cost effective though at $90
true - I prefer the KDK style - but no one makes aynthing near small enough.
tried flea bay?
dang dog will be in high heaven tonight T-Bones
OK, 2.2.2 is installed and verified with AXIS
BBL - have to go deliver a table...
But as usual things change .... If I configure only X the system will not load complaining about missing axis.1.motor-pos-fb
that too big?
that seems like the size I have
AXA I think they call it
jmkasunich: I use the inserts on this page http://www1.mscdirect.com/CGI/NNPDFF?PMPAGE=675&PARTPG=GSDRVSM&PMT4NO=35348947&PMITEM=08443418&PMCTLG=00&PMT4TP=*LTIP
I use the bxa and cxa on my lathes
I just dug in a box and found a pair of matching MXL pulleys
better - the bore fits my encoder shaft
how good can that be?
so, if I bore one of them out to 12mm, it will go on the stud in place of the gear that is there now
the other one on the encoder
rig up a bracket that will hold the encoder and let me adjust belt tension, and I'm in like flynn
sounds like your almost ready to thread
anyone know what ini lines have to be set to HOME_INDEX_ONLY
daveengvall__: I think you set search vel to zero, index vel non-zero
hey JMK .... time to buy a lottery ticket ... :-)
I'll go try it
I'll just have to use my nuts for now I mean half nuts...
What worked on 2.1.7 doesn't work not ... 2.2.2
hmm, grabbing this thing is gonna be tricky
the timing pulley
I gotta bore it out, and face off most of the hub
I took a piece of 6061 and bored out a hole to fit
then split it
thats what I did when I needed to bore the pulleys for my steppers
just wasn't looking forward to doing it again
unless you have some V jaws
I have them now...
from flea bay
even with v-jaws I'd be grabbing it by the flanges
if I make a split sleeve, I can make it short enough to go between the flanges, and thick enough to be outside them
* jmkasunich puts the lathe spindle back together
tried homing .... it goes of searching but I gave up after two complete turns ...if it doesn't find it in one you're dead.
all I really need for now is a way to force homing .... on the non-test setup it will have index plus switch to home on.
hey dave , happy new year
daveengvall__: it sounds like your encoder indexing isn't working right
could be the encoder, the HW, or the hal config
to troubleshoot: hook up halscope to observe counts and index-enable
the driver's index-enable should be connected to axis.whatever.index-enable
that signal will go true when EMC wants to find the index
and the driver should make it go low when the index is detected
counts should go to zero at the same time
jmkasunich: didn't we test and fix motenc index homing at fest?
I think jack e. had motenc
daveengvall__: are you simulating the home switch somehow, or are you using an index-only homing configuration?
yeah - the mazak has motenc too
oh, I thought it was mesa
cradek: he's doing index-only - search vel = 0, latch-vel nonzero
stuart s. is using that successfully (on ppmc)
that is the way to invoke index only, right?
got me, I'd have to rtfm
I figured you might remember since you wtfc
dunno how far you read back, but it looks like I managed to scrounge the stuff I need to mount my spindle encoder
threading here I come (after a bit of machining, a bit of assembly, a bit of wiring, and a bit of config)
did you do something on your lathe to ensure that the toolpost is in a repeatable spot each time you mount it to the table?
or do you just not remove it?
HOME_SEARCH_VEL zero, HOME_LATCH_VEL nonzero -> index-only
good, thats what I thought
I try very hard to never remove it
harder to do when you mill on the same table
you won't be so lucky
I'm gonna have to come up with a plan
a couple dowel pins and holes in the table?
(will you want to rotate it?)
I also just looked in some boxes that I've had stashed for several years - my dad gave me some carbide insert type lathe tooling, which I'd like to be able to mount repeatably
that's trouble too
its worse than that
there is this truncated pyramid cast iron thing that supports the toolpost
that is keyed to the t-slot, so it won't rotate, just slide
but the toolpost can rotate on top of that
for CSS, your reference tool's tip should be very close to the centerline when the X axis is at 0
* archivist thinks tool point touch off
yeah that would work fine for everything but CSS
archivist stays up later than I do
CSS uses the tool table
only 1 am
yes for CSS you have to use TLO
cradek: can I touch off with the reference tool, then use the table and get good CSS results?
or are touch-off and TLO mutually incompatible
no, the reference tool has to mount so it zeroes at the axis zero
you can use touch-off (G54) to tweak your diameters but not for much else
emc needs to know the radius. all it has is axis position (and TLO)
toys to make! http://www.homemetalshopclub.org/projects/scotchx2/scotchx2.html
those other offsets are internal to the interp (for better or worse)
true - whenver I touch off, I enter the radius, but EMC doesn't know that - I could be entering anything
I'm having a thought (dangerous)
my t-slots are closed in the back, by the motor mounting plate
I could make a t-slot cover that is just the right length to touch that plate and the front of the toolpost
double duty - spacer for proper position, and keep crap out of the slot
is any integer number of 123 blocks about the right size?
wouldn't work, they won't fit down in the slot, and the plate doesn't stick up
making a piece is no big deal
cradek ... HOME_SEARCH_VEL = 0.0, HOME_LATCH_VEL = -0.05 HOME_USE_INDEX = YES , HOME_IGNORE_LIMITS = YES, HOME_INDEX_ONLY = YES
it searches but doesn't find.
daveengvall__: did you read what I read about halscoping index-enable and counts?
guess I missed that ... oh ... duh!
will do that after dinner
halscope is really powerful but takes thinking ... hard on me. ;-)
gotta go cook ...
see ya later
I bet his index-enable isn't hooked up
I like his idea of having a fake stage
whose idea, what fake stage
of course, his fake stage isn't very good
[01:47:42] <eric_U> http://cgi.ebay.com/Schneeberger-Ball-Slide-Dual-Carriages-248mm-Rail-New_W0QQitemZ130187453325QQihZ003QQcategoryZ55826QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
Off topic but still pretty cool: http://www.youtube.com/watch?v=l4S6q9OG3JE
somebody buy that
I have about 5 sets of those
I thinks its from that guy
maybe not... mine are all used
seems like an outrageously low price
(5 days 2 hours)
didn't notice that
might be a mistake
I wonder how big it actually is - nothing in the pic for scale
err... I mean the rails
I was thinking more like 1" wide and 24-30 long
* jmkasunich loses interest
yeah... the sets I have are 1" wide
I really gotta figure out how these damned offsets work
(explains the price)
too bad you can't use some other bearing on rails
I've crashed my lathe twice in the last half hour
home it - display reads zero
I'm about an inch from the work, so I say G0 (or G1) Z-0.5 to get a bit closer
and it merrily rams into the work, the display winds up at minus a lot more than 0.5
ah lathe work shifts
mind if I try to help you jmk?
okay, how are you setting where Z zero is?
so far I haven't - all I've done is homed Z
Z has no home switch, so homing it simply sets machine zero to the current position
because I dont know emc good enough, I have to sorta let you teach me this, what I do know is a bunch of ways do stuff with cnc lathes
there might be (probably is) a G54 offset left over from the previious time I ran the machine
and I'm sure thats the problem
yes thats exactly it
but IMO, if there is such an offset, then when I hit home, the DRO should display the G54'ed value, not zero
try this, no, home is a machine position
it cancles out g54
however, if during an operation g54 is implied Im betting
if I home, and the display reads 0, I should be able to MDI G0Z0 and have it not move
so do this with mdi, g0 g54 z5.
then compare your position to the work zero, and adjust the g54, however, try go z5. before you include the g54 and see if the resulting move is the same
gezar - my problem isn't in my methods
i know its not
my problem is that the DRO is displaying G53 coordinates, even though G54 is active and MDI commands go to G54 coordinates
I can't see how that is a good thing
are g54 and g53 the same in the offset table?
g53 are machine coords
g54 is an offset coord system, and right now the offset is not zero
I think its a bug that the display is using G53 coords while G54 is in effect
does emc have a "work shift" type command like a g10 i think?
yes, thats whats used to touch off - it sets G54 (or whichever coord system you specify)
gezer - this is a very simple issue - immediately after homing, the DRO is displaying G53 coordinates even though G54 is in effect
and your g54 is probably -.050 from the front of your work piece right?
or whatever your wanting to face off
it will be
I didn't get that far
I was moving to that spot so I could issue the G10 which sets G54 = 0 at that spot
but when I tried the move it crashed into the work
because the old G54 = 0 was inside the workpiece, and the DRO didn't tell me that I was at about Z=2 in the G54 system
the DRO said I was at zero
oh hold on
EMC remembers G54 offsets from previous runs
did you set your tool to the new parts work face before setting the new g10 for the new g54?
