you can be sure there'd still be at least 11 by the time they learn
I say let them play on the darn thing, it's nearly impossible to hurt yourself with it anyway
the coolest toy is the elevator that rises up out of the floor
my kids used to love that one
some stupid schmuck took the bolts out of the lock hasp
and then didn't put them back in so the rest of us couldn't use that trick any more
done - hearing protectors are a wonderfull thing ;-)
far straighter and smoother than the bandsaw too
no problem and I bet you have some chips in your underwear LOL
pockets and hair anyway
I always end up with a few down my shirt
The big trick is to have the min number of teeth in the material to prevent heat build up and gumming
since this is so thin, I went the other way
just barely raised the blade above the table
it didn't try and raise the material up as you cut?
the only way it would lift the material is if the trailing edge of the blade was trying to cut
the more teeth in contact with the material the more force it takes to hold it down.
the blade is pulling it down
if it is high enough the force needed is mostly pushing as you lower the blade the angle of cut gets real shallow and tends to push the material up out of the blade.
what thickness did you cut
that's pretty thin, I've cut up to 3/4 and when you get a little thicker the forces become more pronouced
1/4 is pretty easy but 1/2 gets to be a bear if the blade is not all the way up
ah, that I can understand
glad you got it cut
well talk to you guys later gotta go eat dinner
good night all
time for me to find some dinner too
cradek: you around?
what kind of two sided tape did you use for cutting PCBs?
I'm gonna get some tape to hold the rubber keyboard to the mounting plate, might as well get something that I can use in the future
I think it's for holding down tiles
the kind that works the best has filaments in it
you have any brand name or product name?
has a little more thickness than the non-filament kind
'manco' - has a picture of a baby duck? thing?
that sounds like "duck" tape
[01:00:47] <jmkasunich> http://www.stickwithhenkel.com/
<-- formerly manco
this one? http://www.duckproducts.com/products/detail.asp?catid=1&subid=7&plid=729
ah, maybe this one: http://www.duckproducts.com/products/detail.asp?catid=1&subid=7&plid=27
double sided, fiberglass
yes I'm sure that last one is it
the green box looks familiar
(the box of mine is long gone)
any markings on the inside of the cardboard core?
or jsut the duck?
just the brand name/address and duck
thats probably a specialty store thing isn't it - carpet or tile store?
or just go to home depot?
I've found it at several generic hardware stores
I'm sure I'll discover something else I need
[01:12:42] <toastydeath> http://www.stanleyvidmar.com/Vidmar/Products/Vertical+Lift+Systems
right after I get back
i can't figure out how you'd keep the trays upright
or right after they close, more likely
in that system
ha i figured it out
Whats shakin in here on a saturday night?
I've been at the body worlds exhibit all day.
Looking at dead plastified peel open bodies.
is that a car thing or the plasticized people thing?
ah - that one
It was pretty cool
what city are you in/near?
Made me wanna quit smoking
NC, I assume?
np - I figured you didn't mean Charlotte, VT
Where are you? How did you hear about body worlds?
I'm in Essex Junction VT (slightly northeast of Charlotte)
I've seen the Body thing in several cities I've visited in the last couple of years
ads anyway, haven't been to the exhibit
You should check it out
yep - just have to do it on a "good day"
That Gunther dude is gotta be one morbid somebiatch
it's a little odd to consider that you're looking at dead people, not plastic models of them
Yeah it is odd.
They arent in glass either, you can youch them
But you could
Man I told the ol' lady about your servos. I think I'm gonna have to buy a few off ya for experimental purposes.
I need to find a source of cheap industrial scrap/
old ball screws etc
I have some acme screws from a Bridgeport, plus I think I have a couple of short ballscrews
don't have a clue what I'd sell them for though
Well most folks seem to think highly of their ball screws.
Not a cheap component.
that's why I have the acme screws, I put ballscrews in my BP
I just love having some of everything.
I could be a white Fred Sanford.
or black, for all I know
I think I might spend some time tonight trying to coax some coherent control out of these floppy motors I have.
I still dont have my drive chips.
[02:22:56] <jmkasunich> http://jmkasunich.com/cgi-bin/blosxom/shoptask/got-keys-12-08-07.html
Local biz sucks. I'm ordering online from now on.
its starting to look like a CNC
It's the computer lmao
The computer is starting to look like a cnc
the computer is still on the bench 8 feet away
it's even almost grayish-metallic colored
computer goes in the box on the back
you can see the X axis stepper at the bottom of that pic
is it trying to hide?
2:1 belt reduction is in the yellow box
the round hole is where the leadscrew clutch/reverse lever used to be
(when it used change gears for power feed)
jmk, what do you think of that machine? Is it rigid enough to be useful?
define usefull ;-)
people do usefull things with sherlines that you can pick up with one hand
all machines can be useful
this is more rigid than that
but it is NOT a rigid machine by real machine standards
Well yeah, are you a machinist by trade?
i've seen some ridiculous setups on those machines
I've made a bridgeport whip in the wind like a noodle
no, I'm an electrical engineer
however, I've done enough work on the side using that machine to pay for it about 5 times over
Have you used a shoptask toasty?
and dmess is, if he were in here
it isn't a bad lathe
People keep telling me I'm gonna kill those little machines case Im used to big heavy duty stuff.
if you are used to machines that a manufacturing firm would have used at any point in history
Heres the one I hope to have by too long. http://www.lathemaster.com/LATHEMASTER8x14Lathe.htm
then yes, no matter how great of a small machine you buy, you are going to be a bit hard on it
Dogfishguzzler: this was made on that machine: http://www.metalworking.com/dropbox/_2001_retired_files/Worm04.jpg
everything except the gear teeth, I did that on the van norman
Right on jmk, nice work.
dogfishguzzler: what's your price range
Right now a grand.
Ive spent ALOT of time looking at small lathes and that one has gotten great reviews
Plus it's american made and damn near tool room quality.
It's not a hardinge but.....
and what exactly is "tool room quality"
are you SURE its american made?
like i don't want to be too big of an asshat but there's only so much capacity you can get out of that style of machine
you can machine whatever you want
Toasty: It will typically have less than .0002 taper out of the box, its rigid at 200#'s
jmk: Yep Im sure.
Thats the question now isnt it
I assume over the entire length of the bed, 12 inches.
i doubt that
in a very big way.
