toastyde1th is now known as toastydeath
[02:04:49] <skunkworksemc> http://imagebin.ca/view/zqFQG7q.html
finally got it into a case
nice latency numbers
what hardware are you running those numbers on?
it's an ecs goal3
cheap too, about $80 for motherboard, CPU, and 512M memory
price went back down
Seems cheaper than the old work computers Ii am messing with )
I don't see the one with memory at the moment
only the Sempron+MB for $40 after rebate
must have sold out finally
($50 before rebate)
again. I think they're still making them
you and skunky pimping them full-time
I've seen them go in and out of stock several times
I only bought two - he's the hog
it's price sensitive, drop the price $5, and the horde strikes
this is an pci-e video card.
are they selling them with the heatsink yet?
skunkworksemc: seem to be OOS
(not using the onboard video)
the combos I bought had CPU fans
atom is back in stock
and there are many many fanless PCIe video cards also, if that was your question :)
most of the USB ports aren't high speed
the goal3 w/ memory they have had didn't have a hsf
(on the D945GCLF2 or whatever)
that's because you are special
yes, we know that
that atom would look good with a mesa 5i20 in it
although hdmi would be nice
no - random - I meant that I am using a pci-e video card with the goal3 motherboard. ... and when I (and SWPadnos) bought ours they came with processor fans which they haven't done lately
I seem to be having a problem getting my point across (all day) :)
that's because the rest of us aren't special, we get it, ok?
I am special - my mother told me so.
you don't wanna know what grandma unterhausen used to say
cradek: this one's for you: http://hackaday.com/2009/02/01/generating-g-code-with-common-lisp/
now, that is a hack
I wrote a similar thing in order to eliminate redundant repeated words for the very memory-limited BOSS machine
I see at the bottom he says he does offset curves
I'll have to check back to find out what he did about that
hooking my oscilloscope up to my running cnc make the drive run out, is this normal?
or is my lpt kicking a signal out to my encoder causing the problem?
I hooked pin 11 & 12 of the lpt to the encoder on my drive,
UncleG: maybe you're injecting interference when hooking up the probe?
that would be my suspicion, however my question is, is it that expected?
or is my lpt being problematic
I've certainly seen similar things happen, but generally only with lines that are not well grounded, terminated, and filtered
I would like to test the encoder with the scope on EMC however I cannot since the drive is kicking out when the encoder is attached
UncleG: good question... is this an otherwise well functioning system?
I have a situation where after about 20 minutes of running fine the drive just runs away and faults out
hmm, not sure what to suggest there
My suspicion is that the encoder is missing a resolution some how and well. zip....it runs away
does it also happen if you disconnect the drive from the PC and short the command input to ground?
the encoder shouldn't be used by the drive. the drive more likely uses the tach only
so the tach wires could be flaky
ok.. im a bit lost.
here is my setup...
you have an older velocity mode amp, and there's tach feedback from the motor to the amp
anyway, go ahead. I'll code fro a bit :)
old standalone CNC system all things self contained, I wired into the encoder with scope
scope causes drive to run away immediatly
otherwise drive just runs away on its own after 20 minutes or so.
I want to catch it in the act with scope.
what kind of encoder is it?
accucoder a,a' b,b' z,z' 1000 cpr
ok, how did you hook the scope to it?
you may not be able to use the parallel port directly then, you have a differential encoder
SWPadnos: I'm guessing position runaway, not velocity runaway
yes, but it's a velocity mode amp with tach feedback
pin 11 and 12 from lpt wried directly to a' b' which is what is also connected to drive
oh, I didn't read back, I thought we were debugging an old system
we discussed it a bit last night, and I still remember bit sand pieces
SWPadnos was around last night for my repeated failures :)
sorry, what does an lpt port have to do with it then?
GLentek servos and GA-370 drives
the idea was to get something connected to EMC2, as a learning experience
just encoder feedback at first, to see the numbers change on the screen as the shaft is turned by hand
SWPadnos kind of.... I may be using EMC2 full time at my shop.
(hooked to spindle input via stepconf)
providing this old cnc box keeps giving me troubles.
that would be good, but as I recall you wanted some baby steps last night :)
gotta start somwhere
I think a Mesa card would be a great purchase for you, if you can show yourself that the amps work (and it's the CNC box that's bad)
one way to do that is to disconnect the amp from the computer and short the command input to ground (I don't recall the pin numbers)
ok it just ran away again :(
if the servo takes off after 20 minutes in that configuration, then the tach and/or amp are bad (or the connection between them is)
what is the command input connected to?
ok.... disconnect the amp from the computer? im sadly lost.
you mean commanded velocity right?
I do, yes
UncleG, we don't know what you have connected where, or what you're doing with the PC/EMC2, or even what "runs away" means exactly
I have nothing connected to the pc right now, the cnc system is self contained and running on its own.
All I was looking to do was get a readout on the scope of what the encoder did at point of crash
ok, have you disconnected the encoder from the CNC box to connect it to the PC?
I left it connected to the cnc box
so it's like a Y cable from the encoder to both the CNC and the PC?