I NEVER GOT THAT FAR!!!!!
my very first move after homing the machine crashed it
im not trying to be rude or offensive im just openly thinking
I understand exactly what happened now (didn't before)
I homed it, G54 was in effect and I was at about 3" in that coordinate system, but the DRO was displaying G53 (machine) coords, and showing 0.0
I saw 0.0, said "I need to move a half inch left" and MDI'ed G0 Z-0.5
since I was actually at Z=3.something, that started a 3.5" move
would the dro show different numbers if you set it to relative instead of machine?
I believe it is at relative
yep, it is
if you switch it to relative is it the same numbers?
it is already at relative, and has been all along
now that I've issued a G54 manually it is displaying the proper (G54) numbers
I think this is a bug - it doesn't display the right coords immediately after homing (might be only after the very first time you home)
i didnt mean to be of no use
I need to talk to cradek or jepler about his
it might be a bug, or it might be that I'm stupid and its doing what its supposed to do
gonna see if I can reliably repeat the "bug"
(without repeating the crash)
this is odd, should I be able to hear the motion signals through the sound card?
like a plain beep or just noise?
like the frequencys being generated are litterly exiting the sound card in my linux box, enter the sound card on my other box that I sit in front of, then come out to the headphones I have on
if I adjust the feed rate override, I can hear the change in the frequency, ie the pitch gets lower as the default program runs
so the two computers are independant?
could be ground loop-related (or something like that)
I would look for grounding problem
ive been hearing strange things form the sound card lately, and this is the first time with emc running and my headphones on
you could also install ferrite cores on some cables to reduce parasitic interference
well, the noise comes across wether emc is runing or not, but when emc is, Is when I can hear pulses
do you hear other noises that could be generated by your linux box?
yeah, odd sounds
I used to hear the hard drive seeking in the sound card of the same computer
yeah its sorta like that type of stuff
but when emc runs, its as if I have my ear pressed up to a motor
but across two computers, that's new to me
well I have the output from this box emc, connect to a sound card input on edge
in a line-in input?
so I dont have to wear multiple headphones
and if you disconnect this link?
do you still hear the motors?
well, yeah, the sound on edge goes away, but If I plug in to emc i can hear it
with the headphones that is
wait, with speakers it doesn't do it?
i dont own any speekers
but it would if I had them
well if you hear the hard drive seeking, I doubt there's much you can do
its neat I can hear acc/dcel
maybe you could try to put all inputs to mute
on the linux box
yeah I tried that
when i have the volume down is when I can hear the axis motion really good
does your second box runs linux as well?
I know there's a feature in linux where you can have devices that are shared over the network
so you could feed the "sounds" from the linux box to the other one over the network
there is just something wrong with my emc boxes mobo I think
I stalled buying a monitor on sale, now they sold out
maybe a bad ground or something I dont know
hopefully one of the stores has a blowout sale on 19" monitors tomorrow
anyone know any good ebay sniping programs?
Pluto Servo Implementation
SWPadnos u alive?
Does anyone have any experience with the Pluto servo board?
if you mean "anyone", yes.... if you mean "anyone who is in the channel right now", not so sure
sorry, didn't mean to be a smartass
I've seen some discussion that it's not working in the current emc?
no problem .. sort of new to this for of communication
I thought that was pluto stepper that had issues (its alpha quality at the moment)
Guest587: no problem
sometimes people are actually working on a machine
usually its better to just ask your question
it might be something that a person can answer, it might not
so you might not get an answer for 30 minutes
if you ask "does anyone have experiance", they have no clue what your real question is
so unless they are true experts and can answer even tough questions they stay quiet
hi Gamma-X - yes, I'm alive
well i have been trying to implement the pluto servo for some time now and am a little stuck on tuning up my motors, re-configuring some of my enable inputs and overall tweaking of the machine.
thats pretty general
ok , I need to change the enable inputs on each axis from FALSE to TRUE in the HAL so that is active low instead of high at start up
you want the external hardware pins to be active low instead of active high?
do the pluto outputs have -invert parameters?
(you use halcmd show to find that out)
Jeff had mentioned that the DIR output (in the pluto-step config) is XOR'ed with the bit output it shares
this isn't pluto step (I hope)
if the same is true for the servo enable lines, then you would set the bit output that corresponds to the ENABLE line to 1
I don't know if it's true though
I don't either
err - the same
I don't know the specifics of the pluto at all
but inverting a digital output should be pretty much the same regardless of the output device
there should be a parameter something like pluto.0.digital-out-N-invert
I think my problem is I dont how to change the halcmd parameters
while your EMC is running, open a shell
and type "halcmd show param"
that will give you the whole list
"halcmd show param pluto" will give you only the ones that start with pluto
"halcmd show param *-invert" will give you only the ones that end with -invert
if you find a -invert parameter, its value will probably be zerp
you can set it with "halcmd setp whatever-the-name-is 1"
halcmd show param "pluto*invert" will give you all pluto invert parameters
and you can put "setp whatever-the-name-is 1" in your hal file to make that happen when you start EMC
yeah i can see that parameter i want to chage in the axis under the hal tree ... I will try changing it this way... is there a graphical way of changing it???
I don't know
use a graphical terminal ;)
do not fear the command line ;-)
I haven't looked at all the pluto FPGA source, but it looks as though the enables are just I/O bits that are connected to the motion controller enable outputs
yeah i think thats where my problem lies.. I can chage things fine from axis and the ini and hal files however digging deaper is the problem
ah ok - the pluto enables are input bits that enable the PWM. if you want an external enable signal (to a motor drive, for example), you connect the enable signal to the output pin as well
it looks like the HAL bit inputs are internal to the driver
yeah thats what i need the motor drive need the enable signal so that everytime i turn on my servo amps the motors dont start running
internal to the driver???
yes, "driver" in this case being the software that lets EMC communicate with the pluto board
oh ok .. so can i change this information in the driver then?
first, the axis enables should be off when you start EMC, so the pluto should output no PWM until you enable the machine (machine on turns that on, I think)
if you need an "active low" signal to enable your drives, then all you need to do is change the -invert parameter of whatever pin is used for the motor enables
like `halcmd setp pluto-servo.dout.09-invert true`
yeah so then what we have just been discussing is the correct procedure.. changing the function in the halcmd
heh - that's not a function ;)
so you would add that setp line to the hal file in your EMC config
I don't think there's any GUI you can use (other than a GUI text editor) to make changes to the HAL file
oh ok i can... so them it load that information everytime i load my configuration starts...
ok well that solves one of my problems
what gui test editor would you recommend?
go to Applications -> Accessories -> Text Editor
or Programs - whatever the main menu is
oh just for editing the .hal file
oh ok i have already been doing that for the pluto servo.hal
i just didnt realize that you can add `halcmd setp pluto-servo.dout.09-invert true` into that file to change the HAL information
ah - in that case, wherever the axis.n.amp-enable-out signal is connected is around where you want to stick the invert setting
leave "halcmd" out of that file though
that's for a terminal
so just "setp whatever-the-name-is-invert 1" (without the quotes)
"halcmd" is a program, and "setp pluto-servo.dout.09-invert true" is a command that tells halcmd to do something
right ok then im not confused then the halcmd was only for the terminal
a .hal file is a list of a bunch of commands that you want to do when your machine starts up
right its all making sense now
ok so my next question is any good reccomendation for tuning motors ?
sorry - couldn't resist
do u have adam1 's email adress
Gamma-X, when was that conversation?
around what date
I don't see any email from him
swp do u remember what he did at all?