Dogfishguzzler: show me where it says that is an american lathe
Yeah I kinda do too but for a thousand bucks its as good as I can do.
its an import, I'm about 99.98% sure
okay jmk brb
if you were talking about a swiss lathe, i could accept .0001" in 12
but even a hlv or 10ee isn't holding that.
any lathe or mill you get will do work.
but it's not going to be fast work.
that number is probably what you get if you stick a dowel in a collet and indicate it right at the front of the collet
(on a good day)
like the first thing that tells me it isn't rigid
is that the distance between the bedways is smaller than bed to spindle centerline
that's like, the very first thing you look at
the second thing is that the carrage isn't wide enough
you can do a ton of stuff with it, just not very fast.
you may not have a comfort level outside the capacity of the machine, in which case you will be a very happy owner.
Yeah I figured it was gonna be a slow worker
fast is the difference between hobbiests and pros
they aren't selling that lathe to pros ;-)
jmk: I can't seem to find confirmation of it being american made either. I'm sure I read that though.
that's where a lot of people are happy using hobby machines
like jmkasunich alluded to, it's comfort level
i am very comfortable cranking a tool in .5" and running it .040" per rev
in mild steel
that machine would not enjoy doing that one bit
I've been a machinist all my life, even since I was a fetus. I have to get myself some machines or I'm gonna go insanae. If I could have three phase here I would just go to an auction.
but it would enjoy doing .1" at like .01"
one of the people reviewing it was very pleased to be able to do 0.040 depth of cut
and in the end the parts are going to be the same
probably a few thou per rev
dogfishguzzler: then get it!
Dogfishguzzler: do you have 240V single phase?
(stove or dryer outlet?)
Yeah I love popping off a nice chip and I know the hobby stuff aint gonna do that.
jmk: Your talking about buying a phase converter.
actually, about building one
Yeah I've thought about that
i personally can't stand to work slow, so i'm saving for bigger iron
rotarys can be done on the cheap with a surplus motor
the only other thing is getting the machine here.
toastydeath: you like "chips that go clink" ?
not necessarily, i just hate taking 80 passes
i'd rather take one rough pass, a semifinisher, and a finish pass
or just one, if it's not a lot to go
if I want to do heavy work, I do it on the van norman (mill work)
or the 13" south bend in the model shop at work
we have a 3 hp leblond at school that i use
it's like 16"x40"
very cool lathe, accurate as hel
the SB at work used to be mine
but since it was in the garage, I never used it
so I sold it to work
haha and now you can use ut
which is cool, since I can still use it
I should try to convince my friend to CNC his Takisawa
if i need to turn big stuff there's a 20 hp gearhead mori seki CNC at work
that doesn't run too often
and a 7hp manual Birmingham 19"x60"
about the same quality as the Leblond
very, very nice import brand
but priced for the quality.
yep, japanese (semi-obviously) brand, good stuff
his is a 5HP unit, weighs about 3500 pounds
[02:44:24] <toastydeath> http://www.fw.ucalgary.ca/sciwksp/images/Takisawa.jpg
yep, that's more or less the one
cool, I think I used all 2HP at once for the first time
also, I have chips in my shoes
I've actually had my BP spindle get shut down from the overloads in the scontactor
that's never good
you'd think that would be hard with a manual machine
I don't see any way the control can notice when that happens...
is your manual 2HP too?
like, you need to crank the tool through the work, so it should get difficult as you're loading the spindle more
I think it's a 1.5 actually
i've never STOPPED a spindle except on the Haas at school
i've always had vibration problems first on a bport
That reminds me, If Im threading with EMC is it counting on the RPMs being right or is it getting some feedback from the spindle?
I have another motor to stick on it with a VFD, I think that one is 2 or 3 HP
I cant wait
Driving me crazy
Thats not funny
How would you like if you have no machinery?
I do have one lathe actually, lemme see if I can find a pic.
i don't have any machines either man
i live in an apartment
[02:52:11] <Dogfishguzzler> http://lh6.google.com/compuslave/Rr25lC14t6I/AAAAAAAAAQw/Lc1sQC-oHh4/s800/temporaryLathe2.jpg
Damn Toasty, I'm sorry dude.
it's okay, i work in a machine shop
trying to save for an american pacemaker while paying for school, etc
Goodtimes right here. http://lh5.google.com/compuslave/Rr25Yy14tmI/AAAAAAAAAOI/91tk34ofLAw/s800/HaasCarnage.jpg
How old are you Toasty?
oh damn you got all kinds of time.
nice carnage tehre
hah yeah i do
[02:56:04] <Dogfishguzzler> http://picasaweb.google.com/compuslave/VariousPhone
it just bugs me to not be able to walk out into a shop and be able to make whatever, whenever
Yep me too
And after work I'm in no mood to hang around
I love the trade though. I love making stuff.
Wait till you guys see my perpetual motion machine.
I am seriously close to closing the loop on a steorn type device though.
It's not perpetual motion persey but it will run for days
how about i just say "i'll believe it when the patent is filed sucessfully and scrutinzed by the scientific community"
No patent, GPL
If I nail it then it goes public domain.
then the second part of that stands!
Right on, I hear ya. I don't know what to think of Steorn's claims.
I find it hard to believe that the laws of physics somehow dont apply but we'll see if anything ever comes of their claims
I dont claim such a thing. But I do have a magnet motor based on some of their public patents that will run by it's self.
i'd just like to build a tool storage system instead of a toolbox
You need to save your money and buy an insert vending machine
That's a joke ofcourse.
What do you mean 'tool storage system'?
it's got shelves inside
on a vertical ring
press a shelf number or enter a tool name
and it rotates to bring the proper bin up to the access port
oh snap, that would be alright.
there are pictures of really big ones
[03:08:14] <toastydeath> http://www.armedforces.co.uk/newimages/amhdatapik.jpg
just want a little toolbox sized one
they're not that complicated internally
That would be a cool project
put some car batteries in the bottom for when you disconnect the power
and use one of those low-power embedded things to run it
I want to learn more about embedded systems etc.
they have a ton of classes at my school
Whcih school is that?
itty bitty community college.
real heavy engineering emphasis
At my age I think hands on is the only way I can learn.
well everything they do has a lab component
they keep their machine shop in running condition and have fluid power labs
and other crap
it's well stocked for hands on
If I were rich I would goto school constantly.
But working AND going to school is just too much.
I love learning stuff. Im not sure why it's not as if it produces anything most times.
I think I need a shrink.
that first sentence made no sense.
I spend lots of time learning useless stuff.
Does that make sense?
well cut it out?
try to learn useful stuff.
As we say here in the south 'I'm eat up'
It's interesting stuff but generally useless.
