:) "Human Interface Device, i.e., one of those terribly slow things whose data rate is limited by how fast humans can twitch." excerpt from a uk adc using pic with hid desriptors
tomp: You dont have a dHID? SWPadnos does
Mine's a Humanoid Interface Device though
SWPadnos: Direct to the spinal column
Just images the spam, viruses, and trojans! lol
The ULTIMATE in peer-to-peer computing!
hey all. little help on hal?
i'm trying to follow Anders jog wheel example on his site. his hal file starts by 'loadrt threads name1=tt period1=100000 name2=tt2 period2=1000000000'
any interterrestial communicators... other than mine???
when i put this in a hal file and call from inside my .ini file i get an error of
insmod: error inserting '/home/noel/emc2-trunk/rtlib/threads.ko': -1 Operation not permitted
i've definitely been rtfm'ing for quite a while. just feel really stuck or thick.
that's a really long period isnt it?
don't really know. it's for counting the pulses on a hand turned pulse generator..
also you have to be careful when creating threads, make sure that they're a multiple of the smallest period (fastest thread) and load that thread first
load which on first? the short or the long?
dmess: like a shortwave radio?
or a tight beam gravity wave transmitter :D
Roguish: what happens if you just open up a halcmd and type those commands?
yeah.. that one... i could use 1
shortened the period by 00 and same error.
Roguish, have you looked at dmesg?
hey there's a good idea
that's where you find out the real problem loading modules
what's that command again to see dmesg?
command duh not found
Roguish: periods over 2^32 nanoseconds (about 4 seconds) aren't supported, and ones over 2^31 nanoseconds (about 2 seconds) may or may not work properly
that was only 9 zeroes, or just under 2^30 ns
oh my mistake
is there a way to clear the message and get a fresh listing?
sudo dmesg -c
clears what was there, the next dmesg will only show "new" messages
or you can run dmesg | tail -30
to get the last 30 lines
and i shortened it to 7 zeroes and get the same error.
Roguish, are you running this hal file after loading EMC?
it is called in the .ini file
is there a line in this file that does "unloadrt threads" before the threads load line you pasted?
add one :)
SWPadnos: threads is not used in the sample configurations; they all use the threads created by motion
hmmm. good point
Roguish, what's your BASE_PERIOD?
but I am sure that dmesg does have the ansewr
Roguish: did dmesg say anything?
is this a servo machine or a stepper machine?
sure, about 19 pages worth
m5i20 with real servos.
would you put that up on pastebin.ca or somewhere?
pastebin the output of `dmesg | tail -30`
ok, are you sure the BASE_PERIOD is 50000 and not 500000 ?
halcmd: loadrt threads name1=tt period1=100000 name2=tt2 period2=1000000000
halcmd: show thread
Period FP Name ( Time, Max-Time )
999979628 YES tt2 ( 0, 0 )
99938 YES tt ( 0, 0 )
in 'halrun' (emc not running), the proposed 'loadrt threads' line works fine
what's the link to pastbin?
[00:31:33] <jepler> http://pastebin.ca
in the context of emc, usually you do not want to create additional threads; instead, you decide whether each new function should be on the existing "fast" thread or the existing "slow" thread, which are called servo-thread and base-thread respectively
or even the really slow traj thread (I think that still exists)
however, any standalone .hal example (not using emc) will have one of these 'loadrt threads' lines because without threads you can't do anything
[00:32:56] <Roguish> http://pastebin.ca/708630
(in short, you probably dont need the threads module)
hm this experimental 126.96.36.199 kernel is spewing "Bad page state in process" errors into dmesg :(
it is 500000, not 50000 :)
why would it be 500000?
sorry, been a long day at the computer
[00:33:54] <jepler> http://pastebin.ca/708630
[ 2927.456283] HAL_LIB: ERROR: new thread period 100000 is less than existing thread period 494480
because it's a servo machine, and doesn't need to do step generation
a servo rate of 2 KHz is plenty fast
(unless you're doing HSM)
would a fast base thread that did nothing hurt anything?
the base period is 50000, the servo period is 500000 (10x base period)
threads must be created from fastest to slowest; the requested creation order here is: 50000, 500000, 100000, and 1000000000.
but 100000 < 500000, thus the error
it's just wasteful of CPU resources, and it can affect the phase of PID control in the slwoer (= lower priority) servo thread
there is a fast base thread, that's not the issue
I was wrong :)
so, basically he has to change the hal file to use the existing threads, right?