I do remember discussing using linear scales, but I think I only helped get him going - like where to connect the feedback, etc.
im tryin to figure out the pin layout of the connector going to the control.
I think he did any further configuration
oh, I don't know any of that
did he wind up gettin rid of the scales and usin encoders?
u see him around at all?
I don't think so lately
how would I figure out what connector does what? lol there hooked into a pcb inside the servo amp box?
a PCB that goes to other connectors, or a PCB that does something? (ie, it has components on it)
some are goin to power it looks like and the others just spread like wildfire
ok, that may make it a little easier
well, I don't know enough about these controls to even begin - I don't know what connector you're talking about actually ;)
its like a 50 pin mil connector hooked into the back of an anilam
it's best if you can identify the major parts inside the control - servo drives, feedback conditioning (if any), power supply, etc.
i doubt theres a wiring diagram...
also spindle controls and limit/home switches
SWPadnos thanks for the help and info i'll keep you posted
theres a crap laod of stuff in there lol
Guest587, ok - good luck
Gamma-X, sounds like more work for you then ;)
there has to be a wiring diagram somewhere, ask anilam
SWPadnos sorry got disconnected
did u say anything? its just a maze in that box! lol
no, but eric_U had a good point - ask anilam
your best bet is to leave as much alone as you can
if a wire comes out of the box, you should be able to leave that alone
i want to just take 5 pin connector and hook most of the pins into the mesa cards
not really a good idea
do u think that anilam crusader wiring diagram can tell me where those leads are and what they do?
eric_U y not?
because someone took a lot of care to design that wiring
you want to pick it off the machine at the easiest place
I think he's talking about the connector between the machine and the control
that sounds like the best spot to me (if it can work)
it's probably nicely connected to a terminal strip inside
and there are probably relays and such that he wants to keep
no no - there are two boxes, with a cable between them
thats why i want to just hook up a female adapter to that 50 pin mil connector and bolt it to a computer case...
I think the idea is to disconnect the cable fromthe Anilam control, and make an EMC-driven connector that matches the pinout
then wire it as needed while still retaining all parts intact
I'd want to see that pinout
it sounds like a good spot, if all the info can be found
and what's inside the second box
that's the box you want to keep, correct?
did you ever post pics of the inside of this box?
im just gunna sell the control box parts on ebay
make a quick buck
i just dont know how to figure out what wire is what, im tryin to think ahead so if anilam dont know.
if they don't know, then I'd scrap it and just use the iron
the only reason for them to "forget" is if the unit sucks, IMO
they probably know since the new 3000 m is a direct replacement control for that machine.
unless the 3000 m controls servos by itself and so on.
heh - "Hello, this is Gamma-X. I hae an old xyz295 control, and I'm thinking of upgrading to the 3000-M. Can you send me schematics of the connector pinout for those two controls? Thank you."
ahhhhh thats good
actually, leave out the "to the 3000-M" part, and it's not even lying
just misleading :)
looked like a bendix mil connector
goin to bed night all!
for Gamma-X2: http://www.cnczone.com/forums/showthread.php?p=389249#post389249
man I wish I had one connector to tap into
that would be nice
there are two problems with making keyways on the lathe
1) takes a long time
2) you have to grind a tool for every key width you want to make
I haven't decided if I'm gonna do a keyway on this pulley or not
my boring head, bars, and shank came from enco yesterday
the shank doesn't fit the head, the head doesn't fit the bars
it's totally wrong, their first big screwup for me
you ordered the wrong stuff, or they shipped the wrong stuff?
the head doesn't match what it says in their catalog
oh, I read your last line as "the first big screwup for me", implying you screwed up
they say it takes 1/2" boring bars, it actually needs 12mm?
"their" screw up
its made in china, they don't know what an inch is
they say it has 7/8 threads for the shank, they're about 1 1/2
so, I cut a test bearing seat with an end mill and it seems ok
(I don't know how much out-of-round is too out-of-round)
depends on the bearings
are you making bicycles or engraving spindles?
something in between
not practical to put the part on the lathe for boring?
no, not my lathe
how far out of round do you think it is?
hm, don't drink coffee at 9pm.
why why would you say that say that?
gezar: I could feel 'some' variation with a hole gauge
if there wasn't a bearing pressed into it I'd measure it some more :-)
if its uniform around then its not so much of a problem, the bearing gets hot expands a bit into the hole, and you hope that expansion is even around the circle,
baring conditions that quickly cause problems are too big a hole, too tight a whole, not so much how round within .0005-.001
it's very much within .001
what are you machinig?
SWPadnos: NO REASON AT ALL! ok =)
cradek: you awake or have the shakes?
just still awake
cradek: Yeah, I know that one.... it gets cold here at work, so I need coffee at 1am, been getting instant decafe, not the best, bot I can go to sleep when I get off work at least
that makes me sick
I bet that's awful :-)
time to go throw up and get some sleep :)
Eh, not much choice
even Startbucks is better than instant decaf
SWPadnos: you gonna bring me a cup?
unless you want to pay for the flights
then shut the hell up! ;)
I don't THINK so!
I will shut up when I want to shut up, and not before
I think I'll shut up now. good night.
Gawd, I've created a monster
maybe you'd better have a little talk with my mother about that
SWPadnos: I already did
hmmm. you're still alive, so I figure that's not a true statement
SWPadnos: we both agree that you need an ass kickin. And she's just the one to do it too!
obviously, you've never met my mother ;)
SWPadnos: Never actually has to ben physical... she could just whine you to death
as all mom's can
oh - she's been working on that for decades
SWPadnos: But I got rope and can hog tie ya freight collect to her garage
she doesn't have a garage
SWPadnos: oh go to your room!
ok, so my real-time latency seems to have been caused by the legacy usb setting in the bios
I could have told you that.
Disba;e everything in the BIOS you don't need.
they didn't call it that
this was the most complex bios I've ever played with
it had three settings for usb, "onboard usb" "usb keyboard" and "usb mouse"
I'm assuming the last two are what they usually call "legacy usb"
and power management was rather confusing, but I think I may have had it enabled
why enable it that is
probably because I never looked in the bios before
Why do you want it enabled?
You want to disable ACPI and PC
that too =)
one of them there suns a biotches
have you ever tried to disable an APC UPS?
I had to crush one once
Disable no, disassemble yes
actually, I just crushed it after it wouldn't stop beeping
didn't have to
hit the reset button =)
zomg I made hackaday! http://www.hackaday.com/
batten down the hatches!
lerneaen_hydra: which link?
the nixie clock
fenn; yep, network use is quite intense
how's the site holding up? does it appear to load relatively quickly?
I seem to be maxing my 1 mbit connection almost constantly
not that that's so hard to do
seems slow from here :)
slashdotted slow or just slow?
opti bf20's are pretty.
hmm.. is it me, or is there a broken link in that article?