So what's your deal with EMC? Do you use it at school?
i don't use it at all
You hang here because your fellow machinist are here?
it seems to be a neat group of folks that like machining and getting projects done
Same here, but I do tinker with EMC.
cradek: you owe us photos or a blog post of what you are working on
I took a little video but I'm sure it is awful
just pics of the current state would be interesting
* jmkasunich waffles
its time to cut a hole in the electrical box
hard to un-cut if I put it in the wrong place
(besides which I'm not looking forward to jigsawing thru all that steel)
#ubuntu is a helpful place sometimes but most times it's so busy you catch hell trying to get a question answered.
i really prefer manpages and google for linux help
hi all http://www.biertijd.com/mediaplayer/fullscreen.html
this is my NEXT adventure
haas has nicer displays than they did when we bought our mill
the nice haas controls, i dunno man
i don't like the control itself
but it has some real nice features
it's very bittersweet
and i am having sort of a crush on these vertical storage systems
if i can build a real simple one that would alleviate a lot of the "home shop" issues
i'm going to make one of those
i'm thinking all you'd need is a 3ph motor and a toggle switch
i already have everything in uniformly sized 12"x17"x4" rectangular plastic tubs
naw it'll be computerized with a search engine
that would be the ideal
but i know myself and i won't get past the "build the simple version"
maybe i'll add on
well, just having everything in boxes works pretty well
well taking advantage of vertical space
but sometimes the categorization scheme breaks down
and it's then that i really wish i had a search engine
are you considering a vertical lift system or a vertical carousel
robot or just wheel of crap
Anyone recognize this quill drive on this clausing cnc mill?http://cgi.ebay.com/1990s-CLAUSING-KONDIA-PROTOTRAK-CNC-VERTICAL-MILL_W0QQitemZ150192956759QQihZ005QQcategoryZ12584QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
i would do wheel of crap
well, i'm moving into a room with a single lane swimming pool (empty)
currently that space is being wasted, so i'm going to try to take advantage of it somehow
cover it up and install a robot, clearly!
i found pictures of the linkage they use to keep the bins upright in the carousel
and they're just triangle linkages
on a revolving chain
care to share those pics?
sure, lemme find it again
[07:44:48] <toastydeath> http://www.diamondphoenix.com/images/verticalcarousel/verticalcarouselChain_lg.jpg
what you want sounds more like a vertical lift system rather than a carousel
the lift systems are the ones with the picker robots
that ride up and down on rails, and grab crap off shelves
yeah i'd rather have something with less mechanical structure
i think the pickers are harder to implement/error prone
the carousel has high uptime but is slower
i think something like a forklift that shuffled the boxes around would work
that's pretty much what they use
then it places the correct box on an elevator that sends it up to me
its pretty much what i do by hand
some folks apparently have 20' long carousels
that have multiple loops
and they keep all their bar stock on them
really heavy duty carousels
forklift drives up, types in the job number, the right stock comes around, and off they go
delivery works the same way but from the other side of the carousel
supposedly in japan they have underground parking garages that store the cars on carousels
i think i saw videos of that!
those are ridiculous
"apparently the company who sold the software to the Hoboken Parking garage was not fully paid and stopped updating the software regularly causing the malfunction"
"so the lesson to NYC is you can fire a human w/o pay but you cant fire a computer system w/o paying first"
the real high-tech solution is better public transit
yeah, i wish we had that around here
like some kind of pod that can link together into trains and then go on a rail or tube system
or just some trains
in the first place
that works but i dont think americans will go for it
we won't go for much
"what? don't eat so much? nonsense."
trains don't work so well if you commute from suburbia
and you cant exactly un-build suburbia
europe/japan never had enough land in the first place to have this problem
[07:56:35] <fenn> http://gizmodo.com/gadgets/gadgets/vws-automated-parking-garage-173049.php
Alex you around?
would it be feaseable to use an M41/M42 command to shift the scale/PID values for the spindle drive?
[10:29:45] <alex_joni> http://www.diamondphoenix.com/images/verticalcarousel/verti
hmm, what's M41/42?
M41 = low gear - M42 is High gear M42 P# could be used for someone with a lathe with a full gearbox
I would use
I would use M62/63 (digital IO's)
then do the stuff in hal based on the outputs you just set
it also allows for reversing the spindle direction to correct with way is CW for a Bridgeport type head mill
that last one should be fairly simple in hal
just set an invert bit, which drives a scale = -1 on the spindle speed (if you use analog output), or sets a parport.*.pin-invert if you drive it digitally
right but you want to use the M62 to tell HAL?
it's already there, and working :)
what command are they using on the Mazak to shift gears?
I don't think they use any commands
it's done in ladder based on spindle speed
In my case I have to shift manually.
OK most new machines auto shift based on the Sxxxx Value
I'm a bit against adding some more g-codes for things that can be done with general solutions
what I would like to have is user-defined mappings
so that you could define: M41 = M62P1
but then most don't use gear boxes anymore - they use multi winding motors
M42 = M64P2.. whatever
this should be done with o-words procedures
so you could in theory define whole sequences
any gearhead machine i've used provided m-codes for the gears
(just throwing that out there)
one for auto select, and then one for each gear
toastydeath: and I bet it wasn't the same codes for all those machines
actually it has been
all in like, the 40's
exactly the same?
so far, yes
but like i kind of agree with you?
that adding codes indiscriminately for features is a silly idea
too many possibilities to please everyone.
I think the more standard solution is to perfer..
I have a Hwacheon that will take any Sxxxx value - but if you don't define the gear with a M41/M42 it will stay in that gear
toastydeath: that's why I propose the o-word substitution
what is o-word sub
a procedure defined by the user
think a M98 P macro
maybe in a special file read by the interpreter at startup
dude might it be useful to be able to define m-codes in a config
toastydeath: yeah, that's kinda what I mean
but a bit more complex than that
imagine I want to make a sub for Txx M6
(for a rack tool changer)
my procedure would have the tool number as the input, then generate the proper XYZ moves, etc
and afterwards do the actual changing
i was thinking more like, an external thing entirely
like, the code (you guys use python right?)
map the m-code to a bit of python code
because, for instance
we have pallet changing mills
that use the same range of m-codes (40's)
for pallet crap
it's not python for this kind of stuff :)
O subs can do powerful things - almost dangerous if I understand right.
Skullworks-PGAB: most anything you can program :)
well how would you write a g-code program
loops, compares, decisions, etc
to do a pallet swap?
toastydeath: same as you probably would write a python procedure
Pallet swap is a job for CL
[10:45:32] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Oword
yeah, we have o-word subs in some of our machines
it's just parametric programming
and i suppose you could hack a way to do a pallet change by writing code to watch pins or whatever
Next topic - Know anything about Puma.kins?
toastydeath: right.. it won't be pretty but it should work
Skullworks-PGAB: a little
any program examples out there
Skullworks-PGAB: what do you mean?