that's the solution I'd recommend: put the "fast things" on base-thread and the slow things on "servo-therad"
is there a way to stop motion from exporting threads?
except spelled right
Roguish, I'd remove the loadrt threads line, and change anything that uses "tt" to use the base thread, and anything using "tt2" to servo-thread
fenn, yes, rewrite motion
if you really need a slower thread than the 2kHz one, that can be created after motion
ok, will try that.
ok, got past that. new error 'HAL: ERROR: function 'm5i20.0.digital-in-read' is not reentrant'
take that line out of anders' file
just delete it?
you need to look through the jog wheel file, and take out any line with an addf or loadrt (or loadusr) command on it
why the 'addf' lines?
if that same function is already added to a thread or the component on a loadrt/loadusr line is already being loaded in another hal file, then delete the line from Anders' file
because you can only add a function to one thread, and it can only be added once
you already had a 5i20 read for EMC, or nothing else would have worked
the jogwheel file also wants the 5i20 to be read, so it tries to add the read function to a thread, but it's already there, so you get an error
well, i'lll be. no error (this time!)
how does rsync work? does a daemon have to be running to use it?
or is all the meat in the client side?
why are you asking in #emc hwo rsync works?
what dont you backup your emc systems?
because they're expendable
* fenn puts a no at the beginning of that sentence
thought you said why
I figured some in here might use it is all, just never understand if a daemon is required to be running or not
yes, there is a daemon
is it expensive to operate?
err - I don't think it costs any more :)
I think it sleeps until it gets a connection
ah, ok. thanks.
then all it should have to do is check things like file time, size, and if necessary do some hash or diff on the contents
I'm not sure of the specifics of that though, you could certainly look it up
SWPadnos: Since I only get 1U, I might see if I can toss two systems in one case.
nah - get one of these:
(waiting, stomping feet =)
[01:16:47] <SWPadnos> http://www.aberdeeninc.com/abcatg/Stonehaven-A135.htm
quad quad-core + 64GB + triple SATA HD (up to 1TB each) in 1U
SWPadnos: Ah, no. I mena for reduant systems
not for performance
ahy have redundancy when you can have 16 cores!!! ?
ok, off to work. back in 90
you get 42U in a standard rack, right?
that's 672 cores in one rack
only $4k each
actually, I was surprised it's that low
that can't be
that's a paltry $160k per rack
they're only $11k with 3TB, 16 cores, 32G, and a SATA RAID card
with the less expensive quad core CPUs, they're still under $5k I think
I didn't know AMD actually had 4-core processors out yet
oh sorry i didnt select quad quad
oops - $6500 with quad quads
ohhh.. just what I need sub - par quad cpu's...
$20k with qq 2.0 and 64GB
it's the 64GB that kills you there
go to 32G and it goes down to the 11-12k range
64GB.. what the heck you gonna do with that
oh I can survive on 16GB just fine
need 64 to mae a real cad/cam box...
they really are inexpensive for what they are
that doesn't mean that everyone can afford it (or needs it), but that's a steal for a 1RU computer with that amount of power
* fenn wonders what planet swpadnos lives on
oh whenever I contemplate lots of cores it's for a compile server not cad/cam
this is the same kind of inexpensive as "oh, that Ferrarri was only $150k, what a steal"
ferrari's are supposed to be expensive
they dont do anything but show off how much money you have
like I need to spend $150k on a car that I'm not able/allowed to drive faster than 75 or so
ok send a 2 Gb file in for crunch... you better have ram and HP
dmess: what files are 2GB and need to be loaded all at once?
i loved my dual 200 pro.... and the quad 200 pro...
tool pathes.... make redundant files... working on ALOT of iges translated files... its all an HP thing.. and time back then...
I'd probably get one or more of those as an image processing farm
i find it strange that you can't sell spare computer time on the net
id use it as a desk top... but i'm wierd
I think someone tried that
similar to the SETI@hHome/BOINC thing
like, you can _give_ it away, but you cant sell it
but somewhat paid
yo can giv it away..
Actually Fenn we came up with a business model to sell spare computer time - and redundant encrypted storage.
M$ offered us $250,000 and a R&D office building in Redmond to develope it there...
We figured it wasn't worth $250,000 to sell your soul...
awallin's new m5i20 testpanel ( wiki updated ) http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?M5i20_Halvcp_Test_Panel
not to them..