[11:54:16] <alex_joni> http://www.lerneaenhydra.net.nyud.net/index.php?option=com_content&task=view&id=33&Itemid=28
both seem to work from here
I just thought it's odd to be lernaenhydra.net.nyud.net
oh that's coral cache
that one doesn't work..
and it's the first link in the hackaday article
works for me
oh, the url doesn't work at all?
what if you remove the .nyud.net ?
then it works eventually
coral cache not working then
they of all should
I was hoping they would take most of the bandwidth hit
is there a good (automated) way of chopping up long stringy chips?
this is kinda neat - the expansion mismatch between alu/steel cancels out since the ways are symmetrical: http://www.team-haase.de/images/ebay/cut4000gg.jpg
I cringe inside every time I see a commercial product like that use extruded aluminum as ways
but pretty neat nonetheless
you could machine the little groove it sits in
yeah, but then you might as well get something out of steel instead IMO
fenn a chip breaking carbide tip
yes of course, not making the stringy chip in the first place is the best solution
I like seeing how long I can make them
and put them on the christmas tree?
our machine shop had a christmas tree made of lathe shavings
with helpers we got 10-12ft iirc
i wonder if that's how they make those dish scrubbers
fenn u remember a guy named adam1?
[16:08:45] <eric_U> http://www.cnczone.com/forums/showthread.php?p=389249#post389249
that was me asking in case you couldn't figure that out
Gamma, my buddy HelicoperJohn gave quite a rundown on the people at Anilam and who to call for what... what a guy
I finally found a picture of the antique hand plow I'm restoring, now I can assembly it correctly
fenn: i think tools can be ground to make sure the chip isnt too long , the term i recall is 'chip breaker' http://mmu.ic.polyu.edu.hk/handout/0102/0102.htm
at one place, the lathe operator made chips that ran the length of the toolroom. His name was Joma. He was very protective of his lathe "Dont make Fsck around my machine!" he'd yell
apparently the long chips and the dangerous things people would do to break them is pushing manual lathes out of production in some countries. forget if the U.S. is one.
i didnt know manual lathes were used for production in the us
every day in machine shops all over the us
I guess you'd be surprised
cant beat then for 1 offs
well. 1 off is not production eh?
one-off or production? right
you can't beat the level of education it takes to run one...
hmm i bet pressing the green button is less demanding of one's skill
machine shops in this area have about 2% cnc the rest is manual
The simplest form of chip breaker is made by grinding a groove on the tool face a few millimeters behind the cutting edge.
a jobbing shop producing to order often does one offs
then cranking on some handles
and short production runs
just finished it http://i47.photobucket.com/albums/f163/johnplctech/Garden%20Tools/HandPlow01.jpg
i was sorta hoping i could have some knurled/geared cylinders that chewed on the chips, sorta crushing and cutting them at the same time
heh and the fastest curve production is real hand turning (graver and rest on the lathe)
archivist: how's that?
(i know what a graver is)
we should produce a video clip
faster than cnc?
* fenn scoffs
yes no setup time
go make me a video clip then.. you have five minutes
no camera here
there's a big shop about 50 miles from here that has about 25 manual mills and one vmc and he has an easier time filling the manual mills with people than the vmc
they call it dumbing down around here
well at least they know how to run a manual mill
fenn www.archivist.info/clock third pic top middle all hand turned
yep any farm boy can do that
arent any farm boys left
yes there is but they drive huge tractors now
that's some clock
we added the dial to an existing movement
hand turning is faster for a hemisphere?
the brass piece holding the sphere looks hand turned..
and the moon is
heh a lunar clock
tells you what time it is on the moon
a pawl moves the moon twice a day
fenn: you could use a bent piece of cold roll rod with a dumb operator on the end ;)
tomp: for doing what?
uh, i dont get it
so it wobbles around off-center and you can't get a long chip?
no you reach in and grab it with the hook and pull it out
i'm contemplating how to build the enclosure for my small cnc lathe
its typical, seen it a zillion times
i could have a tray that slides out, or a chute that empties into a bucket
neither approach works very well with large hairballs of swarf though
and a conveyor belt seems kinda overkill
so i though maybe i could put a "garbage disposal" in the chute
if hiarball is fine enuf ( like steel wool ) then burn it
heh not bloody likely
chip breaker http://siposinc.com/
neat little angle adjuster
that was a joke, but maybe you can keep bottom open enuf to drop chips into container. just enclose upper
is this the concrete machine? ( then bottom is already closed )
actually the way it is right now, the concrete base hovers over the chip tray so i can slide it out
but when i have a big hairball sitting in the tray, it gets stuck
i think a chip breaker on the tool is the way to go
lerneaen_hydra: nice nixie!
Ziegler; thanks :)
hmm, the manual doesn't seem to document G41.1
the way I read the quick ref is that it takes the diameter (or radius) from the g-code instead of from the tool table
but it doesn't go into details - must there still be a D word? must there still be a tool table?
[19:06:37] <skunkworks> http://www.linuxcnc.org/docs/2.2/html/gcode_main.html#sub:G40,-G41,-G42:
ah, thank you
the various "orientations" (L word) are for lathe tools, right?
sorry.. I think cradek will have to answer that.. but I think so.
I remember there being pictures somewhere of the tool layouts..
me too, but not where
I'm doing a mill program right now, so I bet I can simply omit the L word
alex joni, jmkasunich - Just to let you know I finally got the automatic tool changing on the Sabre 750 finished. I've taken some video and when edited will post
I'll try it cutting air
buckie555: very cool - can't wait
Yes I'm very pleased with it - it could do with some speed optimisations but it works well
does anyone know if it's possible to preserve the value of a variable within a hal file and restore it when next run? At the moment I have to ensure that when EMC is fired up the carousel is in the correct starting position i.e. index 1 since my carousel only has an incremental encoder and nothing absolute
unfortunately no, at least not without some hackery
one piece of hackery that might work (totally untested, I'm making this up as I go along):
well shit - if you're gonna ask a question, hang around for the answer
ah, you're back
sorry - dodgy mobile 3g broadband at the moment - back again
did you hear my initial answer?
unfortunately no, at least not without some hackery
one piece of hackery that might work (totally untested, I'm making this up as I go along):
1) have a hal file that runs on exit (emc supports that, I think, but you'll need to read manuals to find out how to invoke it)
actually it could be a script that runs during the shutdown process - but it must run while HAL is still loaded and the value you want exists
it does a show command that writes the value(s) you want to a file(s)
the save command actually writes values in the form of commands that can be invoked later, but I think it either writes everything or nothing
you could use save anyway, and grep out the vars you want
man, I'm definitely making this up as I go along ;-)
that would work - what about relaoading them on start-up
are these params?
(that you want to set)
no - it's a signal
what drives it?
this is not a HAL problem, its an EMC problem
well currently it starts at 1 and then is incremented and wrapped at 21 as carousel pocket pulses arrive
you could attempt to set the signal in HAL, but EMC will always assume that tool 0 is loaded when you start up
ok, I'm starting to see what you have
there are actually two variables
oh no - I'm not concerned about what emc thinks is selected - just where my carousel is
one is the tool that EMC thinks you have loaded (or is asking for), and the other is a counter that you use to know where the carosel really is
so I'm back to my first question - what drives that signal? a counter block or something?
you cannot "set" a signal if something is driving it, cause as soon as that "something" runs, the value you set will be overwritten
hey who knew where i was supposed to email for axis suggestions?
emc dev list?
that would work
there is no axis dev list
k thank you
(that I'm aware of)
okay so emc dev list
buckie555: what drives the signal?
gone again? get some CAT5 man!