[10:48:34] <Skullworks-PGAB> http://www.skullworks.net/include/DSCF0272.JPG
I run these at work
Skullworks-PGAB: me too :)
I have a divx vid here somewhere too
not stäubli .. but anyways
it's actually 'ae'
not sure if it came over right :)
they bought out Puma
Skullworks-PGAB: it depends what you want to do :)
anyway - we just use the teaching pendant to define the moves
the robots I work with are for arc-welding
right, same on most/all I've seen so far
there is a teach-in script in emc2
sweet - MiG or TiG?
both, but mainly MIG
sometimes plasma cutting
how much fixturing does it take to get the robot to pick parts up right
or is there some sort of vision thing going on
toastydeath: depends on the parts
sometimes it's simple gripping from predefined points
sometimes it's vision based de-palletizing
(those two are quite different)
the predefined points thing probably wouldn't work for us
toastydeath: sometimes it's even more complicated than that :)
we have these random sized boxes we stick the parts on
but it gets expensive quite fast
we don't use vision directly on the robots - but the robot does hold the part in a vision optical measuring cell to be sure any punch press features are currect.
like 2 stage localisations (1 overhead mounted camera to roughly locate the boxes, then a camera/laser scanner on the robot arm that find the accurate position of the box to pick it up)
like it would be easy for us to define a small spot to put the boxes in
but it would be the robot's responsiblity to locate the individual part and pick it up
they'd all be aligned kind of neatly
Skullworks-PGAB: to get back to puma/6R types
toastydeath: no problems :)
Skullworks-PGAB: if I happen to start emc2 with a puma-typed robot, I'd probably exchange the intepreter
I don't think G-codes fit an articulated robot
(but that's probably not a very easy thing to do..)
well I was wondering how EMC would deal with a puma type - being 6 axis + romtes for clamping ect.
well.. g-code supports XYZABC.. so the 6 axes is taken care of
you can even extend it to 9 axes with UVW
(e.g. linear actuators that move the robot around)
the problem is writing the g-code :)
right - but movement is different in different modes - world/joint/tool etc
Skullworks-PGAB: I know.. that doesn't really work right now
it only moves in world
just wondering if you knew of anyone playing with that
G0 should move in joint world imo
yup, I do..
there's a guy in germany that built one himself
he does it like this:
fires emc2 up with kins
and uses the teach-in script
I was thinking of making one 1/3 scale
he joggs in world (after homing), and uses the teach-in to record axes/joint locations
afterwards he adapts the stored points into a program (by hand, iirc)
then he starts up another emc2 config without kins, taking the joints only program
and runs it
he said he's quite happy how it works
(this hack he did is because of some still to be resolved bugs in emc2)
I have no prob hand writing out huge 4 axis G-Code but the way things reverse and all on these robots... Which way is up again?
Skullworks-PGAB: haha, yeah :)
I guess for it to really work we would need: a different interpreter, a different GUI (at least for the teach-in stuff), fixing emc2 bugs/misconceptions
are those bugs just in pumakins or what?
fenn: no.. conceptual bugs
don't think its the KINS as much as the UI
for example.. the max-speed, max-accel are used in world space
instead of limiting each joint
ok i think i understand how that method works around the bug
we have 16 puma type, 2 GMF of 1 type and another smaller GMF unit.
what language is used to program robots? (did i ask this already?)
fenn: it's manufacturer specific
I'm thinking of making a scaled down version for giggles
and it doesn't have any resemblence from one manuf. to another
(some people build model steam engines, I build machines...)
but steam engines _are_ machines
anyway, every nerd on the planet is thinking of building a scaled down robot arm
fenn: I'm not :)
but I am pondering about getting an older big-sized one
um, you're not a nerd maybe
yes - But I could earn major geek points to have this unit with vaccuum grips changing out optical media on my PC's...
Skullworks-PGAB: yeah, yeah.. that scores really high
Skullworks-PGAB: well.. if you do start building.. maybe it wouldn't be so hard to build directly two
I know a couple of emc2 developers who would really love to have one (and will probably make it work really good with emc2 :P)
the GMF's are much simpler than the puma types
I keep forgetting to shoot some good pics of those
acronym doesn't ring a bell
still RRRRRR ?
there was a shot of one in the first Terminator movies... yes there that old...
[11:15:34] <alex_joni> http://www.wiu.edu/engrtech/images/robot.jpg
Skullworks-PGAB: here's one I installed http://www.robcon.ro/300730/sIMG_8774.JPG
ok - thats half size of our small GMF
here's another one http://www.robcon.ro/301589/sIMG_8759.JPG
the robots we have are always the same size (roughly), they get extended with linear axes for big parts
you don't need/want really big robots for 30 pound payload :)
these pumas are not too big - and are really light weigth thin wall aluminum castings
roughly 4-5m working range diameter
just well designed to handle loads within there design parameters
Skullworks-PGAB: still scary
when they start to move with 2-3m/sec
ours are a good 10 years old for the young ones
Has anyone reported any strange behaviour when homing to an index pulse using a m5i20 board?
buckie555: not lately
homing in general had bugs in 2.2.1 IIRC
Skullworks-PGAB: not homing, the config files generated by stepconf
I'm pretty sure it was working in 2.1.7 but since I've upgraded to 2.2.1 when I try to home it seeks correctly, backs off correctly, finds the index pulse and then a second later send that axis into violent random movement
something got pooched but is fixed now
if i home with index param turned off it works fine
oh.. that's unexpected
well random in the sense that it commands the axis full speed one way then full speed back and forth a few times - it's a little scary
buckie555: right, I think I know what's happening
but it shouldn't have been changed from 2.1.x to 2.2.x
is it likely to affect the x and y axis as well, I'm scared off trying for damaging the mill
what do you think is happening?
fenn: it resets the axis position
and motion doesn't compensate for that
I definately had x and y homing off an index correctly in 2.1.7 but wasn't using the index on z at the time
so it sees a big jump in position, and tries to catch up
buckie555: where are you located? US?
no - UK
it would be best if we try to get a grip on jmkasunich later on
he should know this stuff the best
otoh, I'm trying to see if I can find anything right now..
for the short term I can just rely on the switch contact but a fix would be very welcome
I can come back online later today when jmk might be around if that helps. By the way I checked the index pulses in halscope and they all appear nice and clean.
for the time being I've disconnected the z axis motor coupling to prevent any further damage so am in a position to test any potential bug fixes before I reconnect
buckie555: ok, this helps as a bugreport
good.. that's nice
good work on checking the signals - assume nothing.
could you try to get a halscope snapshot of what is happening?
the movement is so severe that it overshoots my separate limit switches and hits the mechanical stops
sure, other than the z index what other channels would you like me to monitor?
there should be a cmd and fb position for the z axis
i know the ones, anything else?