SWPadnos:Im making progress on the usc setup,have it turning my power enable contactor on,fiving to put power to the dc power supply
gonna spin a motor soon
[02:23:21] <jlmjvm_> http://imagebin.org/10585
heres a pic of the usc build so far
thanks,been a long weekend
JVM - what voltage you feeding those vampires?
[02:28:32] <Skullworks-PGAB> http://stage6.divx.com/MAX-MOTOR-SHOW/video/1493031/mad-driving-skill
( don't know if the streaming divx works in Linux yet)
these in the pic are g320 servo drives,they are getting bout 35v,using the factory transformer,has 45v servo motors
or did you mean the vampires in my bp stepper mill,they are 62v
is that pic bid enough,first time ive used gimp to shrink 1
detail is good enough for most stuff
the Power section in the upper left is dark - as well as the contrast on the geckos leaves them dark.
the Maxtor is perfectly lit
[02:40:30] <jlmjvm_> http://imagebin.org/10586
and the white table kinda set the lighting balance
heres another view
I have 3 203V's and a 320 coming on tues
what kinda motors you hooking em to?
not sure why though
impulse buy - the sale price was too good
don't really know yet
i hear ya,thats when i got my vampires
should have ordered the 320,s then also,but didnt have purchase order yet
I'm going to be converting my Grizley/Seig X3 to servos (which I have) but likely will buy Jon's amps for them.
kinda wish i had went that way,was gonna be a dual parport setup with ferror off the 320,s,but was told it would be hard to split the encoder signa
and already had the 320,s
I only ordered 1 320 to play with
I might use it for an A axis rotary tabled
i have 3 mills in the field using them every day,only 1 has failed due to a crash
jlmjvm_: Well put a fence around em so more cows crashing into them
sounds like my friend old place - cows kept tipping his sat dish over...
they thought its a scratching post
we are in the country for sure
"The mill goes Mooooooooooo"
wouldnt mind converting my mill to servo,but havent found any motors at a good price yet
I almost wished I had bought that place - but things at work had me nervis
[02:55:03] <Skullworks-PGAB> http://www.surpluscenter.com/item.asp?UID=2007062621461319&item=10-2351&catname=
these are new
are good for about 600 oz-in at stall
37V "Nominal" for continous duty - but I see people feeding about 60V via PWM
ive seen those before,wondered if they were any good,the price is right for sure
they are Skewed rotors - real servos
very smooth - no cogging
thats what i was wondering about
I bought 4 from them - then found they are on ebay for the same new motors even cheaper
sounds like my luck
(sold from Mass.)
/window size 25
i may get 3 of these when i get done with this retrofit and do a pwm system for my mill
note: the rear 3/16" shaft is too short for most encoders - I'd make a 1/4" press fit extension
or I should say I will be
it is small for sure
i wonder how well they would work with my 1 to 1 pulleys
I'm going to make complete new rear covers with an encosed area for the encoder and seperate area to act as a conduit junction for the motor pwr wires
I'm going to run them direct drive
if they are 600 oz they should work good
im glad you showed me that
If you want pics of the insides - I can shoot some this week
I should also post the solid model when its done of the replacement rear covers
that would be cool,may wanna make some
Solidworks - one of the few good reasons to still use windows
mastercam for me
Skullworks-PGAB: and notepad.exe
its all i use it for
mastercam is good for the toolpaths
but we build the IGES/DXF/ etc in Solidworks first
yep,made a bunch of parts over the years with it
tahts the way to do it
used to say I used notepad to write all my html
but sometimes I cheat these days and dump an Open Office doc to HTML - as bloated as M$ almost...
you can make models in mastercam,but it wont keep up with a real solids package
but it is handy
I have a full install from a company I do work for - have to log onto there network to "check out/take out" a license key.
they can have it installed in every work station - but only 3 active seats at any one time.
so it uses a license server system.
Once EMC tapers off - only adding support for new I/O and such - the G-code is almost fully mature. The next big hurtle is to build a powerful linux opensource CAD/CAM package.
that would be awesome
Hey guys, can someone give me a quick pointer to start pluto_servo using HAL, I've read through the manual and can see how to do so with other systems like the Mesa5i20 but there is no info on launching the pluto
well not that I found anyway...
the lathe_pluto config uses the pluto_servo
[07:11:19] <alex_joni> http://cvs.linuxcnc.org/cvs/emc2/configs/lathe-pluto/
that should be good for a starting point (only 2 axes though)
I started doing that but as my mill is pretty much ready to start testing I was going to get the configs done to suit it straight up
and what's the problem?
