anyone know how to set a default spindle speed? so when you run a gcode M3 turns on spindle even when no Sxxx was set.
you could probably put an S word in the startup g-codes in the ini file
not sure of the details or syntax, but I know you can specify codes to be executed on startup
in the ini file, RS274NGC_STARTUP_CODES =
oops - RS274NGC_STARTUP_CODE
in the EMC section
thanks. that sounds promising.
one of the machines at work doesn't do T codes properly
and it's the one i am using
it has a tool magazine, not an umbrella
but it won't seek independantly
so it will sit on that line for like 30 seconds before it does anything else, no matter that the tool has changed and life can continue
also i wish i could rapid the A axis while other axes moved
that would also save time, but axis unclamp doesn't work that way =(
sounds like a plc timer, instead of using a sensor to end that step of the tool change.
I am new to EMC2, I want to run a bridge port (with analog brushed DC Servo & drives), Is it possible?
what hardware is preffered, or parport can do it?
to use analog servos you need something that has D/A converters, and you probably need hardware to count the encoder pulses
there are several supported options
what drives does it have now?
you need feedback (scales or rot encoders) and amplifiers what do you have?
It has Heidenhein control 155, and its gone now, so motors are from SEM (Small Electric Motors Ltd.)
so no drives?
tomp: it isn't, it starts immediately after the magazine finds the next tool
Analog drives are there and mototrs run if applied analog voltages
toastydeath: whats the 30 secs always?
that's just what it feels like
because it's a big tool magazine and it takes forever to seek the right tool
so it'll swap out tool 2 for the next operation, then go seeking 3
ah, its "slow"
but instead of machining while it seeks, it waits
none of our other magazine machines do this
they all machine while the magazine is seeking
toastydeath: does it get separate M6 and then Tn commands?
err. maybe vs:vs
what we've been doing is putting the Tx on the longest line nearby
so that it seeks while something else is going on
Guest571: you still need encoders, s\what plugs into the 155?
The motors have rotory encoders and Tachogenerators, Tacho goes to Drive Card and encoders to Controls.
toastydeath: "pre-waiting" :)
Guest571: and are the encoders heidenhain? thats >may< be special
whee, i got a 2nd monitor and can goto Hardy with spare partitions
Tomp: Yes, they have plate Dr. Johannes Heidenhain.
hmm, does it say "11uA" or "1v" ?
Tomp: seems to be differential A, B and Z pulses.
all have A B Z, but if '11ua" then it is sinusoidal ( not TTL at all ) if '1V p/p" then >might be ttl, handy with a scope?
any other markings ( i use heidenhains )
all will have a /a b /b z /z
Tomp: wait I open the cover and see inside.
i'll look for a 155 manual ( awallin around? )
Tomp: No other markings. The original control has a Card that is working so I can get TTL pulses.
is this a control or a Digital ReadOut ?
It is a control
in any case, TTL is good, you can use Jon Elson's Paralele Port control for the analog command to the amplifiers and it can read the waveforms from the encoders. You can also use solutions from Mesa and others.
and if you ever need manuals... http://www.heidenhain.com/index.php?WCMSGroup_2243_177=846&WCMSGroup_2238_177=2243&WCMSGroup_846_177=565
It means pwmgen and RC circuit to convert it to analog?
I think Jon Elson's does that, yes some simple filtering to turn pwn into analog OR you can build fancier but you have less help
the Mesa has it done in hardware on an add-on board
There's more options for you to study, been to the wiki?
[00:50:00] <tomp> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Supported_Hardware
I have set it up, to generate pwm pulses on parport and converting them to Analog value,
and encoder feed back to parpot and counting in the software, this setup is working.
but there is a problem
sounds like you're3 done :) what are you loking for?
At slower feed rates (50 mm/min) the sizes are ok. but if we increase the feed rate to 200 mm/min, the sizes are less
could you mobe that fast with the 155 when it worked?
(2 ipm vs 8 ipm)
yes, with 155 it can go upto 8,000 mm/min rapid and cutting what I used is 3000 mm/min
ok, now with a scope can you watch the inputs at those 2 speeds? ( maybe they breakup past your interface?)
Inputs of what?
the parallel port
do you do any conditioning?
for encoders or pwm?
... " 200 mm/min, the sizes are less" i think we need to look at the encoder inputs to the parallel port
"conditioning" meaning, i thought these were differential...
The conditioning may be in the card (that I used from older control)
No, it is 6x6 inch card to which encoders are attached and then TTL outputs come out.
Sigh, looks like 8.04 is no better than 6.06, still doesn't recognize my wireless dongle :(. Looks like I am going to have to figure out how to get a RTK for 7.10 or get used to lugging it back and forth, sigh....
in any case you can separate the counting issue by moving the reader head without driving the motor, just debug the ability to read at 3m/min
debug the ability?
Guest571: yes, make sure the output of the 6x6 card is good at 3m/min without moving the motor by pwm ( separate the issues)
Ok I will check that when I will be at workshop.
this parallel port with encoder and pwm, was it controlled by EMC? (hal?)
or your own custom software?
you can make sure the pwm section works by itself by sending pwm to the filter to the amplifier and measuring the tacho output
(dont connect the motor to the axis for this test )
more info than Heidenhain has on the 155 http://faculty.etsu.edu/hemphill/entc3710/heid-op/h-chp1.htm
Tomp: sorry elecricity failure.