I just wired into it with lpt to see if I could get a readout on the scope
I agree with SWPadnos. to debug this, disable the velocity command and see if it's stable
is that not possible to do?
disconnect the amp from the CNC box and short the command input to ground
I wouldn't try to hook the encoder to the old system and also a pc
no, that also sounds like a bad idea
though we do know that the encoder is 5V, which is good :)
there are too many ways that can be bad news (ground loops)
also you'd be messing with one side of differential signals
if you lose encoder feedback it will run away (as you know)
but there are other reasons - so you need to test one at a time
I dont know what the command input it :(
*scrambles through schematics and wiring diagrams*
no docs for the amps?
so you have all the info I have :)
search for "AMPLIFIER"
if I had those amps, I might consider registering on the glentek site and getting the manual
looks like pins 1 and 9 are command signal and ground, respectively
there is a jumper for differential input
not on the parallel port
that's not how I read it
are we talking about "GLENTEK GA370 CONNECTION" pins or "SHADOW 9 PIN D SHELL CONN" pins?
I'm just throwing in interesting tidbits from the glentek manual, I'll shut up now
I was referring to the SHADOW connector for pins 1+9
that's not a parport
I know that
I forgot that i was stfu
on "SHADOW 9 PIN D SHELL CONN" 1 is velocity command, 3 is common
there is effectively a Y cable between the encoder, the CNC controller, and an EMC2 PC
I saw the 9 in "9-pin connector" and thinkoed
unhook pin 1, strap to pin 3
wait 20 minutes
ok :) I can handle that :)
if you can get to the screw, turn it - it should feel stiff and fight with you
if that works, get a 1.5v battery and hook it both ways between pins 1/3. you should be able to drive the axis around at 15% of top speed
wait a sec.
pin 1 on the board or pin one on the D connector?
look at the link SWPadnos pasted
you should understand what you're doing
(that link is from UncleG last night)
SWPadnos: sorry for butting in
(troubleshooting is fun)
no problem. I'm trying to fix some completion code
butt away :)
actually I should go to bed
Pin 1 and 3 on the amp board or the d-connector end that is attached to it?
"SERVO COMMAND -10V to +10V" and "GND. / COMMON"
looks like the SHADOW 9PIN D SHELL CONNECTOR from that manual
it depends on what you disconnect though
k, ty I was looking in the wrong spot in the man
it's quite hard to read
the intent here is to disconnect the CNC box from the servo command, and short it to ground (for a zero velocity command)
it is 2 and 4 on the board where I will be doing the shorting, so thank you i am working on this.
don't short it if it's hooked to the cnc
short the input to the amp, NOT the output of the cnc
right - make sure that the (BROWN) wire gets disconnected from terminal 2
hahah 26: POWER FAILURE DETECTED A DIVIDE BY ZERO ERROR OR A TRIPPED WATCHDOG ERROR.
error 26: I don't know what happened - maybe you do?
maybe a single NMI or something (divide by zero could cause a trap that's similar to an external interrupt on whatever CPU it is)
ok, the drive is turning.... very slowly.
with the command shorted?
or a 1.5V battery
that creep is normal
that's what all the creeps say
null needs adjusting
there will be an adjustment marked 'balance' or 'offset' if you want to null it out
you might have to do that if you want to wait 20+ minutes
I do have a balance option
before turning any pots mark where they are currently
tweak till it stops...
I wouldn't mess with it
I would, since it's marked offset, you're sure it's the right one
but, I also understand why someone might not want to :-)
I did that with a servo hydraulic system and it was just a good way to get oily
the position loop in the cnc is what normally holds position, it doesn't care if there's a little offset
integrator term gets rid of it
eric_unterhause1: thankfully I've never had to mess with a hydraulic setup - seems crazy
scope is kicking out a nice square wave now.
hydraulics are awesome until you have to disconnect an oil line
did you get the axis stopped?
cool. now we wait!
UncleG: how did you get the scope working, was it your hal code?
more specifically, I go to bed
eaxis stopped encoder kicking a wave
how might one log the entire 20 minute wave? :)
what is an eaxis ?
one of the waves keep getting a funny spic in the middle of it.
I wonder if that would be my culprit.
or I just don't know what im looking for
you're seeing this on the computer?
oh, btw should only phase b, be giving the wave, cause A has nothing
that may just be that your encoder is on a line
A never has anything?
just a line
B has a aquare wave
UncleG: A should be 90 deg off of B
UncleG: i.e. shifted about half a square
its not doing anything I will check my setup and wiring again
everything is fine.... but.
my B side on the encoder kicks out a bouncing needle on the voltmeter where the A bise of the encoder reads nothing
is my encoder bad/
sounds like it is to me
with only one side of the encoder working, you'll get the speed of the motor, but not the direction, depending on what's actually reading it
but the motor works... for twenty minutes or so...
that's really odd
is it possible it will work forawhile then after 20 minutes it just run away from me?
I mean will it work at all without the A side of my encoder?
I don't see how it could do anything other than go in one direction with only one side of the encoder working
it works just fine for about 20 minutes. then blam... runs away...