I assume I need to do something like loadrt pluto_servo ...
_epineh: start by using that config
and extend it from 2 to 3 axes
o i c
I will try that
ask if you get stuck :)
alex_joni: you up early
Jymmm: I'm up 2 hours already
it's 10am over here :D
then you up REALLY early =)
toast___ is now known as toastydeath
if you are still there alex, I was wondering if it is ok to take 5v signals from the encoders straight to the pluto, i have seen comments both for and against, I know with some 3.3V micro's they advise to put a resistor in series.
_epineh: right, 5V might be too much for it
k, I know with the prop chip by Parallax they advise to use a resistor from 1k - 100k when bringing a 5v signal in, I think I will do the same.
_epineh: sounds good, the alternative is to use an opto
thnx alex_joni, I gtg but I am sure I will be back soon with more annoying questions :P
_epineh: I'll be travelling later today, but there are other people who can probably help you
alex_joni: the pluto-p I/O pins are 5V tolerant according to http://www.altera.com/literature/ds/acex.pdf
-- voltage on any input pin up to 5.75V is permitted according to the table on page 45
Guest139 is now known as skunkworks
how is your backlash coming?
I never worked on the weekend
I am right where we left off on friday
heh - slacker :)
I had a garage sale that was hell.
03swpadnos 07TRUNK * 10emc2/src/hal/utils/pci_write.c: Trivial fix: corrected program name in usage statement.
Has anyone tuned a servo/glass scale system that has backlash in it?
is it a split ballnut? could you take the backalsh out of it?
when I went looking for backlash on my mill, I found it in surprising places. Don't assume it's in the nut unless you've checked everything else
cradek: on your big mill? where was it? in the bearings?
loose slide, nut not fixed to slide......
endfloat on screw
skunkworks: everywhere... it's pretty worn. in X I was surprised to find it in the block where the screw is bolted to the table (screw doesn't turn)
screw-to-block: loose. block-to-table: loose.
sounds like you need lock washers
or a tightening once a decade
at least every other decade
Well I am confident that its in my yoke, Bridgeport style machine...
welcome, anonymous tor people who never say anything
hm, does everyone but me know which part is the yoke?
it's the thing that holds the two ballnuts
Adam1, if you think your yoke is flexing (and you have a manual/retrofit mill), check out Elrod Machine for a replacement yoke that's stiffer
but to answer your question, no, I haven't tuned something like that, and frankly I'm not entirely sure it's possible to get good results
[16:57:47] <SWPadnos> http://www.elrodmachine.com
if you had tach feedback into emc's pid, I think it could work better
I have tach feedback that goes to my drives
the drives are tuned to the motors
emc and the drives arent getting along
how much backlash are we talking about?
around 3 thou
heh not run in yet
if in emc you set BACKLASH=0 DEADBAND=0.003 can you get some stability?
SWPadnos:Can the usc be configured to use 1 ssr for m3 and m4,i have an external relay,dpdt with a 24v coil to use for the direction change.
other than the estop stuff, the SSRs and inputs do whatever you want - just connect them how you want in HAL
[17:10:26] <jlmjvm> http://pastebin.ca/709375
would you use the invert on this?
on the spindle outputs?
I think not, but if the outputs act the opposite of how you want, then invert them :)
I'm pretty sure with the USC you want to use false = external device off. The internal ESTOP hardware forces the SSRs off in ESTOP, so you want the external machine to be off in that state
(ie, probably not inverted outputs)
wouldnt it be the same output,but inverted for the direction change
if you wanted to use only 1 ssr
using one output means that either FWD or REV is always active
using two outputs, you can have neither active
but i dont have 2
well, unless you always want the spindle on, you need to have two :)
0v=no spindle doesnt it
because there's no way to say "off" if the only two words you know are "FWD" and "REV"
0v = commanded speed of 0, not "motor power removed"
I wouldn't be the one to cahnge tools in that situation
[17:30:38] <renesis> http://www.doingitwrong.com/wrong/585_munkavedelem.jpg
Another genious with a welder
[17:38:16] <Adam1> http://imagebin.org/10600
how accurate is the path... lol
Adam1: do you have g61.3 on?
err g64 Pxxx
try it in exact stop mode and see what it does..
or in g64 P0.01(depending on mm/inch)
g64 is active
not sure on value
well, set it to a known one..