I am using EMC Hal.
excellent! with Hal you can easily separate the issues and test. especially since you are already using Hal.
Can parport be used for 3m/min
good question, I dont know, it depends on your hardware AND i still wouldnt get you the best answer,
you'll need to get that answer from one of the guru's
hardware latency test show 18000 ns so I have servo thread of 25000 ns
sorry base thread 25000 ns
whats the smallest unit of measure with your scales ( .01mm .001mm .001" ?) because i think you get 40K counts/sec with that thread (1sec /.000025) and 40000 x unit change /60 is top distance/minute possible ( again, ask a guru ;)
it is 0.001 mm
software encoder counting with a 25000nS period in theory will fail at 40000 counts per second, in practice you should stay below 30000 or so
with 1 micron resolution, that isn't very fast
I be happy if I can get 1500 mm/min
40mm/sec * 60 secs is 2400mm/min???
I am from Pakistan, so it is difficult to get one of the supported hardware.
yes 2.4 m.min theoratically and some margin so 1500/min is ok
"lower your expectations" :) and that still doesnt say how fast the parallel port can resolve that stream. your problem is even at 200mm/min = 3.333mm/sec = 3333 units/sec. so look at the hardware with a scope.
if you have a pulse generator, test your paralel port withpout the 6x6 board
Ok, I will check that too.
where in Pakistan?
your english is perfect
our medium of education is english
you can also verify the motor rpm by reading the tacho (it is some V/krpm)
I have an idea, to make quad encoder counter card (24 bit) and then read it in EMC
need to write some driver
hurray! the checksum is good, now to burn it
Guest ---> shabbir
Guest571 is now known as shabbir
Tomp: I will be back after checking.... Thanks
I have an idea, to make quad encoder counter card (24 bit) and then read it in EMC, needs a driver?
I check the "over ride limits" box and it immediately unchecks. Got a home that is out in left field and can't seem to correct it short of rebooting?
home homes to whever it is (in this case off the table) and then enforces limits, wth?
NEver mind easier to reboot :).
K`zan: this may help clarify the intended function of "override limits": http://thread.gmane.org/gmane.linux.distributions.emc.devel/848/focus=849
when booting the Hardy Heron EMC Live cd, i first get a pop-up "There was an error starting the GNOME settings Daemon..." I had this also on last box I installd to. ignore it?
you're right that it can be difficult to recover from emc having the wrong idea of where the soft limits are, without restarting. To cure that, emc 2.3 will allow an axis to be "unhomed"; on unhomed axes, soft limits are not applied.
had problems with Gnomes because you have gremlins
anything gnome related is graphical, shouldn't hurt much to ignore
tomp: pass that along to alex, I'm sure he'd at least like to look into it
it sounds like something that mightbe ok next time , will copy the text, ok
this is live,so not next time, but after install
a "settings daemon" just sounds like a waste of memory and cpu cycles
Folks, how do I get an axis to home / zero ? Somehow I got my x out by -7.99 at the right place and nothing I am doing will let me get it to zero?!?
Override Limits immediately resets to unchecked...
especially see section 3
cradek: Will check, thanks. I had not problems getting home/zero moved until today - most confusing.
override limits is for jogging off a limit switch. sounds like that is not your problem.
No home nor limit switches yet...
G53 G0 X0Y0Z0 - does nothing...
x is still at -7.99
override limits, still won't hold.
guess I do a goo x8 and then power down and manually crank the table back.
ok tell me what you did after reading section 3
when you hit home you will get an absolute coordinate of whatever is in your ini HOME=. This is probably zero.
then if you clear your offsets you will see 0 on the screen too
this has nothing to do with override limits
hit home axis and all it does is put a target up in front of x display.
Lemme get some gloves, brb. turning a 1/4-20 for 8 inches is hard on the fingers :)
why not just jog it?
stop doing stuff and tell me what's going on, and I can help
wont jog in that direction
or, read that page about offsets
ok tell me what you're doing
if you start emc, then press the jog arrow, it should jog
cranking on the lead screw, halfway there :)
Would NOT jog in the direction I need to go...
ok stop please
when you start emc, it's not homed
if you press a jog key, it will jog
is this much not your experience? if so let's figure that out.
must be g00 x0 does nothing
ok got gloves off, can type again :)
Ok, lemme power up the drive again
In manual control...
x set to continious
is the home icon on?
Is that the target?
ok this is the third time I've said it: when you start emc it's not homed. you can jog any way you want before you home.
OK jogged to physical zero / home
ok so home is like at the left of the table and you jogged over there?
ok now you can hit the home button
now you ARE at machine zero
g0 g53 x0 will NOT move
x now moves to -7.990 rather than zero...
if the X display does not show 0, you have some offsets in effect
are you with me?
If so, I have no idea where :-/. THe dotted box has moved WAY to the left.
ok let's go back to that wiki page about coordinate systems and follow the steps to clear the various offsets
Yes, but I see no place for offsets and unless I fumble fingeres something I didn't set any.
OK, checking again - thanks!!!!!!!!
"touch off" sets an offset, maybe you used that
ok at step 3 on the wiki page, go to mdi, g53 g0 x0
the machine will not move, since you are already at absolute zero
Ah, that may be it.