trigger on index
there are some homing related (I think homing-stage or something like that)
ok fine, I'll try to get some shots
buckie555: been going through a big list of diff's .. can't find anything atm
could you also post your ini's (both 2.1.x and 2.2.x) somewhere?
buckie555: specifically.. do you have m5i20.<board-id>.enc-xx-reset or reset-count, in your config?
buckie555: I think it might be a configuration issue related to m5i20 (after studying the code)
or a bug in the m5i20 driver
* Skullworks-PGAB is going to catch a few hours sleep.
alex are you still there - sorry just popped out of the office
no problem.. still here
buckie555: can you look at your .hal file?
did you adapt it to match: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UPDATING#m5i20_encoder_changes
not right at this moment but I can remember it quite well - i've spent most of the last month playing with it
"The m5i20 "encoder" pins now match the HAL canonical encoder interface. The -cnt-latch, -pos-latch, -idx-latch, and -latch-index pins have been removed. The standard encoder pin -index-enable has been added. The behavior of the -position and -count pins now match the HAL canonical interface."
yes, i copied across the two lines from the sample config that link m5i20.0.index-enable <==> to axis.2.index-enable etc
yes i have no references to those deprecated pins or to encoder reset or reset count that you mentioned earlier
also my home switches definately stay on from the moment they are tripped to the limit of the travel
ok, the m5i20.*-count would also be valuable on the halscope snapshot
I rigorously checked the operation of the switches following observing this behaviour and they switch clean on and off every time
buckie555: I don't doubt they do :)
since i disconnected the coupling I'm now activating the switches manually
i cmd home, it starts seeking, then wait a few seconds, activate the home switch, the motor begins the latch sequence, then i release the switch to simulate it coming off and shortly after it detects the index pulse, then about 1 second later all the craziness starts
buckie555: to prevent harm I would limit the PID output for now..
but if you think the motor is ok with the coupling removed, then keep it like this..
that's a good idea - will do
one thing i did notice was that "sometimes" after the first attempt with the strange behaviour and then me aborting by pressing escape the next attempt succeeds
i say sometimes because about 50% of the time it exhibits the same behaviour the 2nd time
buckie555: maybe you abort it after it managed to zero the encoder count
oh absolutely - i have no intention of reconnecting the coupling until i'm confident this is fixed - it's already stripped the locknut off my original ballscrew causing the whole z-axis to fall to the floor
then the next time it won't see such a big jump..
not a nice experience with a 4 tonne vmc
quite - luckily the y axis guards took most of the impact saving the spindle and table from damage
as it turned out it was a blessing in disguise as the ballnut was badly worn so the z axis ballscrew needed replacing anyway but that's another story
now I've installed a brand new replacement i don't intend damaging it
is there a way of saving a trace from halscope or do i just need to do a screengrab
I think in 2.2.x you can save a trace
but a screen grab is faster for now
03tissf 07v2_2_branch * 10emc2/docs/src/hal/ (10 files): French translation
03tissf 07v2_2_branch * 10emc2/docs/src/ (Master_HAL_fr.lyx Submakefile index.tmpl index_fr.tmpl): French translation
03tissf 07TRUNK * 10emc2/docs/src/Master_User_fr.lyx: French translation update
buckie555: gotta run for a little while, will read when I get back
03tissf 07v2_2_branch * 10emc2/docs/src/Master_User_fr.lyx: French translation update
03tissf 07v2_2_branch * 10emc2/docs/src/gcode/main_fr.lyx: French translation update
03tissf 07TRUNK * 10emc2/docs/src/gcode/main_fr.lyx: French translation update
Anyone know what causes the popup window Sorry I stepped out when on here?
BigJohnT: it's a gaim thing
there's a timeout for the idle in the settings somewhere..
you can also disable it completely
found it thanks
BTW, face.py version 1.3.1 is up on emcwiki now and includes fractions in the entry fields
cool, saw the list announce
(btw, you were right, there's a limit on the upload size..)
The current upload size limit is 215040.
I'm just a tad over that...
I'll try just zipping the files without the installer
that's way better than including executables :)
zipping is worse it's 4.32mb and the installer one is2.53mb
hmm, you have to press 'generate g-code' before 'write to axis and quit'
also, you're cutting with a g0 move
all the y moves
they are all off of the material
you sure bout that? :)
fenn: I was until a second ago
now you have me wondering...
oh, i get it now
it still seems dangerous
in what way?
if you aren't cutting a square piece of stock, or if your stock isnt clamped down square
or any number of things
perhaps I could offer the option to G0 or G1 between cuts as the user requires it
sorry i've been thinking about pocketing code all week
Oh yes for pockets no G0
what are you doing with pocketing code
that was my next fun project
er.. writing a thing that takes dxf and cuts spiral pockets
from the outlines of circles and other shapes?
i have to investigate how dxf entities get grouped into paths
but the idea is to be able to do a pocket from any arbitrary shape
I've been looking into converting dxf for my plasma routines same but different
its not just for dxf files, but svg, arbitrary python functions, any vector format really
i'll have to figure out what's worth doing of course
but qcad is the only real cad program i've got, so dxf looks mighty tempting
I found a python file that generates dxf files so the code might fill in some of the blanks as to the format of the dxf file
is that german?
i thought you were going to say PythonCAD
I think it is the same one
looking for the link
maybe just an old version
I have both pycad and pythoncad
let me look i think it was a different one
here it is CADvas http://members.localnet.com/~blanding/cadvas/
ah yes, the successor to pythoncad
well, something like that
anyway it's much easier to use
for now my next project in python was to do a simple pocketing g code generator
go for it - i'll probably never finish
too many irons in the fire???
dang that is worse than I thought
which comes first, finish the cnc lathe or start the cam software?
you gotta finish the chicken then worry about the eggs
at least with the software it may be useful to others in a half-finished state
and it's warm up here
I'm in the middle of building a portable CNC plasma cutter
where is here
cool, something like a boom on a pivot?
here is upstairs, rather than in the cold basement
no it's made with 80/20 so it's light and easy to move around
this is what i pictured: http://www.cnccookbook.com/img/CADDrawings/PlasmaJib.jpg
has a regular xyz axis
That's kinda neat, how do you handle the xy moves with a linear and rotary?
x = r*cos(theta), y = r*sin(theta)
you'd have to write a kinematics modules i guess.. hack up scarakins.c
that's beyond me at the moment
what is theta
the rest i understand
the rotary axis
the angle of the rotary axis
ok understand now
that seems simple enough at this level LOL
thought I had some pictures of the drive for the Y axis but they are at my shop...
the cnccookbook link as some neat info on it thank fenn
I'm thinking about enabling my drives when I turn on the machine in emc
alex_joni - it's neil here again, are you online?
