serial LCDs are waaaaaaay more expensive than non-serial
Gamma-X: how is it going?
eh could be better. im sick lol
not wanting to tune my machine! haha
tunin that machine sucks! hahaha
Help! My PID still sucks!
join the club'
I thought I had it dialed in, but when I turned down the allowable following error, the lack of tune was revealed
what type of move would you like to see, fast g1 or fast g0
something that shows the problem you want to fix
GNieport1: if turning down the following error limits caused a trip, you need to turn UP the scope gain on the error channel
during tuning, you should be watching pid.N.error
your goal is for that number to always be small relative to the tolderances you intend to work to
0.001" or so for example
something to chew on while I get the shots: I believe I left out a key clue. The encoders feed through the servo amplifier, and it has an internal loop. I have the documents, but cannot tune it precisely due to a large noise signal on the 'scope output pin. I'm afraid the hardware loop and EMC are fighting each other. I am considering dis-assembly of the encoder harness to bypass the amp's "smart" stuff
is this some step-dir servo amp? do you know what it is doing with the encoder signals?
it feeds directly through, but the amp has internal servo loops, with no apparent override.
"buffer only" and "the amp has internal loops" are incompatible statements
damn i need a ciggarette... i quit today
hehe, the encoder raw signals feed directly from the input to an auxiliary output. The amplifier does pay attention to the signals and attempts to influence the motor.
ok, then thats not buffer only
I understand that you were saying "the amp doesn't do anything TO the signals", but it could also be read "the amp doesn't do anything WITH the signals"
you're gonna need to figure out the natire of the amp's internal loop(s)
is there a possible way to had pid setup auto maticly? in the future..
everything (almost) is possible
nothing is easy
use at_pid, and hope it works for your machine
(but that's not a real answer ;) )
auto tuning nice
one person's attempt at auto-tuning PID
who made it?
do u trust it?
I've never tried it
when I need to tune PID (which isn't often), I just do it myself
[01:16:02] <GNieport1> http://imagebin.ca/view/6GNECtW.html
purple is error, right? 50 microunits per div?
SWPadnos ever use at_pid
what are your units, mm or inches
That is a g0 move at 300IPM, with a g4 pause on either side. Purple is pid.0.error, inches
so that 50u scale is in user units?
what the heck are you complaining about then? you are within 0.000150" of where you are supposed to be for most of the move
what do you have your ferror limits set to?
I didn't know it was a real world number until just now. I get a following error often when the move begins or ends
max is 0.2
see those big spikes in error at the beginning and middle of the move?
they're bad ;-)
do you have any FF2?
no, I tried adding it starting with 0.01 and got more f_error
I mean to say, EMC stopped due to folling wrror limits
the move in the picture - did it reach steady state speed? it looks like it didn't - accel then decel, no cruise period
I suppose it did not reach steady state
(nothing wrong with that, I just need all the facts to interpret the image)
what do you have max_accel set at for this axis? (in the ini file)
G0 X 2.500 F300
G0 X -2.500 F300
G0 doesn't use F words
G0 always goes as fast as it can (machine limits)(
what are the velocity and accel limits for this axis in the ini file?
5.3 IPS axis limit, 20 IPS^2 accel
so zero to full speed in just over 1/4 second
it must be very near to full speed when it begins to decel
I'd suggest setting the following error limts high again (0.1 or more, for MIN_FERROR and FERROR)
GNieport1 just use at_pid
then run this move or one like it several times, experimenting with FF2
Gamma-X: please do NOT make that suggestion
Due you think it might be productive to remove the encoder from the amplifier
GNieport1: I dunno
do you have a URL or anything with amplifier info?
please post it
hahaha just an idea
im sick im goin to bed ill talk to u all tomrow when i can hopefully get some time to tune my machine.
if the amp is using the encoder to close a velocity loop, you probably want to leave it connected
have a good night. see you Friday or so
ur comin to ny?! ull see me?
see your nick on IRC Friday or so
[01:28:40] <GNieport1> http://www.idcmotion.com/support/Manuals/b8001_15.pdf
damn hahah it was worth a try!
later SWPadnos and everyone else.