G53 G0 X0Y0Z0
OK, that gets everything back to 0,0,0
the dotted box is still off, but that may be something in view!
maybe hit reload?
SO now I keep my hands off the "home axis button
reload did it!
yes you should jog to the "left" and home, then from then on, use touch-off
that way you can set soft limits to keep you from bonking into the end of travel, even though you don't have home or limit switches
not sure what touch off is yet, if it has to do with a probe, I aint got one :(.
nope, see #2.2 and #4 on that page
Thankee, will read that whole page.
touch off is the easy way to set an offset/work coordinate system
it's how you zero on a workpiece
Getting the machine dialed in slowly. Rebuilt the Z axis over the past couple days and it may be better.
ok, welcome, glad you got it figured out
Err, thanks but let credit go were due, you figured it out and lead the blind out of the darkness :)
mastering coordinate systems and tool offsets can take some time! but it's so important
Sigh, I never had to worry about that, I wrote the code, but they wouldn't let me near the machines. MUCH to learn.
Lot more to this than doing a pretty drawing and running off a tape :).
Lets see if it will draw the EMC2 AXIS thing. Got a pen refill in the chuck.
I suspect I am going to have to level up the table...
This whole thing is SO flakey.
But fun :)
[04:31:51] <K`zan> http://wrlabs.shacknet.nu/~vw/MyMachineShop/PipeDreamMill/PDM-V2/
On my desktop :)
Yeah, that is starting to look like a VERY apt description...
A Bridgeport it ain't....
the flex shaft spindle looks clever
It came that way.
An affordable Foredom :)
have you measured the table flatness yet? that might be a challenging part
OK lets move the Z down to the table and see what happens.
The table is very flat, just in which dimensions are the questions(s).
These larger steppers are a problem...
they unbalance the table. Had to Gorilla Glue some serious weights at the other ends.
I should, I guess, have gotten the floppy steppers and the single step drive I was looking at. Would have eliminated some of these problems.
floppy steppers seem good for nothing but floppy drives. I bet you did the right thing.
Well, yes, much more capable steppers (~= 60 in/oz) but also much bigger / heavier.
Here goes :).
hitting escape stops it :-)
so does the square button :)
Retracts from table contact and then never goes that far back...
ok you need to use touch off
Nothing in the EMC windows or dialogs is copyable :-(.
Lemme move it back.
you want to cut into the table?
No, got a piece of paper taped to it and a pen refill in the chuck (not turning).
ah ok. looks like the splash screen cuts 0.010 deep. this is a good number for your pen I bet.
so touch the pen to the paper and click touch-off, then accept the default of 0.0
now 0.0 is the pen on the paper. -0.010 will be "into" the paper a bit.
Ok done, smoke test #2 :)
when you use touch off, you will see the program move in the preview, with respect to the cone. this gives a good feel for where it will cut.
Why do I want a big red ESTOP switch?!?
in fact, to see where a thing will be cut, jog the cone over to it, and then look at the machine to see where that is
(do you really have 7 inches of travel??)
End to end -10"
ah good, it will fit
Y maybe 5
Z 2.5 more or less depending on how the rebuild affects things.
A bit much pressure and tore the paper once, and out side of some paper wobble I done got it :).
Pix to follow in just a sec
oh, good damn, not bad damn
There may be hope :)
Fray had to come over and check it out, I guess a nuzzle means I done good :)
or wanted to make sure you were ok after all that strange noise?
[04:53:19] <K`zan> http://wrlabs.shacknet.nu/~vw/MyMachineShop/PipeDreamMill/PDM-FirstSuccess/
I'm about 9 feet above cloud nine :).
hey that looks great!
Fray doesn't know what it is but he sure is happy for me :). Most companionable and sweet boy cat I ever met :)
Paper slipped and caught once, but NOT shabby :).
yep they can be such good friends
I never liked cats until one adopted me and proved all my previous exeperience was with utter asshole cats...
With a braincase that small, it is most interesting as to just how smart they can be.
I bet they grow up according to how they're treated
so I bet it's partly your fault he's so nice.
Yes, I always treat my "pets" as family and the rules apply to everyone and they respond just like my dogs always did for the most part.
True :-) True :), but I had good material to work with :-) :-) :-).
I think the easy part is going to be getting the hardware together.
I better get to bed... goodnight and congrats on your running machine.
Getting the CAD end together will be fun/
Okdokee, rest well and thanks VERY much for the help and direction!
I think I'll float away and revel in the success :-)
I guess I need to run through section 3 to reset that offset...
you can use machine / zero offset / G54
or, just leave it and set the next one you need later.
Thanks, just ran that.
just remember to jog "left" before you home
g0 xo y0 :)
just eyeball an origin point
yeah, g0 g53 x0 y0 z0 before you shutdown will save you some trouble.
then you can home right away next time.
copied that into my notes :). Thanks.
welcome, goodnight again :-)
cradek: Were you the one who sent me those caps?