I sure wish I knew more about this stuff.
i.e. you could command the motor to go, and it'd go and look like it's working, but if you try to reverse the direction the decoder would think it's still going the other direction
UncleG: it's certainly possible (but I don't know how likely), that the encoder works for a while, but stops working once it heats up, or something
UncleG: I'd imagine the motor heats up to some degree after running for 20 minutes?
but after a quick poweroff and then back on, its back on its wheels
surely not long enough to cool down
hmm, not temperature related then
jst_home is now known as jst
changing my fault range from 1 inch to 3 inches triples the the time it will run before crashing out
does that help you pinpoint the problem?
jst is now known as jst_home
UncleG: not immediately :(
Yes, Im in the same boat.
UncleG: is it still running there, still with nothing on A?
it not "running its powered on and at a a standstill
just kicking out my square wave
UncleG: oh, it's at a stand still and you're still seeing a square wave from the necoder?
also, should there be a spike in the middle of the wave square ever?
every few waves I do see a spike in the middle of the sqaure.
UncleG: ok, then what *could* be happening is that the motor happens to sit right on the edge of a step on the B channel in the encoder it could be oscillating back and fourth and generate a square wave, but I can't imagine that would be at a consistent frequency etc
UncleG: in an ideal world there would be no spikes
UncleG: this is an analog scope, right?
this is the EMC scope
oh, hmm, then I don't know what to really expect there
what happens if you try to turn the motor by hand? Assuming you're able to w/o loosing a hand :)
the wave turns into a scratchy mess
if the motor does turn, you really should be seeing things happening on both channels
but still only on the B channel?
lemme get the drive turning...
I can do that with the balance
well, adjusting the balance does change the wave length but still no A wave
I can totally stop the wave/make it extremely long
but still no A
Im still really curious about this spik in the encoder.
you don't happen to have access to a real scope by chance?
If I zoom in on the wave it shows me that the encoder drops its high wave for a split second
I'm at a total loss as to what's going on there I'm afraid :(
no biggy that is my current state of mind on this whole situation about 85% of the time.
you don't happen to be in northern California by chance?
if you'd be close by I'd offer to bring a real scope over to see if that sheds any light on the situation :)
well, that's a bit far away...
I do appreciate the offer, however distance does create a problem :P
UncleG: I'm pretty new to this stuff, but I've designed my own DC servo controller, which is working quite nicely on my first prototype board here, so I do know more than nothing about encoders n' stuff
but I've never dealt with a differential one
danged ole servo setup anyhow
is this on an old mill, or what kind of machine is it?
And index machine
it's a differential encoder?
not sure what kind of encoder this is
maybe you are looking at the wrong line
what is a distinguishing mark on my encoder to tell me what is what?
UncleG: um, what's an index machine? rotary index table kinda thing or something entirely different?
UncleG: if you have two wires per channel, you're likely looking at a differential one
kind if like a turret lathe
UncleG: if you have ground, power, one A, and one B (plus Z on some), it's a non-differential encoder
UncleG: oh, ok
this is my lineup a,a' b,b' z,a' 5v, common
sounds differential to me
the ' lines are probably return, i.e. no interesting signals
unless they aren't
it is always possible to have a dead light in the encoder
sorry to jump in the middle, but do you have a manufacturer and part number for the problem encoder?
I have another encoder here I am going to compare it to.
I just need to finish this soldering job up.
so you are looking at A and B or A' and B'?
but guess what~
I just hooked up this other encoder and I get a reading from both a' and b'
UncleG: progress! :)
the only problem with this drive is that it runs continously....
providing I can switch the encoders (which is nearly impossible due to design) I should have one fully functional drive
when a unipolar stepper is rated at 3.8A per phase, that is per coil?
so, just to make sure I am clear, if I were to drive the motor in bipolar mode, that would be 1.9A, right?
should be 3.8a
if you have 3.8a limit per phase, I don't see how you could hook it up so it has lower amperage rating
So my motors are 6 wire unipolar, I can run them in bipolar mode by not hooking up the 2 power wires, which doubles the coil resistance because the two coils on the A and B sides are connected together, make sense?
or you can use the center tap
which seems to be recommended
use the center tap in bipolar?
a little, but ok
what drives are you using?
using linisteppers right now, which are unipolar, I was just trying to make sure I understand how much current to feed the motors
so I need to feed my motors about 1/phase more, maybe I can get them to spin a bit faster :)
they get hot
speed is more of a function of voltage
I am almost at 10X rated voltage right now
the motors dont seem to get too hot yet, the controlers get really hot though
what controller are you using?
[06:36:05] <JustinXJS2> http://www.piclist.com/techref/io/stepper/linistep/index.htm#KITS
I thought that was the stepper motor itself, sorry
ah, no worries, motors are vexta
is there any sort of calculator for figuring out feed rate VS spindle RPM?
you need cutter diameter in there somewhere
and material being cut I would assume?
there are tons online
search for feed rate calculation
oh yes, loads of calculators, thanks for the tip
Encoder phase A is not quite half of phase B on the scope... any suggestions on what to do?
nm, that question. I should be reading.. ttyl all tomorrow.