I was thinking of removing the encoder, and maxing out the internal P gain
Teh problem is that I cannot effectively tune the amp loop without the scope connection working.
that front panel looks suspiciously like a step-direction drive
it is, also analog input
when you say "without the scope connection working" are you refering to halscope or an external real oscilliscope?
let me guess, the analog value only generates a pulse stream- DOH
dunno, I'm reading
I read fast, but there are 41 pages, please be patient
the amplifier requires a real-worl scope to tune
do you have a matching IDC motor, or just some random motor?
do you have their software disk?
matching motor, and I have the servo tuning software running right here
and scope, with a nasty 60Hz hum
I hate smart drives
in this case, i do as well :)
They are pretty sweet for indexing automated machinery, though
did you follow the "Quick Start" procedure _exactly_ before hooking the drive and motor up to EMC?
are you running in velocity mode (Mode input NOT jumpered to common)?
it says you can set the velocity scale from the "Inputs and Limits" screen
did you do that, so that +/-10V corresponds to roughly your desired max speed?
I have, I programmed the maximum safe ballscrew speed there
what speed is that? (in ipm or ips)
5.3 IPS, 1600 RPM
thats the max safe speed?
you don't believe in safety margin do you? ;-)
with a bit of safety amrgin
yes, i do, I have seen a lot of machinery explode over the years.
the max I calculated is something like 2200 RPM, but I made an assumption or two
and it's 38 years old
I'd try FF2 while watching these spikes
ok, I just got to "Diagnostic Monitor Port" (page 22)
is that the port you have your real scope hooked to?
its supposed to be 2.5V, with +/- 2V swing from there
what are you seeing on the scope? millivolts of 60Hz, or volts?
and is the DC level at about 2.5V, or somewhere else
I am seeing about a 200 mV 60Hz noise superimposed on the feedback signal, which is swamping the overshoot spike
what parameter are you observing? (that port can be used to see lots of things)
I have a ground loop or something, but cannot isolate it :)
the tuning software has a routine that applies stimuli
Square wave for the p loop
Degree rotation for the position loop
Degree rotation for the integral loop
stop a sec
from the manual, talking about the diagnostic port: "Actual velocity, commanded velocity, velocity error, position error, commanded torque as
well as other parameters can be displayed with ease.
it can display only one of those things at a time
which one are you trying to observe?
"Tracking (positon) Error
something is wrong here
according to the manual, when using the analog input, you can only run velocity or torque mode
yeah, I bought the wrong amplifier
so there is no way there can be "position error"
If I set all the gains to zero, the motor does not spin
do you mean the manual I'm reading isn't for your amp?
it is the correct manual
then why do you say you bought the wrong amp?
i bought an amp that does not play well with EMC
from what I can see, it should work fine
but it needs to be set up properly
what kinda amp?
I have one of those
never used it yet
Really, i'm committed here. I have not been able to nullify the 60Hz hum, no matter what. The best other alternative to tune this is to guess at values and listen for oscillation
GNieport1: you want the amp to be in velocity mode, with EMC doing the position loop
I agree, there is no option to make the amp "dumb"
velocity mode is dumb enough
you did say it was an IDC motor, right>?
page 23 of the manual: On the IDCMotion disk set that came with this drive are motor data files (ex.
B23_220.mtr) that supply the drive all the necessary motor data. This data includes motor
winding information, pole count, and encoder resolution. These files also contain default
tuning parameters. The tuning parameters have been optimized for use with IDC motors
and actuators. They work very well with inertia ratios up to about 15 to 1. Most
applications require NO ADDITIONAL TUNING.
so - why do you think you need to tune the amp?
When I loaded the motor file, all values were zero
that would be a problem
the gains for the drive's internal loops?
Kv, Kp, Ki, Kfv Kfa
do you have one of their motors?
the file you loaded is the right one for the motor you have, right?