Save it for later, go get some rest.
cradek: or you can use the POSITION_FILE in the ini, that saves the position on shutdown
when I run ./configure --enable-run-in-place --enable-build-documentation
it gets to this line "checking for dvipng... none"
and gives me this error
configure: error: no dvipng, HTML documentation cannot be built
do I need to change something on this line
sudo apt-get install build-essential
did you run: "sudo apt-get build-dep emc2" ?
it works fine without the --enable-build-documentation
Sounds like it's just missing from the depends list. I set up for building docs a good while ago, but don't have dvipng on my machine - guess it's a new requirement.
what is the depends list?
i maybe got a job running bigass wood router CNC
hahah guy was like 'hey you dont have a problem like, putting the wood in, and like, starting the machine, and stuff, right?'
like he had interviewed some pro draftsmen or engineers before me
BigJohnT: I'm assuming there's an input file for the "build-dep emc2" and called it the "depends list"
thanks I'll look
i got one more pcb job to do then im going to try ubuntu 8 emc
hopefully new kernel and autodetect know wtf is going on with my hardwrae =\
hardwrae (c) rencorp, all of the times
BigJohnT: I googled dvipng - there are hits at sourceforge if you want to install it manually for now.
is loop code eccentric in anyway in emc?
um, m98 and m99 i think
I did a search on my synaptic package manager and it show up there, thanks for the help
well it made it past dvipng to the next line and can't find convert... at least I have a clue as to what to do now LOL
BigJohnT: try apt-cache search dvipng
that should tell you the package name where it's included
(but maybe you don't have universe turned on, so some stuff didn't install..)
is this for a 2.2.5 checkout? or for a recent CVS?
yes universe is on
tex4ht - LaTeX and TeX for Hypertext (HTML)
dvipng - convert PNG graphics from DVI files
apt-get install dvipng ?
ok, that's why build-dep doesn't fully work
I installed with with synaptic packager manager
(the build-dep is for an installed package: 2.2.5 in your case)
and the build dependencies have changed for pre-2.3
you can look in debian/control for a list
(if you care..)
in my emc2-trunk folder?
yeah, that one
hmm.. I don't see it there
what platform are you on?
what am I looking for in there?
I was actually missing 3 things dvipnp, convert, and one I forgot
I found convert was part of imagemagick AFAIK
ah, I remembered
build-dep only gets you stuff needed for building the package
(which only contains pdf docs)
the extra stuff you're installing now is for html docs
(you can also use ./configure --enable-run-in-place --enable-build-documentation=pdf)
or something like that
ok, it seems like it worked, but just the pdf's would be ok....
Oh Cool! I just noticed an index has been added to the pdf manuals
BigJohnT: Will you be coming to the CNC Workshop?
I'm not sure at this point. We finished one automatic yesterday and have 3 more (these are the same) automatics to have done by July 1
We have most of the parts started and I have 99% of the engineering done so it's a toss up at this point
It would be nice to meet you. I like what you did with the g-code generating python scripts, so I hope you get done in time!
alex_joni: I did get one error man9/pluto_servo.9:106: warning [p 144, 4.3i]: can't break line
mshaver: it would be nice one day to roll that work into a real cam program that can import dxf files etc...
actually 14 errrors on the same page...
That's my hope as well. One of my biggest problems is having a CAM solution I can give to my clients. None of the commercial stuff, even disregarding price, really suits me. Too wierd of a GUI, too high a learning curve, bad g-code output, poor support, etc...
and most people's requirements are simple 2.5D, rectilinear type parts!
SheetCam is not bad and pretty cheap too. I agree with you on that most are not intutive to use
yep, that's probably one of the best ones.
gcnccam is at least on linux but it's very young in it's development
Well, it's out to the shop with me! Lots of things to scrap & blow up today ;) !
mine is version 0.4.4
true, there's a lot of promising projects. It's hard to decide which star to hitch your wagon to though... Back later, keep saying interesting things so I have something to read when I come in the office for coffee! :)
I have to shut this computer down for a bit so I'll ttul
* dimas_ just finished soldering an interface board
So, has everyone updated their keys yet? LOL
alex_joni: Have them ISO's been updated?
JymmmEMC: the correct way to address the ssh issue is to use the normal update procedure.
jepler: you still have to destroy ALL existing keys and regenrate them.
JymmmEMC: what's your point?
jepler: Has that been done?
JymmmEMC: when you install a fixed openssh-server package, the host key is regenerated if it was a vulnerable key.
if you used key-based authentication, you have to manually go remove that compromised key from any .ssh/authorized_keys files where you put it
jepler: That can't be, then anyone who uses keys only to get into a box, would never be able to get in. The keys have to be regenerated manually
I've already ran the updates, and don't use keys only, and theys are not changed automagically.
what do yo mean by "fixed" anyway?
JymmmEMC: if you put a compromised key in the ~/.ssh/authorized_keys file of any machine, you have to manually remove it yourself
JymmmEMC: by "fixed", I mean the openssh* packages that Ubuntu/Canonical released in order to fix the vulnerable key generation problem.
jepler: Right, put many ppl aren;t aware of that, and barely even read the "are you sure you want to add this key" dialog =)
...when connecting to a new box
this is the fragment of the 'postinst' script which is intended to automatically cause the host key to be regenerated: http://pastebin.ca/1021082
Just be glad that this didn't happen to GPG, we'd be fucked if that was the case.
In just 30 minutes, we found over 5000 duplicate keys
jepler: Maybe I missed it, but do the ISO have the vulnerability or have they been patched already?