I set g64 p0.01
looks like crap still
well, it might just be the previes..
what feed rate was this at?
I think the display programs only sample the position "every once in a while", so you won't see the exact path on the display
run it at 10 and see what you get (and realize that if your motors are up to it, then it will be nearly identical at 60, even if it isn't displayed that way)
I have a question regarding acceleration, how does one set that...
[AXIS_n]MAX_VELOCITY in the ini file ...
SWPadnow, I set it lower and it follows closer. It still deviates thought.
err - MAX_ACCELERATION or ACCEL ...
run halscope and look at the following error output parameters for the axes (there may be one overall as well) - that'll tell you how far the position actually deviated, regardless of the display update rate
I'm pretty sure you're seeing a userspace samplibng issue, not an EMC trajectory issue
XPOS-CMD vs. XPOS-FB?
no, axis.n.f-error I think
though I don't recall now whether that's the actual error or the ferror limit at the current velocity
I have axis.n.motor.pos.fb
axis.n.f-error is a parameter
I know that :)
it's an output from the motion controller that isn't expected to be needed elsewhere in HAL
but you can still scope it
hm, got any link to the servo amps skunkworks are using?
So I want to scope that
and compare it to?
Adam1: that's the real deviation.. of the axes..
Adam1: not what the userspace program things are going on
anonimasu: this stuff? http://www.electronicsam.com/images/KandT/servostart/?M=A
ok so how should I setup halscope to read that properly
uh - I can't walk you through that right now. maybe someone else can ...
type halscope, click on servo-thread, click on 1, click on parameters, click on axis.n.motor.pos.fb
hey its a pin not a parameter
now in the upper right you can click on 'force' to make it sample
or you can trigger on another value, like Xpos increasing or something
you don't want motor-pos-fb, you want axis.n.f-error
which is a param
fenn: though I'd want the brd file :)
[18:04:29] <fenn> http://www.electronicsam.com/images/KandT/servostart/schem/amp.brd
I think that is the last one I cut.
did you try it much?
only up to 10a and 60v so far. (been busy)
much/does it work ?
that's about what I want :)
it doesn't get warm at that..
100v and 7a peak
though that board will be a pain to toner transfer :p
Red is axis.0.motor-pos-fb, Green is axis.1.motor-pos-fb, blue is axis.0.f-error, purple is axis.1.f-error http://imagebin.org/10602
thats doing the emc program at 60ipm
what's the scale?
ferror looks pretty close to 0 to me
hard to know without knowing the scale :)
yeah I dont know what the scale is
let me go again
clock on one of the channel numbers
and the scale will be shown where it says "scale" under the vert sliders
the scale is independent per channel, so set them all to the same value
they are all the sam
ok, make the two f-error plot something like 1m/div
how can i do that?
the 'gain' slider
click on the channel number, then drag the scale slider
err - right, gain slider :)
ok this is with 10m/div http://imagebin.org/10603
the blue is axis 0
.06" of error on that initial acceleration - looks like you need to do some tuning...
yah - that does look pretty bad
I like to make a program that uses G0 and G4 to go back and forth, short moves at full speed (with a delay in between so it doesn't bust itself up)
then you can trigger on the motion and get a nice display while you fiddle the tuning values
when you get it to follow well, try longer and short moves to make sure it's decent for those too
Well I am not sure about the tuning process really....
Ive read and read and looked and looked. its more art it seems
it's not art, but it's hard to explain remotely
first get a good stable plot like I suggested
then fiddle with the numbers and get a feel for what they do to the plot
in no time you'll be good at it.
well thanks fenn
* fenn tips his cap
i think with backlash in your system you'll have to be careful not to overshoot at all
so start with lots of D
yeah that's going to be problematic.
ok whats lots of D compared to P
what's your P?
100 right now
lots of D might be 2
its sounds like my servos really hate lots of D
at first, try I=0 D=0 and turn up P until you get a bit of sustained oscillation (be careful!) and then add D until it damps out
vibrations like crazy
yeah too much P or too much D will oscillate
hehe.. estop is your friend!
I get oscillations around P425
with P=425 I=0 D=0 ?
try leaving it there and add D a little at a time until it stops
I tried increasing D untill it stopped, I never got very far and my drives stopped working.