Thanks for looking through that document, I must have messed up the quick start.
Let me reload it for fun.
can you decouple the motor from the screw? either remove a belt, or a coupling, or something?
yes, pull the oldham disk out
I strongly suggest that you decouple, and using ONLY their tools, software, procedures, etc, get the amp working as a velocity amp with their default tuning, and a scaling such that a 1.5V battery applied to the input terminals results in 240 RPM
the default tuning is indeed blank
when you can make the motor turn 240 RPM with a 1.5V battery, then its time to connect to EMC and start tuning the position loop
where did you get the disk (and the drive) from? can you get a new copy?
I think eric and I have both asked and you still haven't answered - is this one of their motors?
I downloaded it, I do not know if it is still online, I will check.
Yes, it is a matched motor
did you buy it from them (warrantee - their motor, their drive, their problem if it doesn't work), or is it ebay, dumpster, etc?
they charge $500 to match your motor to their drive
Ebay drive, secondhand motor
I have two drives from ebay, different sellers, both act this way
I'm not sure what to suggest at this point - what I can say is that you want to keep things as simple as possible, and that means keeping EMC out of the picture
I'm sure the root of the trouble is that the amp internal loop is not tuned.
until it works right with a battery as the input source, don't try it with EMC
thanks for the thoughts :)
GNieport1: I bet you are right, but neither of us is really qualified to tune their internal loops
thats why they have matched motors with the tuning params preset
when you load the motor file, do any of the settings look correct?
All of the settings load correctly, except that the loops gains are not set
do you have a link to the manual?
are these position loop gains, velocity loop gains, or torque loop gains?
[02:01:04] <GNieport1> http://www.idcmotion.com/support/Manuals/b8001_15.pdf
that is the basic drive motor
there is another manual for their tuning software
the drive manual doesn't list any params or anything else
(I don't have the URL for the tuning software manual)
Velocity, Position, Integral, Velocity FF, Acceleration FF
it's not online, let me put it up on ftp, brb
bastids: on their manual download page: "Servo Tuner - See ServoTuner Help file"
he could autotune :)
does their software have autotune?
kvtune kptune kitune and autotune
where did you see that? in the manual?
oh, their page numbers, not pdf page numbers
yeah, found it
[02:16:14] <GNieport> http://home.cinci.rr.com/bog/stuff/b32.pdf
no manual for the software
do you want a screen shot? :)
we figured that out: manual is the help file
I can get a photo of the loop structure, but I think we are beating a dead horse now. I Must fix the scope tuning procedure.
Or find the defalut values somehow
have you tried to run their auto-tune?
The button is grayed out. I guess the user manual was optimistic about it working by ship date
there may be something you have to set first
are any of the radio buttons for torque/velocity/position set?
[02:23:20] <GNieport> http://home.cinci.rr.com/bog/stuff/servo_tuner.jpg
Yes, I set every option for velocity.
did you enable?
mcmastercarr is nice, but a friend just turned me on to Misumi http://www.misumiusa.com/
( i was looking for small linear guides, rails and bushes ) not bad pricing (except for flat plate stock ) and they publish prices
after mechanically disconnecting, that is
I like misumi, but I have never ordered from them
that was a different computer, I installed the software only to see if it included a help file in the archive.
auto-tune gonna be grayed out until you enable the drive
misumi has awesome quality
eric_U it has always been grayed out
even when you had it enabled?
ok, I'll move on to the next bogus theory then
hwo about the theory that enabling didnt enable
that's a crazy theory
not simply bogus
what was tried didnt succeed. thought it was enabled but wasnt
I found some defaults
all you have to do is ground enable, don't you trust GNieport1 to do that much correctly?
I loaded the "step and direction" profile, instead of the analog profile
so GNieport1, since tomp2 says we shouldn't trust you, did the disabled label change to enabled when you enabled the drive?
the big question, do I try EMC one last time?
I thought it did at first, but it was just the wind
if the companies software can't get it to move, don't mess with emc
the motor moves excellently with the software.