JymmmEMC: as far as I know, the 8.04 live cd was made before the updated packages were released. we do not reissue the cd just because updates (even security updates) are released.
that is, the emc2 8.04 live cd
(as far as I know, canonical/ubuntu also don't reissue cd images just because security updates are released)
I just looked, they don't even have mention of it on the front page or release notes. I don't feel that's right. Especially due to the impact this vulnerability allows.
alex_joni: messages seen at boot time with 8.04/emc2 live cd http://pastebin.ca/1021098
JymmmEMC: is this your page? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TestedHardware.
could you post your working emc system?
tomp: It was an attempt to, yes.
could you post your system to it? I'll prep mine, but they're bad news
What, you no like my Acme system?
JymmmEMC: the following is not pointed to you:
I've installed the live CD to 2 systems now, both get overruns,
and I've gone thru the troubleshooting page, turned off APM ACPI USB OnBoard video...
I installed 6.06 to 4 machines, and all had overruns.
I has not been running the latency tests previously, so some of the results are back-checking.
Therefore I think that mobos that run EMC2 w/o overruns are not common.
Please post any better results so users dont waste time and money on stuff that wont work.
the top live cd refers to 8.04 with emc
Ok, so why not add those systems to that page?
Tried that yesterday with the hope it would deal with my wireless dongle. No luck. Interestingly the stock ubuntu 7.10 does. LOL, sigh.
fine, a page full of bad new coming right up
tomp: The top of the page DOES say "good, bad, and ugly"
LOL, appropos :-)
tomp, I've seen that problem "every once in a while", on 6.06 and 7.10 systems
(the "error starting settings daemon" thing)
on both EMC2 and non-EMC2 installs
* JymmmEMC dictionaries SWPadnos systems with the 32K dict file =)
* SWPadnos gets out the 64k dict file
my main concern is a system that works, i want some confidence that it is possible
SWDamn dude, you need to get some QUALITY p0rn in there
the quality pr0n is on a non-net-connected server
tomp: The reason for getting bad results too is 1) So ppl are aware of it, and 2) It could be a BIOS bug as well.
SWPadnos: Share the wealth, QUALITY p0rn is a rare thing.
you'd have to sign an NDA
time to start hacking
anyone want to post one viable system on that page while i prep the text about the ones that do not work?
SWPadnos: been there, done that.
Picked up a 1F/16V cap at the thrift store yesterday :). $5.00 :)
A little research shows it is supposed to be for "stiffening" car audio power (lights dim on the bass passages).
Personally I''d like to see RPGs for those folks, can hear them 3-4 blocks away...
Seems like the 16 volt might be an issue with an alternator.
yep. check eBay for things like fuse holders and high current distribution blocks, and they're almost all for high power car stereos
2 more and it might do wonders for the drive power supply :0
don't get too large a cap
that will destroy the transformer
Mostly I got it because I always just wanted a 1F cap :)
Silly me :)
Built one years ago that occupied several boxes using individual caps - not practical...
This is about the size of a TALL beer can.
wow - DigiKey sells 5000F caps
(only 2.7V, but still)
:-(, figures,finally reach a goal and it isn't even significant anymore :)
RUn a LED for the rest of your life on 5KF :)
Might do that. Other than final gratification, I have no idea what to do with it. I will NOT have a car stereo that need it :)
okay. That's the last of the rm -rf .ssh
Now to see if I can get back into any of the other boxes.
hmmm. Q=CV, i=dQ/dt=20mA=C dV/dt, so every second it loses 0.02/5000V = 4 microvolts
assuming a 1.7V LED (for simplicity), that gives 1V loss before significant dimming, so 1.0.000004 = 250000 seconds
which is only 3 days
I wonder if I did something wrong. that sounds awfully low :)
I'll enjoy looking at it, I guess.
err - that should have been 1/0.000004
just don't charge it up then drop a wrench across it :)
Could use it for a spot welder I suppose....
internal resistance may limit peak current to spoil the fun
OK, I think we are down to having a 1F cap on the shelf :)
Connectors it comes with are scary enough.
hi, I'm experimenting with the gtksourceview-2.0 functions that come with ubuntu 8.04, http://imagebin.org/18234
cool. I think this was with rev 1.0, so maybe you could help update it :) http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Highlighting_In_Gedit
my question, is the cfg-string obligated with the hal_parport component, or does it use a default base adress :)
yes, i wrote that wiki page
oh - ok :)
I recall helping someone on IRC to get the regexes correct :)
(but I could havebeen remembering something else)
indeed that was me
I think the parports are scanned in the order 0c3bc,0x378, 0x278 if no cfg parameter is specified
I guess I could look at the source :)
oh, nope. cfg is required now :)
gtksourceview2 has an "extended" option, so the regexp ignores spaces :)
Is it impossible to get some integer data from Virtual panels?
maybe. I'm not sure if the slider/knob controls have s32 outputs or only float outputs
heh - you asked if it's impossible. I said no, meaning that it is possible :)
maybe I hate integers ;)
but I'm not positive now :)
tomp: Happy now? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TestedHardware
thanks for any effort, dodnt look yet, still wrtiing
the hardest part will be adding al the emc components, functions and pins to the .lang file :)
JymmmEMC: thats nice, thanks, thats a micro board like rayh's, maybe i shouldnt try these normal desktop
assargadon, the "Scale" control (a slider) has both a float and an integer output
JanVanGilsen, I think the loadrt word should still be hilighted on those "should not pass" lines
but not the module name
I have found over time that different via MiniITX boards produce quite different latencies.