yes, we were just discussing your issue
managed to get a halscope snapshot?
hi - oh great - yes you can view the screenshot and hal logs at www.freedrive.com/members/8il35w2d5sowc/
it looks as though motor-pos-fb exhibits a glitch when the index pulse is detected
the glitch is not present when home with index is set to false
what a great tool halscope is for debugging such issues
Guest619: it seems as if pos-fb is beeing reset on index (which is good), but pos-cmd doesn't (which causes the jump)
yes i see what you mean
i can upload my hal and ini files as well if it helps
the "glitch" is supposed to happen
that is the encoder resetting itself to zero when it detects the index pulse
the problem is that the commanded position from EMC should also reset to zero, and it doesn't seem to be doing that
Guest619: are you on the EMC pc right now?
or is there some sneakernet involved?
no - it's in the workshop, i'm back at home unfortunately. What were you going to suggest - I have a copy of my entire configuration that i can upload if required
I was going to ask you to get more details from the scope plts
in particular, the exact value of home-state during each step of the process
jmkasunich: there's a log of the halscope
yes i uploaded them as well
* alex_joni wonders if halscope can load logs
I can't seem to actually open the log file (even as text)
as a slight aside - I'm using the embedded java irc client. Can anyone recommend a good client for use on kubuntu?
jmkasunich: there's a download file at the top of the second window
Guest619: it would also be very valuable if we would see the m5i20.*.index-enable pin
jmkasunich: what if the driver doesn't set the pin low.. I think the outcome would be the same as on the scope plot?
yes, far more valuable than the index pin
I was initially monitoring that pin and I'm pretty sure it remained at 0 throughout the entire sequence
index enable? it cannot remain at zero through the whole sequence
not if you are using index
Guest619: xchat or konversation (irc clients)
that initially attracted my attention because the hal file is explicited linked to index-enable so i guess the linking to the index pin is implicit
no, the index pin is a almost useless thing
the actual index signal from the encoder goes straight to the hardware
if you are moving at a high speed, the pulse on index may be very short - the hardware can detect it, but software can't
the driver provides the index pin, soley as a debugging aid - it is a copy of the hardware pin
but no guarantees of you seeing short pulses on it - it might go hi and low again between reads of the board
the index-enable pin is the one that matters
if i command a straightforward f1 move i see index pin go high once per rev es expected but have never seen index-enable high
emc drives it high, to tell the hardware "next time an index arrives, reset the counter and drive index-enable low"
index enable will only go high once, during homing, after you hit the switch
it will go low again when the next index pulse arrives
i was initially sampling at the servo rate of 1ms but increased the sampling rate for scope just so i could get a bigger window - 4s wasn't enough
it will never change state when you are running normally
oh, that is not good
you MUST acquire a sample every servo period if we are going to be able to get to the bottom of this
trigger on index-enable?
you don't need to see the entire homing sequence - you can trigger on index-enable (or, if it is truly not going high, on home-state)
oh ok, I guess i'll have to temporarily increase the servo period then - which has motor tuning implications i believe
no no no
no, leave the servo-period alone
don't change periods
run the scope at one sample per period
we only care about the 50ms or so immediately surrounding the index pulse
ok, so I'll leave the servo period at 1ms and use that thread for halscope
this is why I said "argh" when you said you're at home now
interactive debugging only works when its interactive
if index-enable isn't going high and i trigger off home-state what value of state should i be triggering on?
lemme check the code
I can get another plot posted within a couple of hours. Is there anything else you'd like me to check whilst I'm there?
is that computer on the internet?
ok so trigger at > 14
lemme put the complete list of states on pastebin
no - no network conn I'm afraid - and no web access at the workshop at the moment
so you have to put it on a USB stick and bring it home, and there is no way to talk to you while you are there....
[18:43:56] <jmkasunich> http://www.pastebin.ca/810135
that list tells you what EMC is doing based on home-state
Guest619: isn't there another PC networked around the workshop?
no not at the moment. There will be in a week or so but not at the moment. Yes currently I'm tooing and frooing with everything on a usb stick
Guest619: darn :)
I have a emc setup here for testing but not with the mesa cards!
maybe jmkasunich has an idea what else to log & record
well, we'll do the best we can that way. put "internet connection for the shop" at the top of the to-do list
well I don't mind it taking a couple of days. - Definately
home state, pos fb, pos cmd, and index-enable, are the key items
sample once per servo period, and trigger on index enable going high (which should happen as state goes to 15 or 16
(some states are invisible, because EMC is there for less than a complete servo cycle - 15 might be one of those
so you might see it jump from 10 or 12 to 16
but >14 should work for the trigger
Guest619: you're on dapper.. right?
If it helps your debugging i'm 99.9% sure that when I was running 2.1.7 (I'm now on 2.2.1) x and y homed correctly off the index but at the time I didn't have the z connected for testing
14 right or index-enable > high
yep on dapper - standard ubuntu 2.1.6 iso install - upgrade to 2.1.7 then upgrade to 2.2.1
500Mhz PC with 256MB ram
do you remember to try 2.1.6 ?
especially related to the index-enable stuff
i believe homing off an index was broken in 2.1.6 - i think i can remember seeing some chat about it being fixed in 2.1.7 which prompted me to upgrade
it definately didn't work on x and y in 2.1.6 but then did in 2.1.7
[18:51:47] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/hal/drivers/hal_m5i20.c?graph=1
ok, that's what I wanted to hear
the above is the revision graph for the driver
Guest619: could you try to use the 2.1.7 driver with the new emc2 install?
the bold numbers like 1.17 and 188.8.131.52 are version numbers for that specific file
2.2 uses 1.17 of the driver, 2.1.7 uses 184.108.40.206
at the moment i have the z axis coupling disconnected so can test this without fear of damaging anything - my x and y are connected though so don't really want to risk testing them at the moment
Guest619: I uploaded the 2.1.7 version of the driver at http://wiki.linuxcnc.org/uploads/hal_m5i20_217.ko
Guest619: download that file, and put it on the USB stick
ok where does it need to go?
then copy it to your machine (to /usr/realtime-2.6.15-magma/modules/emc2) and rename it to hal_m5i20.ko
remember to rename the file that is there to something like hal_m5i20_222.ko
for both operations you need root rights so sudo will be needed
ok no problem
Guest619: you are only using the first three or four encoder counters, right?