I suppose the right way would be to fit a proximity switch that defines pos 1 or whatever and then go through a rotary homing routine on startup
Neil__: are you buckie555? or just another interested observer?
no same guy - just dropped out
so, what drives the signal? what is actually counting the slots as they go by?
(you said you have this working, that shouldn't be a hard question to answer)
I have - give me 30s and I'll check my hal file
yes it's a counter with a comp to check when to wrap which resets the counter
but unfortunate since the counter can't be preloaded with anything other than zero (reset)
yes - looks like a physical index is required then
probably not such a bad thing - at least it would be bulletproof
dunno how long it takes to wrap around, hopefully not too long or it would get annoying
just OR the index with the comparator output so either one can reset the counter
it can only rotate one way?
at present yes, I plan to optimise it to go both eventually but I wanted to keep it simple until it worked
the whole counter scheme won't work going both ways, unless you add another sensor that give you something like quadrature
oh, that OR I mentioned - don't use it
if you have the index, just use the index
on the same subject is there any way for the gui (AXIS) to display a prompt to the user. I'm thinking something along the lines of "The machine is not homed - it must be homed before doing anything else"
connect something to the index-enable pin that will always drive it high
if you just set it high once, the index will only reset it the first time around
Neil__: unfortunately no
we've tossed around the idea of "runlevels" that the integrator could specify, where you tell emc the levels you support, what you can do in each level, and what conditions are required to move from one level to another
that would be cool
in your case, "not homed" could be a level where all you can do is home (and maybe jog, if you need to get clear of something before you home)
that is a significant change - definitely not something that will be in 2.2.x, and probably not in 2.3.x... .maybe in 2.4
well I've got some other requirements which might force me to modify the source and build my own specific flavour anyway
if all you want is a message on startup, that probably wouldn't be too tough in axis
if you want to enforce doing nothing else until homed, it gets more complex
axis does know what is and isn't homed though, cause it displays the little icons on the DRO
I wanted to go a little further than that and display dialogs that a user can respond to with a number of options and have those acted upon lower down
you should join the emc developers list, and discuss your plans there
(unless you are doing something proprietary)
that's a good idea - and no I'm happy to re-submit. It would be good to give something back
I've got years of realtime c++ experience on win32 but nothing on linux yet - should be an interesting learning curve
EMCs realtime is mostly plain C
I would like to add some machine vision capabilities for a couple of applications
would be neat
* archivist hopes thats for cutter location
I'll ask the same question on the developers channel but before I do can any recommend a good opensource machine vision toolkit in c or c++. I've found mimas so far but would be very interested to hear from anyone that's played with anything else?
I was thinking about a module to detect the cutter shape
I used something at university for robotic applicaiton
never done any vision stuff
it was "ok" as it was my first time with vision
I need the run out of a cutter, for the best centering of the cutter for gear millint
i think it was SDL image manipulation library
harlock - can you elaborate on your cutter shape idea?
well if you have your camera looking perpendicular to the cutter spinning, with a black or white background, you could detect the shape and get Z position at the same time
archivist: anything wrong with using a dti and rotating the spindle by hand?
toastydeath, a pain in the a
I use a microscope at the moment
i'm confused on how that's a pain in the ass
my application is for using a robotic arm to pick up an object that is oriented at an unknown angke and then accurately linish the correct face against a belt sander
toastydeath: archivist is making gears for clocks - the cutters are small
make a crush grinder!
I've seen a vid on youtube of a robot doing something similar but more elaborate with a turbine blade but they used a cmm to probe the blade before grinding to obtain the translations required
plunge grinder... not crush...
dmess: i meant crush
1/2 a thou off center on a 3mm pinion is making waste steel
to get the profile on the wheel
man how cool would that be, a little dremel-sized crush attachment
i have old CBN inserts i use for that
whittle.. shadowgraph.... whittle some more.. shadowgraph again.... close enuf...
i've created a tool post jig grinder for the manual lathe at work... so far we have mastered faces and rads with no burning of the chrome thru lpi and mag...
we tried and died with the dremmel... finally i resurected an old jig grinding attatchment form an old manual b/mill.... air driven sweet little thing...(german... probably before the war)
has under .0001" runout... i just SMILED.... from ear to ear when i seen that...
some of the old stuff is really good
we have a 1940's boley jug mill
i try to hunt for it and keep it from the scrap heap...
it has the best dividing I has seen
OLD jig mils or grinders SHOULD never goto scrap... please.. cnc them and show the world what accuracy IS..
eg its in schlesinger (however you spell it) spec
height setting is utter stupidity though, you need to undo bolts and put spacers in to lift the head frame
ive run moore & wright's.... and retro'd 1... its bomb proof to .00005".... any time.. I gaurentee it... it did .000015" for over 24 hrs in shakedown...
a later one of these http://www.lathes.co.uk/boley%20miller/index.html
BIG machine are like that.. i seen a vertical lathe that had 6 foot spacers to raise the crossrail.. it needed both of the shop cranes to handle the rail... 66 foot swing... 33 rpm max
the boley is small though!
the second one painted up is what we have
SWEET....nice machine.... can i have 1 too.. i could get the av8b wing tip gear op that costs 780.00 and takes 4 minutes..
Ive told the boss that its MINE!!!
id poke an eye out for a machine like that.... ; )
not neccesarilly mine... LOL
whats it weigh??
having xy on top of a rotating table is nice
nice option for sure... you can roll around a profile at will...
not too heavy , it was an easy crowbar move for one person
when are you on night shift next... i'll bring a truck and a friend..
use top xy to find center and offset with lower x and rotate
its good enough to make master dividing plates on
Neil__: there is another issue regarding your problem with the ATC
emc2's internal counter
I think you can always set a parameter when you change a tool (say #5899), then on startup use T#5899 M6
but that probably needs to involve some more special HAL tweaking so that nothing actually happens
could he use an index input and just (at startup) cycle the carousel until tool one's posn is found? (clunk clunk clunk cluck -found it :)
no memory needed
jmkasunich: about the runlevels that seems like a really good idea
yes - the problem at present is that the carousel has no absolute index - just an incremental one so at startup emc has no idea where the carousel is, it just has to assume position 1
tomp - that's the other option I'd considered and I think it's probably the most robust
no need to assume, add a spcl snsr to tool one and go look for it, in either direction
i did that with a 21 posn ( not emc )
Neil__: that's a pretty common setup..
Neil__: you could make emc home it while homing the machine..
but i also add gcode to select tool one before ending proggies :-) saves time, next time
anonimasu - yes see" if homing" do carousel seek routine
the other problem (I think this is what Alex said) is that EMC always assumes you have tool 0 in the spindle at startup
that's where the "special tweaks" come into play with a parameter setup
cant you write tool number into the var file?
it may make sense to have a tool number parameter that's maintained by the interp
yes, you could - that was Alex's suggestion
but it has to be restored "automagically" at startup, by the interp
hmmm. or you need a new G-code that says "this is the tool in the spindle"
I see. I guess I can get around that by ensuring that I always have a m6tx at the start of every program regardless
SWPadnos: Yes, but why should you have to worry about what tool is in the spindle?