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-02-03.txt
does anyone know about very cheap motion controller card
to drive what
to drive servo system
how many axes?
a lot use cards from http://www.mesanet.com/
actually I am looking for a pci i/O card that gives me a support to connect external hardware (my own counters and DAC's) that gives me access to 32-bit i/o per address
cnc_engineer: not sure I've seen such a card
but in that case you'll have to take care of a driver for emc2 aswell
sorry i was disconnected..
I am looking for a PCI bridge card that gives me access to computer pci bus. And I can connect my counters and dac's with that bus..
cnc_engineer: not sure I've seen such a card
but in that case you'll have to take care of a driver for emc2 aswell
yes, i can write a driver and have microcontroller that can count the encoder and give 32-bit data for encoder position. Now i want a that data to connect to computer PCI bus
alex_joni: 5120 can't do that?
JymmmEMC: it can do much much more
but it's probably too expensive for what he wants
basicly he wants a card with the PLX that's on the 5i20
only PCI access chip.. don't think I've seen one recently
but i want to develop my own hardware for emc as the hardware suggested on emc website are too expensive
I wouldn't say 80$ is too expensive, but it's your choice
I have the software and electronic expertise to develop counters for encoders and DAC out only i need a PCI interface to access the data and address bus
its going to cost more to save a few pennies
cnc_engineer_: a quick google search found Tiger 320
which is a chip that does what you want
here's a board that has that chip: http://www.futurlec.com/PCI8255.shtml
but it's not that much cheaper :)
(oh, and it's supposedly buggy, and doesn't work with emc2 iirc)
cnc_engineer left the room (quit: Read error: 110 (Connection timed out)).
JymmmEMC needs some coffee :)
indeed it is
* archivist wonders what pjm__ wanted last night just after I went out to play
ah yes i was gonna ask about your cnc, what IO ports are u using to control stuff, parport or one of the messa cards?
as its a stepper machine
and Im a cheap bar steward at the moment
hehh me too!!!
i'm gonna do some further testing to see what I can clock into the parport reliably
for spindle encoder
I need a better paying job
or some activity that pays
i'm sure its possible to generate pin money making custom ally boxes for microwavers
Im not sure this thing is rigid enough for that
but I know I can follow through with a vna test :)
ah yes ;-))
having cnc at home is pretty handy i must say
a mate of mine in the US just finished a cnc conversion, basically to cut out 'D' connector holes for his projects
of course also running EMC2
heh quicker to make a punch for some hole types
ah yes but why do it easily when u can build a CNC to do it!
mine is far to light in column construction at the moment
good morning.one question.the another cnc when i write m3 s2000,stop the read next partprogram line.when the spindle revolution ok read the next step.because in emc the interpreter read immediately the next line after m3s2000?
perhaps your question in your native language
motioncontrol: emc2 doesn't wait for spindle to reach the speed
if you need that, you will have to put a G4 pause
or use a more recent emc2 (emc2.3 will have a pause to reach spindle at speed)
you can also use the motion.feed-hold to control emc2 if you want it to be stopped until spindle-at-speed is true
still going http://imagebin.ca/view/yxPk_e.html
skunkworksemc: how long?
ok thanks alex
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-02-03.txt
alex_joni: over night
02:04:49 -> 12:44:27
heh - too early for math'
I am lucky I got the 10+ hours right..
skunkworks: wow, what kind of machine is this? http://imagebin.ca/view/yxPk_e.html
[15:04:27] <BigJohnT> http://cgi.ebay.com/Okuma-LR10-Lathe_W0QQitemZ370152170082QQcmdZViewItemQQptZBI_Lathes?hash=item370152170082&_trksid=p3286.c0.m14&_trkparms=66%3A2|65%3A7|39%3A1|240%3A1318
$750 to load it!
cradek: it is the ecs goal3 motherboard.
This machine powers up sometimes but then shuts itself off after a few seconds. The amount of time the machine stays on varies from time to time. Machine was bought at an auction and painted by previous owner.
haha, it doesn't turn on right, so we painted it
and now we hope to hell it sells for $750.99
yea, did you see his feedback!
cradek: I think it is going to be the new gantry computer. We have been getting by with a 800mhz pentium.
It will be nice to have a speed gui ;)
looks like a very nice system to use
BigJohnT: he only screws about 1 in 20 people...
that ain't good odds in my book
highest loading fee evar!
sure is blue
would really work nicely together with tkemc
what I want to know is the 9900 pound weight before or after painting?
I think it's at least 20 pounds of paint there
you can't really fault the guy for his handling fees, they are stated in the auction
eric_unterhause1: yes you can: it's ebay fee avoidance, which is against their rules
theres handling fee and extortion
shouldn't be grounds for negative feedback though
I might possibly agree with that
but, it's grounds for avoiding the dishonest seller
I thought ebay was dumping fee avoiders
I avoid high carriage cost sellers unless I can collect free
I have gotten some decent deals when shipping is too high
because people avoid the auction
You have successfully submitted a report. We appreciate you taking the time to file this report. eBay will investigate to determine resolution.