It SOUNDS like you "Ladies" are trying to repair a recipe for a dinner this evening or something.
Add a little more salt/butter/sauce/etc until it reaches 325F, then...
In my recipe I think I will have to add 2 more dashes of drives now
lol, there ya go =)
well thats an expensive addition
thats like adding kobe beef
But using quality ingrediants makes a world of differnce
Yeah I turned the oven on too high I guess and burnt the beef
Adam1: you can always put encoders on the leadscrew/servo...
or is it going well>
No my drives arent working anymore, that = not well
skunkworks how is that better than galss scales?
no backlash - as the encoders are on the servo.
Could there be thermal cutoffs/resetable fuse in the drives/motors?
skunkworks: wouldn't scales be better as it could read/compensate for any backlash?
JymmmEMC i sure hope so thoose are $900 replacement :|
Adam1: what drives?
* skunkworks just read about the drives not working - eww
Adam1: 1525-BR ?
I"m seeing two fuses on the side of it
Adam1: brush or brushless ???
[18:56:03] <JymmmEMC> http://www.servodynamics.com/Manuals.html
JyymmmEMC SEM 30M4-59
is it on that page?
thats my servo
Brushed DC servomotor
Cont. stall torque 3 Nm Rated Torque 0 Nm Rated Speed 0 rpm
running IRC warscripts I see? ;)
[19:02:32] <JymmmEMC> http://www.sem.co.uk/files/brochure/MT30M4.pdf
Adam1: brished servo motor
rated speed: 0rpm
max speed: 0rpm
rated current: 0A
isn't that call a paperweight?
I bleww some fuses
"I'm sorry you've exceeded the rated amperage, this voids our warranty"
30A 140V 3400rpm
7.5 volts/1000rpm tacho
3nm stall torque
I blew some 15A Slow-Blow fuses
"Warranty void upon receipt of product"
"This product hasa lifetime warrantee. It's dead, therefore it's no longer in warranty."
Adam1: Is your drives/motors new/used?
jepler: I'll keep it in mind the next time
yes/no it/they is/are
used from an older anilam conversion
Adam1: Do you know it/they work?
They did about 30 mins ago
untill i started turning up D
Adam1: Their manuals are really skimpy, maybe there is another one that I can't find for tuning.
do they hold position, or are the motors free-whelling?
Adam1: Hell, I don't even find the fuse ratings in the manual.
ie, are the drives broken or is it misconfigured?
well the drive will hold, they just bounce off the glass scale counts
I can make the machine do things, it just wasnt accurate
then the drives are fine and you need to tune
however now the drives arent working I need to fix that first :p
ok, I'm asking you if the drives will hold position right now, not 1 hour ago
You know, bullets make great replacement fuses
no the drives dont do shit now
SWPadnos: Right now, he has two blown 15A fuses
I dont even know if they are on
I am checkin the elec. with the mulitmeter
time to get out the voltmeter and see what's powered
I took the drives out they look fine, not burn marks
they are rated 15amp
so.... i cant see them blowing up
did your machine oscilate before it happened?
yeah it was a bastard
your machine might have pulled more then the max rating.
or well, back emf or something
* JymmmEMC thought BEMF was ONLY a stepper thing?
JymmmEMC: it's cetainly not..
JymmmEMC: what happens when you brake a motor?
anonimasu: I dun know.
brake or break?
You shunt the two leads together to make it brake
I am on a fuse run brb
skinnypuppy1334: if a machine oscillates what does the motor do?
anonimasu: look for prozac?
im going to do something else.
im getting an error when i try to start the usc
[19:19:18] <jlmjvm> http://pastebin.ca/709526
you have probably mis-spelled an ini variable name
or the variable you want is missing from the ini
look at like 47 of univstep_servo.hal (as the error message says)
just copied the new files in that Jon sent
can you pastebin univstep_servo.hal and univstep.ini
[19:22:44] <jlmjvm> http://pastebin.ca/709537
[19:24:15] <jlmjvm> http://pastebin.ca/709541
ok, look at the hal file, line 47
(as directed by the error message)
47. setp pid.0.FF2 [AXIS_0]FF2
then look at the ini file, in the AXIS_0 section
there is no FF2 variable in the ini file
the same is true in the other AXIS_n sections
Neither on AXIS_1 or AXIS_2 as well
so you should add a variable, proabbly set to 0
FF2 = 0 <-- put that in each AXIS_n section of the ini
emc is starting now
taht did it
did you see how I found the problem?