Neil__: but.. if you leave a tool in the spindle at shutdown.. you have a problem :)
or in the position.txt file
at startup, EMC assumes tool 0
in my case where the carousel defaults to 1 if emc thinks t0 is selected it will call for a toolchange that will go through the motions but keep the same tool
if you have something other than tool 0, you have aproblem at the first toolchange
if the carousel is smart ( eg a plc ) then >it< can do the rembebering. ( and tell emc )
I don't think there's a mechanism to tell EMC what tool is in the spindle, aside from asking it to do a toolchange
I've got a signal in the hal code that stores the carousel number
it holds the number, I suspect it doesn't "store" it (ie, across restarts of EMC
hmm, a ups and special startup instruction when power fails
shutdown HAL script = halcmd save net tool-num > fred
currently at start up I just need to ensure that the machine is homed and that the carousel is at position 1
Neil__: and no tool loaded from the last time
I think that's a separate issue actually
we had a few embarassing moments at the CNC workshop, where we shut down EMC with a tool in the spindle
there are two problems: 1) EMC has no feedback to tell where the carousel is
so if for example I've been manually sequencing it then I just need to manually advance it to position 1 and then restart emc - voila carousel actual position and position stored in the hal signal now matches
on the next toolchange it tried to put two tools in one slot
2) there is no way to tell EMC (at startup, for example) what tool is actually in the spindle
any objection to adding something like M6.1 "Set Current Tool Number"?
where would it come from? var file?
that's a possibility
SWPadnos: hmm.. I'd rather see it save the current toolnumber to a var all the time
sure - the ideal is to have the interp eep track as a var
then it can restore it at startup
"the current tool number is #5678"
then home the toolchanger if you have one...
you'd also want a hal pin with that number
so the toolchanger logic knows where to put it away at
there's a slight distinction with how my setup works
ideally the current carousel position would be stored, regardless of what tool is in the spindle.
the current carousel position isn't emc's concern
the tool changer logic (ladder, hal, etc) should worry about that
no, but the tool in the machine is..
EMC cares only about what is in the spindle, and what is next
[20:36:30] <alex_joni> http://www.anthonyjudelawrence.com/wp-content/uploads/image/smackdown.gif
ok I should have said hal and not emc
oh, you mean storage over a shutdown/startup cycle?
that would make me nervous
emc definately needs a small parameter table then to store such data, not so much a load file, or acually since
emc uses a config file, then it should be actively written/read from by emc
its no different than not homing after a shutdown/startup cycle - you are gambling that nothing moved in the meantime
i agree - which is why i 've come around to the idea of homing the carousel on startup so hal knows exactly where it is
precisely - in between you could have moved an axis and in between you could have incremented the carousel
it's probably quite correct that at startup it assumes nothing and then relies on a homing sequence to figure out the state of the machine
its a tradeoff though
carousel homming is a good idea, pocket management is another that relates to homiing, but more so the management of tools I dont know if they have that in emc
we just decided that is _should_ remember the tool in the spindle
that is also subject to change while powered off
jmkasunich: yes, but unlikely accidental
don't pay to be a purse snacher in asia
right, and more importantly, there is really no way to sense the change
I mean it won't just happen that a tool fell out, and another happend to fell in it's place
it would have to jump in to place I think...
what it comes down to is that if you can sense something, you should, and if you can't, it's helpfull to remember what it was the last time
well, on that type of tool changer system, does emc put tools back in the same pockets or is it able to keep track of the tools as they are interchanged amongst them selfs?
gezar: one with barcodes/id sensors on toosl
gezar: EMC proper doesn't change tools
I've got it setup so that the same tool always goes back into the same pocket
it just says "gimme tool N"
the toolchanger is implemneted externally, either plain HAL, HAL plus classicladder, or some external PLC
external plc would be the way to go then where it handles whats what, and in the "gimmie tool N" can N be output as a bit?
a pic microcontroler could handle that no problems
doesn't emc have to know some things about each tool like lenght offset and diameter?
in my case hal and classic ladder - I was initially thinking that a hal command could write a signal value to a file and then restore it on startup - don't think this is the most robust route now though
BigJohnT: sure it does, based on the tool table
in the logic that controls the tool changer (I don't have one) don't you have to know which holder is tool 1 - n
do you leave a tool in the spindle after running a program?
BigJohnT: it doesn't matter
it only matters when you shut down emc
in the next program you have another TxxM6
kinda what I was thinking
if it's the same tool, ideally nothing happens
Neil__: how does that work for you?
that's the problem - my example is rather obtuse. If I had an absolute encoder rather than an incremental one on my carousel this wouldn't be an issue at all
Neil__ can you just do a homing on the carousel when you start up?
alex_joni - I currently get around it by ensuring that the carousel is at position 1 on startup - so that it matches the hal signal value of 1 on startup
BigJohnT - yes if I fit a proximity switch or similar to give me absolute feedback that's what I'll probably end up doing
we did a job with a 40 slot carousel and used an incremental encoder. We put a pin and a prox on the carousel and homed off of that.
any pix of the encoder on the mazak carousel, it was absolute ( lotsa proxs and holes on the wheel )
prox's read the holes, i decoded it many years agao, maybe fenn has the phots
the Mazak has 5 "bits"
the low 4 are standard binary, the fifth has the value 13
there are 25 slots, numbered from 0 to 24
5 proxes and a wheel shot full of holes and a geneva wheel so only 1 set under prox at rest
[20:54:13] <fenn> http://fennetic.net/pub/emc/mazak/imgz/LeftSide-ToolDrumEncoder.jpg
wow ancient history mr peabody
another issue related to the toolchange - i ended up going through hoops to hijack the z-axis position command during toolchange so that i could retract away from the tool change height while the carousel moves to the next required position before coming back down to pickup the new tool - the ini file currently stores a tool change position but in my case and many other vmc's the tool change has 2 positions - change height and retract height. I'm very happy t
hat what I've done works but just thought that a change could be made in emc to make it more elegant
bit 4 isn't 16 though, it's 13
why not a analog sensor?
getting analog into EMC isn't trivial - the only interfaces that support that are servo carsd (and my Mesa add-on, once I check in the code)
and only some servo cards
and jeplers newest one :)
oh right - arduino
SWPadnos; what about an ADC?
and a adc module :p
or a avr..
with the pins going into some standard digital input?
[20:58:09] <alex_joni> http://axis.unpythonic.net/01198594294
lerneaen_hydra, I was pointing out that there are very few cards with ADC that have EMC drivers
Neil__: to move to different heights via hal, you might look into what Rugludallur did. He moves the foot of a plasma cutter without using emc , just using hal. you might move the chuck up and down as needed this way.
you'd have to write a driver, and then get the number into a float instead of an S32
SWPadnos - do know whether EMC supports more than one m5i20 card installed. I can see from the manual in theory it does - wondered if there were any practical limitations?
heh use a potentiometer to sense carousel position
I haven't tried, but I can some day
tomp: I believe that is what Neil__ did
SWPadnos; ok, what if you make your own ADC card that pipes into a parport for example?
lerneaen_hydra: be my guest :)
nothing wrong with that
alex_joni; oh so there's no support for that ATM
lerneaen_hydra, yes, we all know it's *possible* to get analog values into EMC, it's just not as easy as digital, because of the lack of supported ADC hardware
tomp - yes that's what I did and it worked well - just thought it was a little clunky
when you get one done let me know
hm, there are a heap of adc cards..
though I imagine it wouldn't be very hard to do
send me a few, and I'll write drivers for them ;)
and that would be really easy to run in linux.
SWPadnos: 150eur or so..
that's fairly cheap..
SWPadnos; I was more thinking of if there already was support
sure, all the Measurement Computing carsd, most of the National Instruments cards, and various others could work
lerneaen_hydra: nope, but it's trivial
alex_joni; from your point of view it's trivial, not for me ;)
comedi supports most of those
I have a very nice A/D and D/A card that works with EMC (once I contribute the driver), but it's expensive as hell :)
a generic libcomedi hal driver would be more useful than one or two random cards
the labview-like part of RTAI
or is that scilab?
uhoh, I saw labview there
* lerneaen_hydra sprinkles the holy water
SWPadnos: which card is that?