Tracks a square wave, even
too many loops are getting stacked.
that's what I get for crashing my IRC client
I don't think you have that problem
Yeah, my real problem is that my first servo install I thought I would get all fancy and go brushless. A dc motor would have been so much more simple.
I watch a lot of people messing with geckos, and I don't envy them
hrm, grass appears greener
any other brushed option is really crazy
what does it do when you try to run under emc?
I didn't try it yet.
did you try the 1.5v battery test?
what are you using for an interface, Mesa?
The full disclosure is that this isn't the first brushless system I have tried to install. I bought resolver-based hardware first, and could not get it to work for months.
so I bought these
Mesa, no 1.5V
what system w/resolvers?
with PacSci motors
that isn't going to work
what is your theory
I don't see the problem yet?
the resolver frequency was ALMOST the same...
I think most of those drives were meant to be matched
not sure about the baldors though
why wasn't a Baldor amp and PacSci motor going to wirk
because they didn't agree on phasing
I think my cables were not good enough
did you research the phasing issue?
the motor would track at full, half speed, but at 10 rpm or less, fault out
I had the motor datasheet to connect the resolver properly
10 rpm = most metal cutting work on old Bport :)
I always wondered how good the simulated encoders would work
the cables are pretty critical
now you've convinced me you should hook it up to emc
I had multishield, twisted pairs
but I could not find exactly waht the motor came with. The old harness was too short. I considered purchasing one from the mfr... $$$
er $$, but I'm a tightwad
probably nothing better than that
although they might have had better flexibility than the wire you used
If I knew then what I know now, I would have bought their cable in a hearbeat
that was two years ago :-P
(I'm setting the options for analog input)
[02:47:03] <GNieport1> http://imagebin.ca/view/wzAVmsSv.html
not as good as hoped.
what are we looking at?
a really loud oscillation when a tapped <jog>
ki might be a little high
this is with their "tuned by guessing" velocity loop and EMC's position loop?
unloaded motor, correct?
could you try the 1.5v battery test? With no emc, no mechanical coupling, just a D battery (to Command+ and Command-) , the motor and an enabled amp.
no, this was with the amplifier loop defaults from the step/dir profile, and my unchanged EMC gains
tomp, yes, 15 mins to pull the coupler
that 15 min may be faster than other things, maybe
jmkasunich: I found some defaults in a different folder
who's in a huury? I've only got about 30 hours in this since Friday
you really want to do that 1.5V test though
at fest, we give away battery boxes
another 2 years, and it will be purring like a kitten
next thing you all are going to ask me to go to the basement and get the tachometer
I'm thinking you need to break out the trav-a-dial
50 mph wind gusts, not sounding too happy outside
where you at?
home of the bungles
I was there over xmas
okay motor is loose, will you all settle for a 1.5V command instead of me finding a battery and taping wires to it?
in fact, I've been refraining from suggesting a way to check the velocity loop tuning using EMC
not really EMC, just HAL
i hope the command doesnt try closing any loop. you should ignore any velocity or position feedback for now.
siggen square wave -> dac -> amp -> motor -> encoder -> amp -> encoder-counter -> ddt block -> halscope
if I unlinkp the dac, may I use setp to change to 1.5?
yes, that should work
I'd verify the voltage with a DMM at the DAC output terminals
(always sanity check everything)
let me put the batteries inthe tach
I lied earlier, i already had the tach out to tune the 0-10V spindle drive ;-)
okay. 1.5V on dac = 445 RPM
er, 1.5 on dac
didn't measure volts yet
Real volt = 2.75
1.5 dac * 1.875 output_scale = 2.75V out
it works thus far
nice job, EMC programming team! :)
is the rpm reasonable?
max rpm is 1600 at 10V
i say reasonable
thanks, all I did was unbolt the motor
any idea how to make Channel A - Channel B actually work on my scope?
subtract? got a tektrnx? its a button
usually you invert one channel and then use add/sum
In trying to deduct the 60Hz hum, the problem is that the scale is different between the two channels.