don't know if they will back you up on that one, rigging fees vary all over the map
I don't understand why people like to paint over cables
[15:20:27] <alex_joni> http://i7.ebayimg.com/02/i/001/2e/b8/8475_1.JPG
I sent a report to a sellers local trading standards office for one odd ebay item
alex_joni, lazy, used a spay probably
that's so annoying
alex_joni: I don't either, no wonder they couldn't get it to work
especially if you want to debug/fix something
they should have gotten it to work first
[15:24:08] <eric_unterhause1> http://cgi.ebay.com/4-Pin-Amphenol-Pigtailed-16-gauge-cable-P26263-E25_W0QQitemZ230323232488QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item230323232488&_trksid=p3286.c0.m14&_trkparms=72%3A1205|66%3A2|65%3A12|39%3A1|240%3A1318|301%3A0|293%3A2|294%3A50
4 pin cable can be used for encoder
stupid seller, but good price
16ga is pretty heavy for encoder feedback -- this table says 16ga is suitable for 3.7A of power transmission.
it's the power cable
corresponding encoder cable has 15 pins IIRC
must be a different class of encoder than I'm familiar with. usdigital offers 26ga cables for its ttl encoders (4 wires for +5V, A, GND, B)
(and up to 500'!?)
this is for an industrial brushless motor
they use differential encoders with hall tracks
today stay at me the second pci card m5i20.i wnat use in one pc two 5i20 card.i have installed and start emc.the terminal eroorr is:hm2/hm2_5i20.1 invalid cookie and hm2_5i20.1 boar fail hm2 registration.i thing is necessary load the firmware in the second card.which?
motioncontrol: depends what firmware you want
i want use the second card for only i/O
then any firmware will do the same
you probably want something like:
config="firmware=hm2/5i20/SVST8_4.BIT num_encoders=3 num_pwmgens=3 num_stepgens=0, firmware=hm2/5i20/SVST8_4.BIT num_encoders=0 num_pwmgens=0 num_stepgens=0"
the firmware the first card is hm2_pci:svst8_4.bit
you can use the same firmware for the second card, but not activate any components
ok i prove one moment
gotta run.. bbl
[15:48:08] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=72801
have problem : hm2_pci is in use
skunkworks: I have no idea what he's talking about
cradek: seems there's some MTConnect component that's supposed to work with emc2
cradek: same here. never heard of the hardware..
but I can't find it at mtconnect.org
[15:57:04] <alex_joni> http://agent.mtconnect.org/
[15:57:49] <alex_joni> http://agent.mtconnect.org/current
that is one informative web site
[16:00:18] <alex_joni> http://www.amit-deshpande.com/2008/06/mtconnect-live.html
[16:01:39] <alex_joni> http://mtconnect.org/download/MTConnect_NEWS_05_23_08.pdf
there is a student competition, that's the problem
i have modification the hal file but the problem is :EMC2 - 2.2.8
Machine configuration directory is '/home/emc/Desktop/emc2-2.2.8/configs/M5i20'
Machine configuration file is 'M5i20.ini'
insmod: error inserting '/usr/realtime/modules/rtai_hal.ko': -1 File exists
Realtime system did not load
Shutting down and cleaning up EMC2...
ERROR: Module hm2_pci is in use
<commandline>:0: exit value: 1
<commandline>:0: rmmod failed, returned -1
ERROR: Module hostmot2 is in use by hm2_pci
<commandline>:0: exit value: 1
<commandline>:0: rmmod failed, returned -1
<commandline>:0: unloadrt failed
ERROR: Module hal_lib is in use by hm2_pci,hostmot2
ERROR: Module rtapi is in use by hm2_pci,hostmot2,hal_lib
ERROR: Module rtai_math does not exist in /proc/modules
ERROR: Module rtai_sem is in use by rtapi
ERROR: Module rtai_shm is in use by rtapi
ERROR: Module rtai_fifos is in use by rtapi
ERROR: Module rtai_sched is in use by rtapi,rtai_sem,rtai_shm,rtai_fifos
ERROR: Module rtai_hal is in use by rtapi,rtai_sem,rtai_shm,rtai_fifos,rtai_sched
EMC terminated with an error. You can find more information in the log files
as well as in the output of the shell command 'dmesg' and in the terminal
motioncontrol: use pastebin next time
i don't know paste bin excuse
rtai wasn't unloaded
because this problem? the rtai use the hm2_pci
[16:08:23] <alex_joni> http://mtconnect.org/index.php?option=content&task=view&id=110
motioncontrol: all those error messages are because you tried to load modules on a system where the modules were already loaded.
you may have to unload by hand using rmmod in the correct sequence
use lsmod to find which modules are being used by other modules
if a module is not being used by another module, rmmod will remove it
using "sudo rmmod" actually
sudo make me a sandwich
can i post the lsmod result.excuse but i don't have expert for linux please.