I guess comedi is the instrumentation I/O "toolkit"
begone foul beast
mine - it's one I designed
and you need a 5i20 for it
actually, a 5i22 at the moment
and the toolchanger has tools in its pocketsessss
[21:03:30] <BigJohnT> http://linuxgazette.net/118/chong.html
SWPadnos - is more than 1 m5i20 card supported at present - I need more axis'
would that link give a low cost ADC into EMC?
Neil__, I'm not sure, but I can test sometime
Well the other card will get used in another project if not but it would be good to know
i currently have x,y,z and spindle - and need a, b as well
i suppose I'm just getting greedy now
BigJohnT: this surely does: http://axis.unpythonic.net/01198594294
Neil__: way to go!
arduino is only $35
and it knows other tricks too
Neil__: I think you can use up to 4
Neil__: did you see the videos on youtube
alex_joni - on one card yes - I thought there wouldn't be any good reason for not being able to use 2 cards
anonimasu - which ones?
[21:09:53] <anonimasu> http://www.youtube.com/watch?v=fWKYQUj5AOs
Neil__: I meant up to 4 cards in a PC
alex_joni - oh I see - thankyou very much, that's ideal
I don';t know if the driver will download the bitfile to all cards, or just the first (or the one you specified at load time)
I think if you program all the cards, the driver will see them all
but that's just speculation
would the Arduino USB board be fast enough for THC?
the code should give a hint to the sample rate
if I could set a float value in pyVCP and compare it to the analog in from the Arduino then move my Z up or down as needed?
BigJohnT: I don't think so
unlikely, un;unless the THC can handle a 1-2 ms (minimum) time lag
it's easier to do THC outside of emc, and only give up/down signals
the difference between whatyouwant and whatyou got IS the basic idea of a servo loop
you could do it with my analog board - it's fast enough
BigJohnT: you set the cutting current by hand.. right?
voltage results from the power supply characteristica, and distance to workpiece
but my point is that you most always set the current by hand..
you could also use a potentiometer to set the height you want
the dielectric constant of dirty moist air ;)
then using 2 simple opamps set as comparators you can get the up/down bits
obviously you'll want a transformer to isolate the voltage, maybe even reduce it to useable values
what's your regular voltage while cutting? 50-100V ?
the voltage signal is 0 -300 vdc
that's a bit much :D
I was planning on using a voltage divider circuit to scale it down
do you have HF ?
use a transformer..
it is a voltage signal for THC's
yay - I finally have a photo of the AIO card online: http://www.cncgear.com/images/AIO%20Card.jpg
hmm.. it's DC.. right?
I suppose I could have rotated it
SWPadnos: Got Light?
SWPadnos: looks good.. almost 800$ worth :D
on-camera flash, with dying battery
and I only charge $850 for it ;)
SWPadnos: what is it?
it's a fast 16-bit 6-input/8-output analog I/O card that plugs into a Mesa 5i22
SWPadnos: how are the ground planes?
sepparate for I and O ?
there are planes for +15V, -15V, and ground, in addition to large planes on the top (signal) layer
is that burrbrown's logo on the chips? or AD ?
yes, the analog grounds are tied to the figital ground at one point
PetPeeve... I always hate userland PCB's that don't have mounting holes in critical points... like the middle to prevent cracked traces.
the mounting holes are placed around all the connectors, since that's what usually stresses a board
tomp, actually, they're all AD, the BB chips weren't fast enough or something
SWPadnos: True, but Murphy's Law trumps any theories you might have in the subject =)
or didn't have true +/-10V bipolar operation
this way, I get to sell lots of replacements when you break your board
true bipolar missing
the AD chips are true bipolar
is it hobby priced?
good night all
SWPadnos: Or, I just say that I found lots of cracked traces on the PCB and leave itl like that.
plus the input buffers for the ADC have huge common-mode rejection
good night alex
see you Alex
Jymmm, that doesn't work with me, because I test everything before I ship it
and I pack well
SWPadnos: nice that you used removable term stips
yes, it's a pet peeve of mine :)
SWPadnos: Oh you should know buy now that you could personally install it, yet one bad verbage post can wreck havoc
do you use the single pin crimps on the wire into those? i've thought thats just an extra connection and have been placing bare wires into them
sure, that's why I don't usually sell to yokels like you ;)
is this a daighter board or connects via ribbon cable
I usually used ferrules when I wire stuff
connects via ribbon cable to the Mesa card
ferrules are my friend
ferrules, thats the word
otherwise it would be internal to the PC, which would make connecting difficult
automationdirect has them at a low price
there's a shield can that goes over the large rectangle, which contains only the analog components
is an ancient scope that doesnt really work right anymore worth keeping around?
tomp, only if you like collecting old junk that doesn't work any more
make it into asteroids :)
oops - that was for gezatr
err - gezar
dump it, and you can buy my old scope to replace it :)
I think it has a blown trace
i have an old military tek that sucks too
I have an "old" HP - a 54622D MSO (from 2001-2002 or so)
wife is wanting me to clean stuff up, I would like to have a good scope, but hell this stuff is as old as i am
it's a really bad thing to sneeze as you're taking a sip of coffee
gezar: 1966 http://cgi.ebay.com/HP-170A-Oscilloscope-Manual-TM-11-6625-535-15-1_W0QQitemZ7631545899QQcmdZViewItem
ype, i even have that too
gezar: save/sell the manual and scope for parts
is it worth my time to mess with it though?
an the plugins
gezar: what is your time worth?
I waste so much of it, not really anything
(tomp put a higher value on gezars time )
gezar: $400 ??? http://cgi.ebay.com/New-Digital-Color-Oscilloscope-OWON-PDS5022S-w-USB_W0QQitemZ140195766490QQihZ004QQcategoryZ104247QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
speaking of wasting time... bbl
its like all the bmw motorcycle parts i have, I can put it all back together but for what?
gezar: 60MHz $650 http://cgi.ebay.com/New-OWON-PDS6062T-Color-Digital-Oscilloscope-60MHz_W0QQitemZ260198919228QQihZ016QQcategoryZ104247QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
my bet is that those are shit. of course you'd only be able to prove it if you stick one next to a good scope
and measure the same signals
[21:44:30] <Jymmm> http://www.owon.com.cn/eng/pds6062.asp
yeah, you've linked there before ;)
Man U Ell http://www.owon.com.cn/eng/PDS6062T%20USER%20MANUAL.pdf
SWPadnos: at least it has real dials
could those be any good?
yep - I can't stand it when the only interface is a touchscreen or a single set of knobs that you have to swap around between channels
daveengvall__: did you get updated?
yes I did ...
man the stuff I have is massive
tuning better? or same issue?
I have tuning ... on jog only that isn't too bad.
gezar: Probably better than nothing at all, but might ask SWPadnos
f-error making sense?
it really wants to osc at 40 Hz with any kind of excuse
I'd agree with that - probably better than nothing at all
I put the scope on the index pulse and can verify it at the encoder plug and on the motenc breakout board
so possibly worth the money, if you only expect to need it for "diagnostic" purposes, versus measurement or calibration
daveengvall__: I goofed around tuning a small servo that wasn't hooked to anything (even added a flywheel) but I don't think there was enough mass to really tune well..
I hope this tunes better with a load
but I suck at tuning though - so who knows.