(there is a known problem in the FPGA for the last four)
yes just 0,1,2 for x,y,z. I'm planning on using 3 for spindle encoder feedback but haven't connected it yet
FYI. I'm using all of the IO available from the 2 7i33 cards
that is a lot of I/O
yes - atc uses most of it
Guest619: this is my project. http://www.electronicsam.com/images/KandT/conversion/toolchangerspindle.JPG
hi folks, i guess you guys might know a bit about robotics then..
and the possibilities of a 3 axis mill, a high speed precision air bearing and a spare 2.2Ghz peecee..
have the hardware almost up together, looking at interfacing and controllong it all now.
depends on the mill.. steppers or servos?
skunkworks - looks like an exciting project.
what are your speed requirements, what's your SCALE (counts / unit), etc
steppers, mostly from a scrapped boregrinding machine.
right I'm off to try this out - will report back in a couple of hours
Guest619: no problem using lots of I/O, as long as you stick to encoder channels 0-3
so you should be fine
Guest619: I should be here when you get back
thanks for the heads up
I might also be here for hours < 4
ok - really appreciate your support
also a sony linear measuring device that has some odd output. looks like manchester encoding on a scope but cant quite get it;
I'm in the US, so I should be here till well after sane UK people are sleeping
Well if we miss each other I'll try again tomorrow. As soon as I have the new screenshots I'll upload them to the same place
Guest619: just make a small comment in here
Ah I stopped being sane when I quit the day job to work for myself!
with my name on the line, I'll notice it later :)
same goes for the other people around I think
right - be sure to mention each of us by name, that will highlight your comment for each of us
getting approx 1100 counts per mm on linear scale.
angle sensor has binary output, about 850 counts per revolution, .
air bearing runs at 36000rpm on compressed air at 80psi.
if your using steppers - you really don't need any sort of feedback.. (although someone on here is using encoders to detect following error with steppers)
what are these air bearings being used for? spindle?
hopefully will be able to make precision components up to 7in cubed
airbearing used for precision grinding or drilling
have had some success making small parts for miniature steam trains and signalling components.
but all manual control, takes ages.. looking to make it repeatable cnc :)
mechanical pcb etching?
had a go with pcbs turned out a nice board, and nice 0.3mm drilling
the air bearing is this sorta thing. airbearings.co.uk/product_range.php
but is FederalMogul / Westwinds type gas turbine airbearing, not electrically operated.
works on gas pressure & airflow.
cant find pic on their site tho..
basically high speed router/drill .
can mill pcb sheet up to 6mm thick but that real slow, 3mm pcb + copper one side can mill about 3inch a minute.
so wont be fast but will be high precision... (other project is precise milling of resonant cavities for antennas)
ok, who wants a free Zip drive
does it have the click of death?
John, what signal should I hook my drive enable up to in HAL?
I think there is something like axis.n.enable? do a hal show sig to see what is available
oops, hal show pin, not sig
the sigs don't exist until your config creates them
i was gonna say, no sigs :)
I see all the m5i20 pins
"show pin axis" will show just the ones that start with axis
(emc has to be running when you do this of course)
not just halcmd
that would explain a few things
start emc in one shell
run halcmd in another
thats the standard way to examine a running EMC/HAL setups
if I am running emc2 from a terminal, are there arguments?
actually you can continue running emc from an icon, just open a shell for halcmd
(I'm used to doing everything from shell)
doing it remotely, never done that before
oops scripts didn't like realtime stuff already being loaded
start emc first, that loads the RT
then just run halcmd by itself
no scripts, no halrun
I had been playing with the halpy scripts
which doesn't unload cleanly
this looks promising: axis.2.amp-enable-out ==> Zenable
so what was the conclusion on the m5i20 driver?
if I get my mill running, it's actually connected to the internet
he's gonna grab some more traces and we'll take another shot at figuing it out
I'm currently running 2.2.1 though
my home switches are currently connected with duct tape, so I can move them
I am insane
I'm sitting here thinking "that zip drive head actuator mechanism is cool, I should save that and use it for something someday"
jmkasunich: welcome to the club :)
what kind of actuator is it?
linear voice coil
I saved the heads off an ibm hard drive because they are big
different than a harddrive head mechenism?
no, but they are huge
harddrive is rotary
zip is linear - a tiny hardened and polished rod is the guide
I always thought it was funny that they asked the iomega engineers where they wanted to live
probably thinking somewhere in utah
answer was colorado
I probably would have said Park City
cool, amp-enable works exactly the way I want it to work
damn, wish I had a pop-riviter
you can borrow mine
or 1 of mine
only $16.97.... seems like a good excuse to own another tool
* jmkasunich is doing radical computer case surgery
what are you mounting in there?
I'm shrinking it
mounting the power supply above the cpu side of the mobo, and the disk above the expansion slots
chopping and channeling your rod
and cutting the rest off
final size is gonna be about 13.5 x 10.5 x 8.5
but she might heat up too...
there aren't gonna be any covers or anything
the whole contraption goes into a 24 x 30 x 9" hoffman style box
and that has a filtered air inlet with a blower
so your just using the guts of the case as a way to mount everything inside of your enclosure
[20:44:41] <jmkasunich> http://jmkasunich.com/shoptask/box-02-13-07.jpg
the PC goes in the lower right
cabling from the maching comes in the middle of the right side wall
don't you love the button hole plugs
the amazing part is that they were already on the box when I bought it
whoever surplused that box must have read my mind
we use a ton of them at our shop when we rebuild equipment
I wouldn't want to buy those new on a hobby budget tho
I got that one for $50 from a surplus place
I see $30 worth of hole plugs in the picture
on the same trip I got a 14 x 16 x 6 box for $25
it had a 50 pin ribbon cable to screws breakout board inside, suitable for use with a mesa board
the breakout sells for $40 ;-)
you ever use this stuff http://web4.automationdirect.com/adc/Shopping/Catalog/Terminal_Blocks_-a-_Wiring_Solutions/Wire_Duct_-a-_Accessories/T1_Series_Wire_Duct_with_Cover/T1-1010G-1
at work, yes
I dunno if I'm gonna use it for this project or not
I might just go with ty-wrap pads and ty-wraps
sometimes it just gets in the way sometimes its great
we use a ton of spiral wrap too that is some neat stuff
[20:52:21] <BigJohnT> http://web4.automationdirect.com/adc/Shopping/Catalog/Terminal_Blocks_-a-_Wiring_Solutions/Wire_Duct_-a-_Accessories/Spiral_Wrap/P-2
its cost effective and we use it to tame cords, wires, pneumatic tubing etc
<Dr Nick>: Hi Everybody!
I sure hope it gets a few degrees warmer (or even colder), or we're gonna have some very unhappy trees
[21:42:14] <jmkasunich> http://jmkasunich.com/pics/icy-branches-1737.jpg
* skunkworks_ read that as icy-britches
it's raining here and the wife is making deer balls
not touching that one
think meat balls but very lean
around here - eveyone mixes a ton of pork with thier deer meat
she make these with straight deer meat and they are good and good for you
if you process it correctly it doesn't taste gamey
BigJohnT: where are you located?
