpler
my name is reason
Hey Jeff I was just reading the irc of a few days ago I guess just after you gave your critique ...lol
chester88: I forget exactly what I said, but please don't remind me
I didn't mean to step on your toes with stepconf
chester88: oh, I need to learn to be a little less sensitive
well i understand. when people change CL with out explaining . Sometimes ...lol But its all for the greater good.
thanks, good night all
micges: ask me tomorrow, I may have a patch concerning this issue
Jeff I have questions pertaining to my patch... where to put a 'preset' ladder folder and what to call it?
micges: in case you look at the logs, http://emergent.unpy.net/files/sandbox/sane-periods.patch
chester88: the other items I copy come from configs/common
emc.nml for instance
so I think that's "where"
"what" is probably "a name without spaces that uses the standard extension of a classicladder file"
I'm talking about a folder that holds the three sample ladder programs.
common/ladder/empty.clp or common/empty.clp both seeme equally OK to me, but the former is better if you think someday there'll be 20 instead of 3.
[Global Notice] Hi all, It would appear one of our US client servers have fallen off the edge of the discwo^H^H^H^H^H^Hinternet and all but vanished! We're looking into the issue now and hope to have it back soon. Affected users ~3K. Apologies for the inconvenience and thank you for flying freenode!
could be I guess. Ok looking in the future a bit what if we build the same idea with pyvcp. would a separate folder be the way to go or just one to hold all preselectable options?
if you're going to show a directory listing then having a subdirectory is probably the way to go
less clutter for the user to see
(though filtering on extension would help too)
but if it's 2 or 3 with radiobuttons like in your CL mockup, I don't see that a subdirectory is needed either
Well I'll start small first to see if it is workable.
MrSipan: I think people are using both sherlines (e.g., sherline 8440) and proxxon (e.g., "proxxon mf 70 cnc package") for 3-axis jewelry type work. However, I don't have direct experience with either machine.
jepler: I appreciate it. Thanks.
So the big question id my patch close enough to be committed to trunk or do you want it completely working. This is why i need advice about the directory, I can get CL to load from a directory but need one committed, if I commit the patch.
chester88: for TRUNK, stuff that's not fully working is still OK, though real soon now I'm going to declare that no more new features can go in until we release 2.3 and start on 2.4...
chester88: as soon as I find the guts to do it
ok i will commit it soon then :) before you make up your mind
chester88: good man, but that means you're committed to finishing it
heck, if you're not careful you'll become the stepconf maintainer
* archivist sends feature requests to chester88
you can manage me!
gotta go bbl
see yo uch
see you chester88
I need some help with a particular issue. I would like to synconize a joints velocity with the curent velocity of the tool tip. I'm not sure what signals I should used to make this happen.
for xyz cartesian machines you can compute the tool tip velocity by taking the ddt of each axis and then taking sqrt(dx*dx + dy*dy + dz*dz). you can see this in several of the "sim" configurations, where a signal called XYZvel is computed
for a noncartesian machine or a machine with rotational axes you would have to do this yourself (i.e., in a custom HAL component). you can use the axis.0.joint-pos-cmd and so forth as inputs to your component
once you have the velocity number you can do various things with it, like feed it into a stepgen in velocity-control mode; or feed it into a PID command with an encoder.velocity-fb as the PID feedback value
I sort of figured thats is how it might be done. So I'm running an all servo system, 7i43, where would I impliment the tool tip calculation? right in the hal file where the PID is setup?
yes, it'll go in a hal file
have a look at core_sim.hal (in /etc/emc2/sample-configs/sim) to see how Xvel, Yvel, and Zvel are calculated by ddts, and then XYvel and XYZvel by hypots
there are related "loadrt"s and "addf"s above
Great, I'll have to try to make that happend in the next couple of weeks. Thanks for your help. (by the way I'm looking at the sim hal files and I think I understand now)
can you tell me a little bit more about what you're doing? I'm curious..
You might be aware of the EMC RepStap group that is using EMC to do rapid prototyping. I'm working on a extruder design that uses a servo motor and extrudes hot plastic at the rate the tool tip is moving.
or, i guess: hot!
Hey I didn't know that-Thought that would be cool!
geo01005: ah, that makes sense.
there is a motion.current-vel output pin. it will give a better velocity output than a bunch of ddt blocks.
cradek: oh hey, thanks for being mr. obvious
you're welcome, caller
geo01005: you should listen to cradek, not to me.
I don't think I have a motion.current-vel output pin, is there?
Jeff : Question about trunk stepconf. when I run it and display parallel port screen, it prints a lot of text about pins to the terminal and doesn't have defaults for output pins. Is this right for trunk or did I screw something up?
chester88: yeah, I think I left some debugging stuff in there. if you can locate it, feel free to rip it out..
ok Thanks. I will look
oh, looks like in 2.2 it's a parameter not a pin, so you can't hook it to a signal
in trunk it's a pin.
geo01005: you reprap folks like to live on the edge anyway, so just go to Trunk ;-)
geo1005 do you have a website about the work with EMC and Restap?
There is not a specific site, if you go to www.reprap.org you will find information about the repstrap project
It's ok, I'm already running Trunk for the 7i43 card to work right
the 7i43 should work fine in 2.2.8, no?
I couldn't get the bit file to load to the FPGA correctly without changing to the Trunk.
geo1005 yes been there. No mention of EMC use that I saw. How will you create G code for objects?
geo01005: that's very odd... it didnt work in 2.2.8, but it did work in trunk?
the 7i43 driver in those two is nearly identical
[23:27:49] <geo01005> http://dev.www.reprap.org/bin/view/Main/EmcRepStrap
There is a work in progress called Skeinforge that creates the g-code.
I'm not sure what the problem was in 2.2.8. I had to load the FPGA config onto the epprom from my windows box. When I changed to the trunk version it just started working right...
Well Got to run.. Thanks for the help
see you geo01005