that scaling seems wrong to me
hum on the Vcmd signal? (and cradek is right, on tek too)
you want full scale on the dac to be 1600 RPM, right?
and 1600 RPM = 5.3 inches per second
hum is on the servo amp tuning ouput channel.
jmkasunich 5 turns per inch
never mind, my mistake
you confirmed my numbers Sat nite :)
I was thinking you were commanding 1.5 _volts_, which should have given you 240 RPM
you commanded 1.5 ips, which gave you 2.75V and 445 RPM, which is correct
* jmkasunich was confused by this: "<GNieport1> okay. 1.5V on dac = 445 RPM"
I retracted on the next line
like I said, my mistake ;-)
okay, I tuned the Kv factor up until the motor oscillated, then subtract 10%
is that a good ballpark?
they all interact
this scope is giving me nothing
too much hum?
is it the scope or the amp?
something in my wiring
does it show hum on the dac output or the amp input? or only hum on the diagnostic terminal?
you don't have anything wired to that do you?
Both amps do it. I have a ground loop somewhere
Just the scope input
try a 'cheater' on the scope ac cord, but dont connect the gnd leg
it's a two-priong cord
"The wpnders of IRC troubleshooting"
you have the scope ground clip connected to the amp ground, and the probe tip connected to the diag port, and you see hum?
"Well, ground the sucker!" lol
Yes, on the 5V square wave
connect the probe tip to the amp ground, right next to the scope ground clip - dp you still see humm?
maybe there's noise in the amp..
then its not a ground loop, and its not the scope
its real - the amp diagnostic output has hum on it
it's BS is what it is.
put headphones on it :)
nah, baby's sleeping
not hum, it just doesnt know the words
it's psychosomatic, you need a lobotomy, I'll get a saw
supposedly a calvin and hobbes quote
[03:44:58] <tomp2> http://www.atmayogi.com/node/506
DIY lobotomy ( this guy went up the nose )
I prefer chemical means if I ever need one
Okay, I changed a few oscope settings and believe I may have done a somewhat serviceable tuning job on the amplifier gains
cartoon? not nearly as good as jmkasunich undressing someone :)
EMC at PID 0,0,0
nite best o luck
twice2: I don't want to know what you are talking about
makes about 32 of us
sorry, it was,,, not productive
i mean my comment was not productive, the conversation was very much so
its afternoon, get up on time!
It could be 4PM and I'd still say the same thing =)
alex_joni: If you come across any electro-mechanical (relay) estop circuits, let me know, will ya?
Jymmmm: I have one I use usually
not sure where you can find it though.. let me look up some specs
not sure that means anything to google
heh, found it
[13:50:44] <alex_joni> http://www.google.com/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Fwww.sea.siemens.com%2Fcontrlbu%2Fupld%2Ffiles%2F3TK2801_02.pdf&ei=mICgR9WVMYWe-QLX392IDQ&usg=AFQjCNHEi0s_9TegLt2y18bV3ROpb6_-wA&sig2=9pYVjYB3EFWNidAt9PIYLg
or a shorter URL: www.sea.siemens.com/contrlbu/upld/files/3TK2801_02.pdf
alex_joni: thats a safety relay, isn't it?
yes, suitable for estop
k, thank you
Jymmmm: it's an older model than what we use nowadays.. but the newer ones are only marked by our mother company
so I'm not sure who manufactures them, or what the designation is
hmm.. a guy from germany reports he's using the LiveCD burned on a DVD
and he's getting much better performances
(purely as a live system)
finally after work
almost lunch here
cradek: around ?
did you guys see the new macbook?
anyone can send me sample dxf you working with ?
* archivist never looks at expensive shiny stuff
for dxf post testing
I only have artioscad and qcad and need more :)
I have not used dxf for about 10 years
I use solidworks now
output is good ?
or need editing ?
not connecting directly to machine yet
alex_joni, er lacks full travel keys on the keyboard :)
archivist: I rather miss the page up/down keys
micges, just remembered I did a direct via dxf to a machine here about 4 years ago, was as good as the toy machine could do
alex_joni, yes I use those a lot
nearly parsed a gerber file to a bitmap already :)
good night all
[22:23:11] <archivist_win> http://www.youtube.com/watch?v=Sebko0n6uJ0
wonder what fraisausa is cutting?