[16:14:28] <eric_unterhause1> http://pastebin.ca/
and give us the link
ok i prove
[16:15:46] <motioncontrol> http://pastebin.ca/1326314
sudo rmmod hm2_pci; sudo rmmod hostmot2;sudo rmmod hal_lib sudo rmmod rtapi
oops forgot a semicolon
ok i prove
isn't there a script that unloads rtai?
otherwise sudo rmmod all the other modules starting with rtai_
i write first string sudo rmmod hm2_pci the error is: hm2_pci is in use
what does that do?
it unloads realtiume
halrun -U, realtime spelled without a u in it :)
sounds like there is something using the hm2 driver
you need to remove all of them with the same command, if you want to do it the hard way
I assume hm2_pci will unload itself
halrun -U should remove all EMC2 related modules, along with RTAI
I guess used by 1 means something is using it, what is it though
motioncontrol: could you put a the result of ps -A on pastebin?
i have restart pc and start emc the error is:http://pastebin.ca/1326318
you have a more complex problem than removing modules. you got a kernel oops
and you're running on a custom kernel
i use debian etch 4.05
what "HAL file" did you modify?
you built the kernel?
you want post my hal file a apstebin?
yes i have patch the kermel 2.6.22 with rtai 3.6
and then you built emc against that?
whith one card 5i20 fuction ok
i don't understan against that
I suspect that the hm2_pci can only load once
i have add the second card in pc and this the problem
the old m5i20 driver can use 2 card?
probably never been tested
i speack with mesa for this problem?
where is your hal file posted?
one moment i poste it
hal file: http://pastebin.ca/1326327
you probably need to change the loadrt hm2_pci line as follows:
loadrt hm2_pci config="firmware=hm2/firmware/5i20/SVST8_4.BIT num_encoders=4 num_pwmgens=4 num_stepgens=0"," firmware=hm2/firmware/5i20/SVST8_4.BIT num_encoders=0 num_pwmgens=0 num_stepgens=0"
the difference is that the two strings are separated by a comma, instead of having the comma inside a single string
I don't see in the hostmot2 manpage where it says to do either of those things. I don't see how to configure two boards.
I'm not sure the manpage says it well
but looking at the source, "config" is an array of strings
one per board
each string has the option of specifying firmware, encoders, stepgens ...
one string per PCI board
check your quotes, that looks like ther emay just be a typo
there may (see! :) )
seems to me that there would be a significant effort to write a driver for more than one board
it's supposed to work for more than one board
even different board types
SWPadnos: "even different board types" seems easier to me
i want use the second card for only i/o
eric_unterhause1, different board types which use the same hm2_pci driver
my machine is big
have more input and output
in conclusion the hm2_pci driver suppor more pci card ?
anyone know who wrote the driver, was it Pete?
yes it does, seb kuzminsky wrote the driver
sounds like it might be a good question for the email list
I'm pretty sure there's a typo on the loadrt line still
I'm not convinced motioncontrol has got the command line right yet
the error had two quotes at the beginning, not one
and only one at the end
no comma between the two quotes
there is an unintended space between the quote and the word "firmware" in the second string. that may need to be removed
there is a comma, the problem is reported by the second instance initialization
yes i writr ","
so the first one was fine
there was a space just after that in the line I pasted - that should be removed
I wonder if we're getting into some weird quoting /parsing issue with halcmd vs. shells vs. hm2_pci
do you have two cards?
(I sure don't)
i have write "firm.....","firm....."
the problem is equal
[17:04:50] <motioncontrol> http://pastebin.ca/1326333
yeah, I can stick a few cards into one machine
motioncontrol, try this: "'firmware...','firmware...'"
ok i prove
error parameter or pin :hm2_5i20.0..pwgen.pwgen_frequenzy not foud.i thing no load the driver
can you pastebin the errors. that looks to me as though the driver loaded but something else happened
yes one moment
are there actually two .. between the 0 and pwmgen?
half of those words are spelled wrong
yeah. I'm not sure where the transcription error os though
I have no idea
makes it very frustrating to try to help
[17:11:52] <motioncontrol> http://pastebin.ca/1326357
that didn't look bad
who is the people is project the hm2_pci driver?
i want question him
send an email to the emc-users list. that's probably the best way
he's also on IRC a lot, but not at the moment
ok thanks for all
motioncontrol, did you change it to "'firm...','firm...'" ?
(I'm wondering if that's why the driver loads now
ok you have 2 card?
I have many cards, but only one in a PC at the moment
I can experiment with that later
one thing to look at: take out the pwm frequency line and see if it gets through the file
you can connection now the second card?
in a few minutes, yes
ok i wait
ok, pleas do this while waiting :)
when you time want?
take only the hostmot2 and hm2_pci lines from your hal file and put them into another file
(lines 17 through 21 in the pastebim)
ok one moment
then halrun -I <the new file name>
that will give you a halcmd command line, and the hm2_pci drivers should be loaded
within halcmd, show all
and paste the result (it will be several hundred lines)
[17:28:40] <motioncontrol> http://pastebin.ca/1326371
ok, now halrun -I <whatever that file is called>, and paste the result of "show all"
one moment but halcmd i write in terminal or in emc.emc no load
do not load emc
halrun, in a terminal
terminal error no such file or directory the command not exist.
you need to source emc-environment
if you do it in the same terminal where you've been trying to run EMC, it should work
i 'm in scripts directory.in this directory this is halrun , but i write hulrun the error is command not found
if the environment variables aren't set up right, you need to type ./halrun
if the environment isn't set up right, there are probably going to be other issues
agreed, but I was too lazy to type that
don't worry about it, I have the machine with two cards in it, and I see that the driver isn't exporting anything
i started ok i have line command halcmd
how do you see what symbols a driver exports?