Poplar Bluff, MO
shiver me timbers but don't it get cold up there
my buddy is from WI and goes up there all the time
he lives in St Louis now
heh. It has been cold. 15 deg f right now.
it's like 40 here and very moist
that is t-shirt weather around here.
* alex_joni heads to bed
good night all
any one have any comments on this http://www.water-4-fuel.com/
actually, fraud seems more accurage
just a a fraud
it's just a small scale hydrogen/oxygen generator that takes more power to run that it give back to the engine...
your better off with a wood gas generator for your car LOL
its actually $97 to download a "book" that tells you how to make that thing
interesting that they show bound books, but what they are selling is a pdf
as pt says there is a sucker born every minute
they don't even pay for the paper - 100% profit
ok now face.py has a Inch MM radiobutton...
that puts G20 or G21 at the start of the file
speaking of pc'
pc's before I am planning on mounting my gecko's and power supply in a pc box
with my breakout board
should be nice an portable
and for windows users I've added a shooting solution for face.py to emcwiki
i have video of the car running on FULL HHO
best use for hydrogen is to allow you to lean your mixture, full hydrogen production is a scam
but it could increase milage for sure...
even just the water vapor wave would smooth and increase power as does water injection an even an old bake and break engine...
alex_joni or jmkasunich a new homing plot is online at http://www.freedrive.com/files/8il35w2d5sowc/Homing2sIndexEnable2.png
there's also a log file of the same name in the folder
looking at it now
index enable is doing exactly what it should
as is pos fb
but pos cmd should be jumping to zero, and it isn't
* jmkasunich is perplexed
cause this has worked for a long time
gonna look at the code and see what I can find
do you have your ini and hal files?
fyi the other driver version that alex_joni provided was exactly the same file size as the one that was already on my system
I just thought of something that could explain what we're seeing, but I need to see your hal files
i did try using it after performing the test you've just seen but insmod refused to insert it - I wonder whether it's compatible with 2.2.1
do you have your ini and hal files?
no problem, give me a couple of minutes and I'll upload them
ok they're uploaded - some of the servo links are moved from core_servo to sabre_toolchange - the comments in sabre_toolchange will explain why
my head is spinning as I try to figure out all the offsets that are being added and subtracted in this code
i know it's a bit of a pain - basically there are two offsets - one for tool change pos and one for tool retract pos
no, I'm talking about lower level offsets down in the motion controller
home offset, motor offset, backlash correction offset, and various sums and differences of them
home offset is 0, i thought i had backlash set to set as well, not sure what you mean by the others?
that is helpfull info
they the other stuff is internal, I'm just thinking out loud
are you talking about references in the ini file or hal files
neither - if you are curious, see http://www.linuxcnc.org/docview/html//code_Code_Notes.html#r3_2
for a block diagram of the stuff I'm talking about
part of my tool change logic in sabre_toolchange hijacks the cmd and fb pins but is only activated during tool change operations and anyway it works fine when homing without index
just thought i'd point it out incase it's relevant
machine zero is supposed to be at the index pulse right after the home switch? (that is what home offset = zero implies)
yep that's what i was expecting
I think I spotted the problem
it's an issue with the order functions are added to the thread
I'm all ears
nbuck: I bet it only is an issue for the Z axis though
thats what I was thinking, but then I thought again, and figured it didn't matter
so I got into the code and never checked the order
you have a couple of muxes between Zpos-cmd and dac
and a mux between Zpos-fb and motion controller
the ones you use to hijack the position for the toolchanging
I was right that order between driver and control wouldn't do that, but the mux sure could
delay, yes, but not screwing up the index-enable handshape
but with a mux between the encoder and control, that would do it
wonder how it worked on 2.1.7 then..
the order MUST be - read encoder, evaulate fb mux, run control, evaluate cmd mux, run PID, etc
ah ok - suddenly light coming on
jmkasunich: did we change the default order for a simple addf ?
addf mux2.1 servo-thread
its hard to figure out the thread order looking at multiple hal files, wish we could see "halcmd show thread"
by default, that adds to end of thread
if you add a number to the end, it puts the function at position N in the thread
if the number is negative, it puts it N from the end of the thread
although using that syntax is brittle and should be avoided IMO
Yes - I had deliberately defined them in the order shown - guess i just screwed up
better to do all addfs for a thread in one place in one hal file
nbuck: I haven't yet figured out what order you actually have them in, so this is still just speculation
yes I think I'll combine everything into 1 file - i was planning on doing that in the end
does the sabre_toolchange hal file add the muxes?
but if this were the case wouldn't it cause homing without the index pulse to exhibit the same behaviour
yes it does
that means that mux probably runs last in the thread
when not using index, the thread order error only causes a 1mS delay
when not using index, the encoder position only gets remembered (not reset)
thus you won't have the jump..
but when you are using index, it means that when the encoder sets its output to zero and tells the motion controller about it (by way of index enable going false), motion still sees the old non-zero value, and calculates the offset according to that
so if you don't home to an index pulse then home offset has no effect?
it has an effect
nbuck: it does.. just gets applied differently lets say
but the calculations are different
without offset, the position coming from the encoder doesn't change abruptly, instead motion simply subtracts out whatever it was to get to the right position
well this would certainly explain why i thought it was working on on x and y - because it probably was!
no muxes there?
no that's right
ok, my suggestion is to merge your hal files, and get things in the right order
well don't I feel stupid - I'm really sorry for tying up both of your time for what appears to be pilot error
yes will definately do that
it sounds like you understand the issues ( which is a non-trivial thing, and don't feel bad about it tripping you up)
nbuck: it's a good excersise at least :)
and it's not resolved.. (so far it's only a guess..)
yes - implementing the tool change has taught me a whole lot about the hal and ladder
one of the things on my "that would be nice" list would be for threads to determine which functions feed which other functions, and put them in the right order automagically
well tomorrow I'll tweak the order and merge the files and see if it fixes it
and let us know
automagically would be nice but then I guess you potentially lose an element of control
after loading your config but before trying to run it, do "halcmd show thread" to check the order
will do - and of course you'll be the first to know
in fact, you should load the old config and so "show thread" to see if the mux is really after control
nbuck: in the "inprobable" event it doesn't run.. halcmd save > file
I'm pretty sure the mux is after it
that sure is consistent with what we're seeing
* jmkasunich goes to make dinner
infact looking at the screenplot again I think you can see the delay between pos and cmd
* alex_joni goes to bed
this time for real :)
alex_joni: do you have that
thanks again guys - will report back tomorrow
alex_joni: sorry, typo. have a good night
thanks, you too