I keep hoping for better methods
SWPadnos: They get geeky on pg71 of the manual
JMK started something on the wiki but I think he got distracted
heh - I like the "Fault Treatment" section of the manual - starting on p. 73
SWPadnos: sorry, pg70
bugge /me needs level shifting on my parport
it ought to be possible to hook siggen -> sine -> an axis and get freq response
but then I don't know what to do from there.
yiou need a square wave to get frequency response
so siggen -> square -> pid
skunkworks: I wonder how it would work to use an induction motor as a flywheel, applying some DC to the windings to simulate friction.
can't you sweep a sine?
but what you want to see is the step response
OK and then how do you get from there to PID?
daveengvall__: I think petev added an auto-tune pid.. I think it is pidat or something like that..
square wave -> PID (position, like you'd get from the TP)
I think the results with that were somewhat variable though
hmmm. I guess I should have cleared the driveway while it was still light out
there you go. I played with it - but my period was outside what should work for that style of calculating.
a summer intern at NIST did some tuning but didn't finish the project
it worked 'ok'
yes - and it was lost. :)
bbl - heading to Costco with the wife
I need to take the next step on homing but don't know where to go.
when homing starts the axis.0.index-enable goes TRUE so it is looking for it but somehow doesn't find it.
daveengvall__; that sounds like all interns in a nutshell :D
projects are usually longer than the allowed time
daveengvall__; tell me about it...
just tuning a servo ought to be easy but ...
I take it you've read the tuning guide on the wiki
(excuse me if you've already said that earlier)
I've looked at the stuff JMK put up but didn't finish. I really need a cookbook for HAL
I think there is a real difference between tuning an STG (ISA) card and tuning with the newer (PCI) stuff.
old nuts & volts magazines worth keeping?
gezar: I think so
gezar: Better if you can scan them and put on cd
daveengvall__; oh you're not using emc to control PID?
what kind of experiences are those with 5i20's having?
sure ... but you get to set the P, I, D, FF0, FF1, FF2
gezar: gimme the name of an article in one of them plus the date
yeah, what's the problem, can't you find a stable value for the parameters?
I'm closer than I was yesterday ... in fact close enough to transfer the contoller to the machine and retune for the heavier axis
and higher resolution encoder
going off to beat my head against the homing issue.
see ya later
gezar: back issues are $5/ea
gezar: per year on cd $25/ea http://store.nutsvolts.com/home.php?cat=105
well if thats the case im just going to toss the 4 issues i have
Oh, I thought you had a whole collection of them
gezar: Too bad they're not letter sized (8.5" x 11"), then you could easily scan them to PDF.
just been testing G0 vs homing and there is a huge difference. Makes me think that when going to the home position it's not really just a G0 move
and this machine goes 450IPM so you really notice it
450ipm would be nice, I doubt I'll get more than 200IPM :(
BigJohnT: don't compare the home move to a combined X and Y G0
compare X home move to an X only G0
jmkasunich: that's what I did X15 X14 and back and forth
your G0 move also has to be long enough to let the axis accel to full speed and decel back down - shorter moves won't hit top speed
jmkasunich: when I lower the speed down the home is as smooth as the G0 moves
how much of a move is "short" depends on your vel and accel limits
I only move 0.5 from the home switch to home and was doing 1.0 lenght reversals
would there be a way to suppress a following error during homing, or allow it to reset?
I'd like to help, but right now I'm trying to wrap my brain around some G-code
that is what prompted my questions from the other day
what are you cutting?
bracket to mount my spindle encoder
first time I've used tool diameter comp
got a profile or just slots?
tool comp that is fun
two helical milled holes, a curved slot, and an outside profile
(I don't mess around)
you got a print?
I have it in easycad, haven't printed it out
or even dimensioned it
I've got it mostly coded - the holes are done, the slot and profile I've coded the path, now I'm doing the o-word loops that will rough and finish cut it
can you post it up somewhere perhaps I can send it through my software
this is a learning experience for me, so having you run it thru your cam wouldn't teach me anything
nope, just thought I'd offer
if you want to jsut for fun or to test your sw, I can post a dxf. but later, after I get this coded
I better leave you alone then so you can focus
the sw is commerical so it works...
daveengvall__ i think jmk suggested using only P to begin with , and to not use emc at first, just hal ( pretty much how youd do an oldschool analog amp ) oh, and null it 1st ( see it dont move when told to 'stay' :)
interesting stuff http://cgi.ebay.com/Phoenix-Contact-Lot-of-DIN-rail-acces-Omron-Contra-clip_W0QQitemZ130187579822QQihZ003QQcategoryZ42899QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
py pyrometer ( long skinny bit ) ?
nope, cut cable
anybody here used arc moves to begin tool diameter compensation?
I have vague memories a year or so ago of working out some of the math, and I'm almost certain it was implemented
but I'm getting arc radius mismatch errors
you starting the arc with tool comp?
G0 to a spot, G41 to turn comp on, G3 to a new spot
jmkasunich: I have but it's been a while
without comp (or with G41 D0) does your program work?
it won't work without comp
or I'm not understanding figure 10 in the manual
[23:02:29] <jmkasunich> http://www.linuxcnc.org/docview/html//gcode_tool_compensation.html
in figure 10, section 1.4.2
the distance from current point to programmed center point is not gonna be the same as the distance from programmed end point to programmed center point
if it won't work without comp it's hopeless with
they need to differ by the tool radius
please look at that drawing
that picture is bogus
you program the uncompensated path
the same arc will work with and without comp
so I should actually be programming the unlabeled point at the bottom of the drawing as my first point?
I better read this text
yeah I think this is all wrong
I feel better
imo the programmed arc is 'current point' to 'programmed end point', and the motion is 'current point' to 'destination of tool tip'
you should program an arc with radius > tool radius that enters the path tangent
this must be why I just program tool path without comp...
the distance from "current point" to center, and "programmed point" to center needs to be the same, like any other arc, right?
thats why the figure is wrong, it implies the programmed radii at start and end aren't the same
the generated compensated arc will be tangent to the path at the endpoint. direction at start and center are different
you looking at figure 10
I don't understand if figure 10 is what the programmer is supposed to do, or what emc is supposed to do, but it's bogus either way
see, I RTFM and it leads me astray
comp311_2.ngc is a great example
it shows uncomp and comp with arc entry
you can see the arc is programmed the same way
in comp311_2 they do the comp move before the arc right?
no it uses single-arc entry
on line 26 g41 g1 y4
lines 11,26 are the same; 26 is with comp
you must have a different version of the file than 2.2.2 has
I've got 2.2.2
line 26 is g3 x2 y3 i1
line 25 is g41
the center of the entry arc is unknown ( not related to your part , just a requisite of the leadin arc ) http://imagebin.org/12798
if you load it without having a tool table, the comp won't show (tool diameter is zero)
tomp: emc calculates that arc properly
I had to change it to g41.1d0.2 to get it to show both paths
the paste was about the pix not the calc
[23:15:41] <cradek> http://timeguy.com/cradek-files/emc/comp311_2.ngc.png
the highlighted line is the entry arc
yeah, thats how it looks after I put in the 41.1
tomp is right, the center of the generated arc is different
as well as its initial direction/tangency
cradek: better pic tho, mine is snarfed
yours is better pic
how to calc ctr?
don't know, I figured it out a while ago :-)
actually I think I put good comments in the source - let me see
doesn't matter, EMC calculates it
cradek: I had the wrong file loaded I had comp311.ngc not comp311_2.ngc
the i j ?
I think you are calculating the i j of the uncompensated arc
time to start the deer chili bbl
search for "imagine" haha
got it, now getting pencil & paper... :)
yeah you will sure need the picture :-)
woot - it accepted my arc (after about 10 tries)
I think I'm getting a grip on it - the manual sure didn't help
I always write the program with D0 first and then turn on comp afterward
I'm definitely not good enough to get it right in one try
I remember 3-4 of us solved this arc entry geometry problem (differently) but mine stuck because I wrote it in C :-)
jmkasunich: working on something fun?
theres the dang inv button on this gnome calc, that took 4ever to see
heh - I wrote my answer with kig, I think
which is vastly different from C