I use show to show pins, parameters, functions
you said it wasn't exporting anything
I mean in a HAL sense, not a kernel sense
is the mechanism different?
for HAL components, they have to expose things they want the user to be able to use
the resul is http://pastebin.ca/1326383
this is called "exporting", but it's not the same as the symbols that are exported to the kernel in a module
motioncontrol, I'm looking at it right now. there is a problem with loading multiple cards
I'll look at it some more, and check in with Seb when I see him next
you have connection 2 card in your pc?
for now, I don't have any answers for you
yes, I put in two 5i20 cards
you have my problem?
no, I have my own problems :)
but yes, I see that there is a problem loading a driver for more than one card
ok swPadnos you will modificated the driver ?
for load more card.
maybe me, maybe seb. I'm working on it now. thanks for the trouble report
thank at you at all comunity i return tomorrow for news?
sure. I don't know if we'll be done by then, but you can definitely check in
thaks alredy and good night( in ITALY)
bummer. it's working here now :)
I need to figure out why
well, that was with only firmware strings - so there are no spaces in the strings and I think
s/and I think//
I also didn't use quotes when loading
well crap. now it seems fine. I wonder if it was just the extra space he had in there
that can do it
you would think the parser would fix that, but not always
parser is too generous a term in some cases
nope. that's not the problem
"call to strtok"
Ive done stuff which is more raw state machine
well, it all works fine. there were several other problems with his hal file
and my testing methodology at first
(note to self: when looking for PWMgens on two cards, leave the ".0" out of the show pin spec)
SWPadnos: so what's the incantation?
om mani padme hum
actually, it works exactly as he first had it, with the exception that the strings are too long
we don't need Chenrezig now
and I'm sure he never answered to _your_ calls
when I made the firmware strings "hm2/5i20/..." instead of ?hm2/firmware/5i20/...". it worked
I also had to change to a different directory, since have installed and run-in-place firmware, but anyway
there are a bunch of pins named hm2_5i20.0.gpio.P3.xxx, but the P3s aren't there using hm2_pci
(and the numbers need to change, I'm sure)
seb changed that lately I think
and he's only adding hm2_5i20.0.* to threads
not lately, hm2_pci does it this way, hm2_5i20 does it the other way I think
* skunkworks finally got trunk..
Question: In 2.2.8 there doesn't seem to be a "dirsetup" parameter for hm2_5i20. It does exist in TRUNK. How far apart are 2.2.8 and TRUNK, and will there be a release soon that equalizes any important differences? I ask because I have to ship a machine to a customer tomorrow (Wednesday) and I want to know whether I should wait for a released version, of give them a run-in-place version that I know woks well...
nothing like waiting until the last minute :)
about 2 months for emc 2.3.x
2.3.0 to be precise
not sure about the dirsetup though
OK, I guess that settles it...
I thought hm2 was pretty much the same between 2.2.8 and trunk
but I must be nuts
running 2.2.8 complains about dirsetup with my hal file that works with TRUNK
hm2_5i20 or hm2_pci?
(man I hate step/dir)
mshaver, you could try hm2_pci instead
you can try hm2_pci instead ...
yes, that's the one that supports all the PCI boards in one driver
the GPIO pin names will chang
I don't know what other differences there are, if any
nothing like waiting until the last minute...
I'll give this a try.
here's an interesting TRUNK commit message: "new 5i20 bitfiles from Peter, with fixed stepgens"
i __NEED__ that!!!
probably not in 2.2.8
then I think the path is clear :)
no, I don't see any additions to v2_2_branch after that commit
or at least the path is potentially navigable
(for the 5i20 firmware anyway)
mshaver: I see the same dirsetup things in v2_2 branch as in TRUNK
judging by grep -e dirsetup src/hal/drivers/mesa-hostmot2/*
I don't know what to say. Maybe it's in hm2_pci and not hm2_5i20, but the need for the latest fpga bitfiles really supercedes all other considerations.
are you sure you have 2.2.8?
yeah, dirsetup is there in 2.2.8
It said 2.2.8 in the error message printout, let me look again...
the latest bitfiles are definately only in TRUNK though
oh ok, I suppose the rest is academic then
that's different from "there is no dirsetup parameter"
having no parameter is a driver issue. having it not work is more likely a firmware issue
when I start up in a terminal it signs on as EMC-2.2.8
or just as likely anyway :)
what bitfile are you loading?
right now a custom one from Peter that slightly preceeds those ones he checked in with the stepgen fix
oh. in that case maybe you should ask Peter for a new bitfile with the fixes
the driver should work though, and there should be dirsetup parameters
my bitfile has the fixes I think. He found the bug on my error report.
I meant "very slightly" preceeds...
what indicates to you that there are no dirsetup parameters?
EMC2 fails to start and complains, "kc6.hal:48: pin or parameter hm2_5i20.0.stepgen.00.dirsetup not found".