I'll look intoo that =)
rayh: feel free to add your results to that page
I wonder if we should expand the board name to include models.
also, if it's possible to have a couple of colors, it might be good to either hilight the instance number differently (parport.0 vs parport.1) or hilight the whole pin name differently for each instance
hmmm. option 2 would get pretty confusing, so maybe just the numbers inside pin names :)
okey that was easy, just adding a ?
rayh: thanks, i found those boards. your info is handy for users.
SWPadnos, i dont know how I'd do that
yeah, me either :)
the TestedHardware page should probably be merged with http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Latency-Test
is there a man-page on the net command?
halcmd help net
or man halcmd
(hopefully at least one of htose places has the info)
Have you guys seen the hardware testing menu item in 8.04. I wonder if we could produce a similar thing for emc2
how IO pins can be used at all? As far as I understand, every "net" should have exactly 1 input...
loadrt at_pid num_chan=num [debug=dbg] , on the manpage it states "If numchan is not specified, the default value is three", shouldn't that be loadrt at_pid [num_chan=num] [debug=dbg]?
JymmmEMC: 1 entry added, will get the other systems next
tomp: If you're going to included video driver, you need to add the version of it too.
tomp: Also, try to keep the same syntax, as then it be easier to parse out into CSV if ever needed.
version wasnt reported in xorg dump, will look
i'll keep same syntax, please give example, didnt understand
tomp: Cpu info, it's one big entry, not comma seperated.
gotch, no commas
example: Pentium4, 800MHz, 533MHz,512K
you want commas, ok
While the board may be capacble of 400/533/800MHz, would rahter know what you are currently running it at.
cpu freq != bus speed
right, i said cpu was 800, i said bus was 400/533/800
But what is it running at?
it's not running at ALL those speeds at the same time.
i dont know how to tell, could you tell me?
afraid not, maybe your manual could.
will reboot and see if it says anything, then re-edit, thx
If you don't know just leave it blank, or better yet, just add an extra comma, like Pentium 4, 800Mhz,,512K
memeory freq is at DDR333, does that mean 333mhz bus?
DDR333 actually means 166x2 (data is clocked on both edges of the clock)
i only get a few hundred ms at first text splash screen, re-trying now
dram clock is 166, yes, host/pci is 100/33Mhz , is 166 the 'bus freq' ?
100 I think
the I/O bus is running at 100 I'd think
tomp: dont sweat it, I revised your entry alredy
so, the column header should be host frequency?
tomp: Just FYI, *ANY* overrun at all, is bad.
well, time to look at the big outdoors (before it starts raining)
tomp: Techincally FSB Freq
tomp: but abbr to Bus for brecity
i got that all overruns mean no emc, yours doesnt work, mine doesnt work, i'm looking for one that does as soon as i get these reports out
I tested my realtime port of my controller driver
and I got a strange errors
it looks like SOMETIMES other data became written to LPT port
not one I put there
I read man about rtapi_outb, and rtapi_request_region listed there
1) is error I talking about can be caused by non-using of rtapi_request_region ?
2) why my code still works without rtapi_request_region ? I thinking it's something similar to io_perm
* assargadon reading rtapi_request_region's man
SWPadnos: thx for the entry, will look for that board
can't find error... I just replaced "normal" calls to "rtapi" ones....
if all drivers use rtapi_request_region (emc rtai) request_region (linux), then you can be confident that only one driver is writing to the I/O region at a time. If at least one does not, then you can't.
(a userspace program which uses iopl(2) or ioperm(2) could also interfere)
jepler, yes, I just read man's about it...
but I still do not understand
I making experiments, thought :)
As always, you catch the error yourself as fast as you ask for help :)
needed delay was 1 millisecond
and thread period was 50 microseconds
those numbers in different files, so it's hard to catch :)
tomp: Are you looking for a system that just works?
tomp: If so, this might do the trick... http://www.weirdstuff.com/cgi-bin/item/62008
I might be able to confirm this next week.
now I should invent good, structured and maintanable way to make "timer-like multiloops" delays instead of rtapi_delay calls :)
JymmmEMC: thanks, yes i am just looking for something that works
tomp: I have some RackMount (tm) boxes at work, but I haven't tested them.
LOUDER THAN A MOBO though.
i have 12" fans 120V~ in some of mine, doesnt bother me after working around presses and big mills :)
i just asked a reseller in Dallas if he'd test a system with the live cd& the latency test, I would pay for testing in anycase.
SWPapadnos, this is what I've go so far: http://pastebin.com/m56d11496
I have a question about steplen and stepspace
Does one know why stepconf is using 0 and 1 for steplen and stepspace?
stepconf uses the "doublestep" approach to step generation, to allow a higher step rate
that means steplen is set to zero
I think i have short pulses while using high acceleraton and speed near maxfreq.
your pulses should all be the same length
how wide is the pulse in this case?
that's set by parport.0.reset-time (or similar)
the value used in stepconf is 5000 ns
is that enough for your drives?