SWPadnos: you RTFEM ?
i didnt realize error 47 was line 47 in the file
it said "error 47" or "error at line 47" ?
that makes sense,was looking for a parameter that didnt exist
HAL:47: Ini variable not found.
a little non descript. but once you know...
Ah, that is a bit ambigous.
I would have thought that meant error code,
thats what i thought
HAL.ini Line 47: variable not found
I was told that the format it is in now is 'normal' linux error format.
(the reason why it is formatted that way)
skunkworks: You mean like man login
is the unix standard
the older versions of halcmd start mesages with HAL:linenumber: which isn't quite as useful
glad that came up,learned something
jepler: If it had said "hal.ini:47: error msg I MIGHT have figured that out w/o prior knowledge.
a few microwave fuses later and I am back to square 2
Adam1: Now to confuse the wifey when the microwave dont work
Now back to tuning.... ugh
hi jlmjvm, how goes it?
going very well,just got a motor spinning on this usc project
i put some pics on imagebin yesterday of the panel im building
[20:42:36] <jlmjvm> http://imagebin.org/10586
[20:43:10] <jlmjvm> http://imagebin.org/10585
alex_joni: how do you set a pid loop when you set up a drive
P=60,I=1.00,D=.1 is what i have
that sounds reasonable
really,was wondering,first time to do this
maybe more D and less I
only way to tell is look at the plot, really
is that in emc
on the machine menu
note that PID tuning needs to change if you test with the motor unattached, then connet it to the machien, change the power supply voltage, change the current limit on the drive etc. ...
you need to tune with the motors and machine in their final state (but learning on the bench is also good)
yes,would really like to know how to use this scope to check what i have
[14:19:12]<cradek>I like to make a program that uses G0 and G4 to go back and forth, short moves at full speed (with a delay in between so it doesn't bust itself up)
[14:19:30]then you can trigger on the motion and get a nice display while you fiddle the tuning values
[14:20:01]when you get it to follow well, try longer and short moves to make sure it's decent for those too
that's one way :)
another way is to just set the commanded position in hal
or try the autopid
where is that?
it's only in CVS TRUNK
not in 2.1.7.
and it's not really production ready
from what I heard, it didn't always produce usefull results
jlmjvm: I would do it like this:
add a couple of signals to watch (axis position, ferror, velocity, etc)
watch ferror, and command a move
you'll see the feedback (cmd-fb=ferror) trying to catch up to the commanded move
let me connect the network to the control so i wont have to move back n forth
I won't be long .. and I'm only talking about theory here..
ok, so the idea is to have some predefined position command
and the feedback to track that as well as possible
say you are at X0, you command a G0X10, it will try to go there (taking into account accel, vel limtis, etc)
the point is to change the parameters until the motors follow the prescription as good as possible
in order to do that, you need to always change only one parameter at a time, and test the result
if it gets worse, go one step back, and try something else :)
gotta run, battery is running out
good night all
is there a way to tune with software and hardware. Something you can just hook up.
Roguish_ is now known as Roguish
im having a little difficulty trying halscope
what kind of difficulty?
added the signals alex told me to add
see a segment up in the corner
its going off the screen,then going away
you are going to have to be a little more verbose
if the traces are off screen, adjust the vertical scaling - slider to the right of the screen
how do you set the trigger
do you have any experience at all with real scopes?
see the button in the bottom right corner?
(the entire far right column is trigger stuff)
click it to set the trigger source
(you can pick any signal that you are observing)
adjust the signal scale if needed, using the vertical gain slider
then adjust the trigger level with the level slider
it will trigger when the signal crosses the trigger level
speaking of scopes,anyone tried the bitscope stuff?
in the rising direction, unless you click the button that says "rising", which will change it to falling
the trigger pos slider sets the trigger point horizontally in the screen - if you need to see mostly stuff that happens before the trigger, then put the trigger point near the right of the screen - if you want to see stuff that happens after the trigger, put it near the left
I have to leave in a minute or two - any last minute questions?
to set a pid loop what signals would you watch
trying to watch ferror
i have xpos cmd,xpos fb,and axis0 ferror
are you getting following errors?
no,i can jog the motor
ok, then why are you scoping?
I have to go, back later