TRUNK works OK
with the same file?
same config files
that's a good indiacation :)
line 48 in my .hal file refers to dirsetup
rtapi_snprintf(name, HAL_NAME_LEN, "%s.stepgen.%02d.dirsetup", hm2->llio->name, i);
^ from 2.2.8
I see the dirsetup code in 2.2.8 stepgen.c, so I don't know what the problem is
yep, that's the code
are you sure there isn't another error earlier?
like the whole driver fails to load, and that's the first time you try to manipulate it?
maybe the stepgen didn't get exported..
yeah, you've got a more basic problem I bet
I don't think so, let me check...
hm2_5i20.0" Inconsistent Module Descriptor!
hm2_5i20.0: board fails HM2 registration
sounds like the bitfile is newer then the driver knows?
if you only knew how to fix that, you'd be all set
hm2_5i20.0: probe of 0000:05:00.0 failed with error -22
aha! error -22!
whatever that means
I do! Use TRUNK!
yep, I like error -22
well, there is that
TRUNK it is!
It's one of those pre-runtime errors
it's a "you wish you could get to run time" errors
EINVAL (is -22)
Is this maybe the (I dont like stepgen version 1) error?
If I "sudo make install" from emc2-trunk/src will it install TRUNK over 2.2.8?
yes, that's not the best thing to do
mshaver: I would advise you 2 ways of doing this
I could make them a DEsktop icon, but I want them to leave the munu launched (installed 2.2.8) version alone...
1. uninstall 2.2.8, then use ./configure (without --enable-run-in-place), then make && sudo make install
but better yet use sudo checkinstall
checkinstall is a program that runs make install by itself, but instead of installing puts stuff in a package, which you can install and uninstall later
option 2. build a proper emc2 deb file
that's pretty simple, as you have most everything installed already
just sudo apt-get build-dep emc2
then debian/configure -r
and sudo apt-get install debuild
if I "apt-get remove emc2" will it kill anything else important?
no, only emc2 and sample configs
and it will remove emc2-dev too
but I assume you don't need that either
I would need that if I wanted to do more compiling on that system, right?
but I think I'll try making the .deb
then if the guy needs updates I'll build him .debs he can install
OK, I'm going to refresh my TRUNK, compile, and build the .deb now!
mshaver: I'm here if something doesn't work :) getting it right the first time is a potential pitfall
isnt it always...
compiling without run-in-place and with build-documentation
no need to compile
building hte package wi
building the package will compile for you
the first thing it does is make clean, so your compile only proves it's working
sudo apt-get install debuild
sudo apt-get build-dep emc2
ok, doing that now
sudo apt-get install fakeroot debhelper
(not sure if debuild pulls those in)
apt-get can't find debuild
sorry.. it's devscripts not debuild
sudo apt-get install devscripts
did you run debian/configure -r ?
is debian/configure -r next?
that configures the building process for the running kernel
(hopefully it's something like 2.6.24-16-rtai)
if you're on Hardy 8.04
now run debuild from the top emc2 dir
compiling now - had to install fakeroot
mshaver: gotta run, I'm sure jepler or cradek will help you out in the next 60 minutes if needed ;)
probably you need debhelper aswell
don't mention it
(open a new terminal, and install debhelper)
beats waiting for a compile to error out, then redoing it :)
just a note if you read this later - debhelper was already installed - may have been part of emc2-dev which is also installed, or maybe devscripts
you don't need to sign the package as it says
that is only needed if you want to stick it into a public repository, which you don't :)
SWPadnos: Thank you for your help the last couple days, I now have a fully functional CNC system again
what did you have to fix?
I routed out problems with the encoder of one drive and compared my results to the drive that just ran continously when plugged in, I found out that the old round lettered connector going to the one drive was shorted out and the encoder was not getting proper signal to the CNC box.
Whats a run on sentence?
well, with a lot of soldering and checking wires in turns out the problem was little more than a bad connector.
heh. I think someone mught have mentioned wiring as a possible problem :)
but that was for the drive that I wasnt even working on :P
it usually pays to check the simple stuff first
The drive I was working on, the A and A' channel of encoder was bad.
a friend used to do computer tech support. something like 80-90% of problems were fixed by the first two questions:
is it plugged in? is it turned on?
It was just shorted out :)
I found the problem checking the connector with the ohm meter it seems the 5vdc in was also putting voltage to 1 of the Phase wires.
My next tes will be to see if I can fix the original drive with the bad encoder and add a third axis to my machine.
check the encoder before removing it from the motor - it's probably a real pain to replace
disconnect the motor/encoder at the closest connector to the motor, power the encoder with 5V from USB or something, and use the same halscope setup as before to look at phase A and B as you slowly turn the motor by hand
it appears that this one may be shorted out as well
the encoder reads A' and B' just fine when competely removed from the "mothe??" connector
alex_joni: just to let you know, if you read back to this point, that it all worked, .debs, installation, all of it. This will be a good way to distribute updated versions of TRUNK, rather than trying to teach people how to maintain a run-in-place setup. Thanks Again!
Im going to check out that "mother??" connector with the ohmeter
mshaver: that's neat
I'm glad trunk packaging is in such good shape already
SWPadnos: s/ is it plugged in? is it turned on?/Did you restart the computer yet?/