2500 shold be enough
then you can set it to 2500 in stepconf
moment, i check parport.0.reset-time
the default is for Geckodrives, I think
I use smc-servo from elm
hmmm. maybe not, but you can change the values :)
setp parport.0.reset-time 9000
that shold reach
now i know why this config works, but the last not
you want the reset-time to be as short as possible, while still running the drives properly
sw: is parport.0.reset-time also used for stepspace?
JymmmEMC: this is very close to same hdwr as that Tyan S2518 , its a Tyan S2510 but only 19$ less cpu less mem... http://www.weirdstuff.com/cgi-bin/item/15184
mean this the signal is set while stepgen is called and reset after parport.0.reset-time
yes. if stepgen says it needs a step, then its output will be set. the next interrupt, the parport pins are written at the start of the thread, then reset later by the parport.0.reset function
with steplen 0, stepgen knows that it can request a step every interrupt, rather than every second interrupt
thanks. I will remove the sikefilter from my controller. then it shold work
heh - good luck
I think its no problen, now i know for what i mast look
basically, in stepconf the step "0" timing is justified based on choosing BASE_PERIOD so that (BASE_PERIOD - latency - steplen - 5000) > stepspace
all worked well if acceleration is under 300 with speed 20mm/s
I found the latency test once, but for the life of me I can't find it now :-(, help please?!?
open a terminal, type latency-test
SWPadnos: Thanks :-) <blush>
[19:32:00] <JymmmEMC> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting#RTAI_Latency_test
see you all later
Thanks es later!
servo is modified. go testing
I'm really confused about the home/limit switches. Must have to do with the limited number of inputs on the parallel port.
I suppose one must create something external to monitor them and send that through the driver to EMC.
Seems that one would want them separate anyway...
Parport is why we piled them all in together.
Trying to figure out what to do for a test next :).
OK, that does make sense if seemingly "chancy".
* K`zan_emc is VERY ignorant of a LOT of this at this point.
You certainly loose the info about which axis and which end that way.
First ?successful? run of the Pipedream last night.
Fill me in on the Pipedream?
printer ports are cheap...
So it would seem, no way to tell which was which. Home seems to make sense and let EMC deal with the limits?!?
Ah, multiple parallel port cards.
[19:49:35] <K`zan> http://wrlabs.shacknet.nu/~vw/MyMachineShop/PipeDreamMill/
K`zan, yes, multiple LPTs
special interface cards with special drivers
Had to go to the desktop to paste that.
Thanks, now without my spikefilter all works well
assargadon: I can see some exciting construction to fully implement that.
"something external", as you told abvout
* K`zan_emc lusts for a big red estop button first though :)
nice job. good to hear it's working.
Thanks, but it is certainly going to need adjusting, of that I am sure. Last pass took care of a lot of it.
Might work out now if I am not planning on doing work for NASA :)
Makes you feel great when it begins to work like it should.
Need to go get a piece of PCB and make a square on it to see how level everything actually is.
And for that I have to learn to write enough G-code or find a toolchain that will do it for me.
I knew I was right to presume that the hardware would be the easy part.
now i must use an hardwarespikefilter
gefink, Where you thinking of putting it in the circuit?
rayh: its a atmega- interruptpin, think about c+ schmitt-trigger
Which value do I want to plug into stepconf from the latency test? Higher of the two or servo or base thread?!?
or better replace hcpl 2631 with hcpl2231 which includes trigger.
None of anything to do with EMC - window/dialog is copyable, arghhhhhhhhh
Ah okay. I see what you are aiming at gefink.
give me the link to good site about rtapi, please
[20:08:49] <SWPadnos> http://www.linuxcnc.org/
I trying google, but somewhy first page fulfilled with not-very-usefull links
RTAPI was written by one of the original EMC developers at NIST, and I think EMC is the only thing outside of NIST that uses it
servo thread shows 51061 so I'll use that one, been abusing the box for an hour now. LONG way from 15000 default :-/.
is there not difference between rtai ant rtapi
Seem that this HP 2K celeron should do better than it does. Compared to the 2.2G AMD it is a slug.
servo thread jitter, yes.
I presume the bold values are what one wants and the highest of the two.
wiki isn't real informative....
program reports jitter, stepconf wants latency.
Perhaps trivial to folks with a clue :).
SWPadnos, you are wrong about "only EMC uses RTAPI"
It's much easier to implement realime delays then I thought :)
I just should use rtapi_wait :)
and some kind of time measuremeNT
quite easy, quite simple ;)
Guest604 is now known as skunkworks
( AXIS "splash g-code" )
Is that just a comment of an AXIS directive?
joint 0 following error ?!?
in g-code, things in () are comments
Perhaps running it with the drive powered down is the problem, although I can't imagine how it would detect a feedback problem.
THanks, wasn't sure if stuff could be embedded in comments (like #!/bin/sh) just wanted to be sure.
this is steppers?
if steppers, powering down the drives won't make a difference
do you have backlash comp enabled (a non-zero value for backlash in your ini file)?
I can't imagine the gcode is a problem.