Gamma you making any progress
anyone know if there is a way to set FERROR through HAL?
there is not. it is set once from the inifile when emc starts.
SWPadnos: you around/awake?
how many pascals is in 10MPa?
could anyone suggest where I get the information to see the link between a physical pin on the m5i20 board and pins described in hal_m5i20?
ds2 you sure you mean 10MPa and not 10Pa? or 145psi?
oh crap, 1Pa, =145psi
cause 10MPa is a lot
yes, I mean 10MPa
might be a start then theres the code too
let me re-read the conversion table... I was getting some odd numbers (was reading a post claiming 10MPa bolt is a very strong bolt)
tomp: thanks I looked at that - the code there does not get me to the point of knowing whats up with the physical pin names.
then mega is 1000, so 10000Pa
is 1PSI == 6894.4Pascals?
I'm rooting around /src/hal/drivers/m5i20
cradek_ is now known as cradek
so 10MPa/6894.7573 == value in PSI, right?
wait, mega is 1000?!? arrg
that gives me a 1.45PSI bolt...that cannot be right
not even an atmospher
this is for a bolt tensile strength
$ units 10MPa psi
wait a moment, your talking about a bolts strenth
oh crap, kilo is 1000, M is 10000 right?
owhite: HOSTM54E.PIN do it for ya?
yes I'm looking at that.
how can that be a strong bolt if grade 5 (normally strong bolt) is like 50-60 (if not more) Kpsi
isn't mega 1000000?
there's onlyt one .PIN file in that directory so I figure that should be it.
yeah its a million, and yes, grade 5 is > 60ksi
but thats only in the pull, the shear is really low
okay, so I have not made a mistake converting... still puzzled as to why a 10MPa bolt is a "strong bolt"
something isnt right
1 MPa, =, 1.0x106 Pa
so a bolt with a diameter such that it has 1sq in will be able to dangle 60,000lb of stuff on it
that guy must have been talking about something else... a grade 10.9 bolt has a tensile of 1040 MPa
(per a table I found)
why I said something isnt right
i so hate metric
im betting he is looking at a small dia bolt
so he just said a bolt of 3mm dia has 10MPa wroth of holding tension..?
no, I was reading a thread on how brittleness is suppose to increase with the strength of a bolt
kind of like grade 8 is suppose to be more brittle then a grade 5 but they were talking in metric terms
did not metion size at all
maybe its a 10MPa diference in them?
Ultimate Tensile Strength 10MPa
Carbon Content 0.9%
that's from the post
that cant be right
*nod* thought I missed something, but guess not.
owhite: i remember having to have the code and the manual together to figure out the phys pin to hal name relations, oh and inserting the board, getting a hal show all then going back to the printed docs... it was not forthright
crap thats easy!
enc-01 A input is not a hal name tho, a lot of those arent hal names
tomp: check out jepler's web page. think we hit the codex.
10.9 isnt a grade its an iso spec
the format is m5i20.0.enc-01-value
they sort like hal names ( take your best gues at - or _)
gracias Jepler... ;)
I'll stick with counting tick marks on the bolt head
this is meant to be a honest question. How would I have ever found that hal_drivers page? Besides lucking out on google.
use the hal show dump for the real text
owhite: linuxcnc.org > documentation > all documentation (html) > HAL > Hardware Drivers
(that actually takes you to a lightly different, linuxcnc-styled version of the documentation)
[23:58:04] <gezar> http://www.rigging.net/Bolts.html
this is also in one of the PDFs in the Applications > CNC menu in ubuntu, but I forget which one
but I have simply memorized all the important URLs
ds2 you ever fix that lathe axis?
gezar: not yet, hurt my back a bit so I am limiting my physical activities :(