Not that I know of. Still setting things up here.
did you create your config using stepconf, or manually
post your ini file on pastebin.ca so I can take a look
[21:19:38] <K`zan_emc> http://www.pastebin.ca/1021324
And the gcode - my first in the last 15 years :):
[21:21:03] <K`zan_emc> http://www.pastebin.ca/1021325
FWIW, last night I ran the EMC2 AXIS demo file with no problems...
you don't have any margin on your max velocity
65000nS period means a maximum of 15384 steps per second
lemme start stepconf
your scale is 16000 steps per inch, so your max speed is 15384/16000 = 0.961
stepconf will overwrite your ini file
yes, that is all that has ever accessed it, to date...
Done it all with stepconf (warnings made me paranoid about messing with it)
I don't use stepconf, and I don't know how it works - but the numbers in this ini file are no good
max speed relates to IPM ?
* K`zan_emc is very ignorant at this point...
K`zan_emc: can you pastebin the stepconf file too?
[AXIS_n]MAX_VELOCITY is the speed it will try to travel for rapids - in inches per second
should be using F1 rather than the F10?
just a sec
no, F10 is fine - that is inches per minute
all speeds in g-code are in units per minute
but ini file and hal numbers are inches per second
(or mm per second)
in the present released version of stepconf, it will write bad .ini files if you enter a step rate higher than you can do based on your timings that you enter on the second (third?) page. Did you enter a really large value for jitter/latency, for step space, or for step length?
(the next version is supposed to automatically lower the speed you enter until it is possible to reach it)
[21:27:25] <K`zan_emc> http://www.pastebin.ca/1021331
I entered the high value for jitter out of the latency test which was 51061 (it rounds to 50000).
you entered 50000ns as the latency -- that has a very big impact on the maximum number of steps per second you can achieve <property name="latency" type="float" value="50000.0"/>
It seemed to work fine with the 15000 default, but I was trying to do this "right" :-)
I mucked about for about an hour to get that value in stepconf, seriously abusing the box.
How about I just go back to the default and let this go - not a clue as to what it really means other than to screw up the process :-/.
OK, you have several choices that should each get rid of the "following error": change jitter to a lower number than you actually saw, or change the max speed to be lower.
given the small size of your machine, you could probably lower max speed and still be happy
if you put a lower number than your system's actual jitter, then there's a chance that the timings of the step waveform will not be in spec for your steppers
As noted, it seems to run OK with the default...
my shoptask only does 48 IPM = 0.8 inches per second
the next version of stepconf/emc will detect this specific problem and automatically lower the speed (e.g., from 1.0 to 0.8)
(or whatever is within the limits according to the values you enter)
OK, so go back anu put in the 50K value and then muck with the individual axis values until it works?
K`zan_emc: changing to the default is not a good idea - sure, it will let you get to 60 ipm, but when you inevitably get a 30000 or 40000 or 50000nS jitter event (rare) it might make you lose steps
don't "muck with" them
do the math
65000nS = 15384 steps per second, and at 16000 steps per inch, your max speed is 0.9something inches per second
so tell stepconf you want 0.8 inches per second to give yourself some margin
Hummm, is that Max Velocity or Max Acceleration we are talking about here?
Velocity I think (in/s) acceleration is in/s2
Done, thank you sir (from 1.0 to 0.8)
firing emc back up
Wonder what the green line with the X on it means?
about 1/3 the way down the first X move.
Works! Thanks folks!
Now do I get bold and track down a cutter and a piece of PCB or cower in the corner :)
I'd vote for both. Set it up, start it up, cower in the corner.
with safety glasses on
LOL, good plan.
Looking to see what thickness the cladding on PCBs actually is - I question my z -0.01 value.
Of course, google is toast here ATM :)
you mean the copper?
I think 1oz copper is 0.0014
Corresponds to what I found, thanks:
1/2 oz. 0.7 mils
1 oz. 1.4 mils
2 oz. 2.8 mils
S0 0.01 is probably overkill :)
this will be a demanding test of the flatness and squareness of your table
Want to get an idea of how level the table is - affixing the PCB with double sticky tape...
jmkasunich: Yes, I know <shudder> I expect to be disappointed but it will let me know what to do next.
put a dish of water on it, the ring edge shows you level... the squareness of the quill motion to that surface is harder to inspect, but a bent piece of wire in the chuck swung over the surface in a large arc will say a lot
or get an indicator and 123 blocks ;)
Level isn't as important as squareness of motion.
I'd put an indicator in the spindle, run it down so you've got room to move up or down, and then move in x and y and read the indicator.
if you don't have an indicator, use a felt point marker - adjust it so it just touches, then run it around and see if the line gets fatter (table hi) or disappears (table low)
moving the part under the felt tip shows parallelism of surface to x and or y way, it would show the same result if the quill was 45degrees to the surface.
tomp3: if you are milling a PCB, parallelism is what you care about (consistent depth of cut)
having the cutter at an angle to the surface has little effect when the DOC is measured in single digit thousandths
we must be speaking of different things, when my spindle isnt square to the table, a normal square bottomed end mill will dig on one side. If K`zan wants a machine that works well, squaring the spindle to the table motion is a good thing.
yes parallelism is neccesary before squareness, and linearity before parallelism, i see i was too far into the geometry
Comcast has deemed to come back up!
First cutting pass went reasonably well but there are still problems:
[23:48:55] <K`zan> http://wrlabs.shacknet.nu/~vw/MyMachineShop/PipeDreamMill/PDM-FirstCut/
hi anyone home?