Hello, i work in a very small shop in norfolk/eastern Virginia.. we cut pile driving jaws, generally around 9 inch, with an old bridgeport milling machine
ive only been working with the machine for a month maybe
and i have no prior knowledge of any sort of machine work
im a fitter/welder
but bossman sat me in the stool and showed me the process to face off and cut jaws
well, the process we have may be flawed
tooling could possibly be wrong
if anyone with any experience with this sort of machine and the tooling required, help me out please, otherwise im sorry for my waste of bandwidth.
by the way, my boss is also a welder, he welds the height of the jaw back up
and we really are in the dark here
some pictures would help
i was hoping i was writing all of this in a channel
but i can take my digital camera in tomorrow
still have... i think 20 teeth to cut in a fixed jaw
i will say
its alot of guess work
i would probably just do it with an angle grinder
i mean, you're going to be slamming it into dirt
doesn't matter if it's precise
we charge a pretty penny
pretty perfect jaws
ah well if that's your business model then keep at it
but they are never exact
they pay to have them almost new
lets say that
bossman has some nice contracts ;p
but eh, what really kills me
is turning the head on a 45, to cut teeth
[00:35:55] <fenn> http://4.bp.blogspot.com/_C5baAbh5DW4/SHb2XhDdR2I/AAAAAAAAACE/1DnARRoyU58/S220/shark+piledriver.jpg
you have to guess the depth
just keep raising
and a small lip will form on the right side of the cut
on the jaws surface
god i love the internet
anyways, i will bring pictures of it tomorrow
more so when stoned, i may say :D
<- high on breakfast cereal
so are you an iron worker also?
i didnt think there were many of us online ;<
thought they were all redneck hillbillies
and didnt know how to use a kom pu tik or
i'm a mad scientist
oh even better!
youre more likely tripping on neverending acid
and dude, let me tell you. that shit never ends.
nah i mostly consume food
what do you do
like math or some shit?
because, i suck at math :<
i mostly chat on IRC
so how did you end up in here then? O.o
it was part of the plan to have robots take over the world
i got you.
im clueless, so yeah, time to go sleep
* skunkworks thinks fenn is on more than cereal
i think so too
and im not even tired
fenn confuses the shit out of me.
i'm just here to help
cnczone.com may be a better place to ask your questions.
lol, sorry for the bother, and fenn you are strange my friend.
it isnt cnc
its motor driven table
with a digi readout for x and y
i'd suggest making a 45 degree block to bolt to the table instead of swiveling the head each time
SkinnYPuP, seconds that
you could make it really big and then bolt all four jaws on at once (or however many there are)
im not uh
i would like to not have to adjust that bitch
twice a day
that head is heavy lol
its a big bitch :<
[00:47:02] <FoZ> http://www.p-h-s.co.uk/Bridgeport/bridgeport-mill.jpg
think of something like that to give you an idea
that machine is actually alot like the one in the shop
a knee mill
a few minor differences
just say you have a knee mill
never heard that before
it'll make it easier to understand what you have for future discussion
my boss isnt a machinist
that bridgeport mill is in a class of machines called knee mills
he used to run an operation like this before
and sold his shit, like 10 years ago
just bought this machine for cheep, jury rigged it to cut jaws
and it makes money
because they have a movable lower support for the table, which is called a knee
foz: if you have to adjust the head a lot, something is wrong
either in technique or with the machine
if you ever adjust the head, you're probably wrong :)
this is perfect
we have six of those and none of them ever go out, even the broken ones
i love you people!
(unless you're tramming it)
so the 45 deg. block is the way?
if you need to mill a pocket at an angle, then you should make a jig to hold the work, like fenn said
just toss it under the jaw in the table?
I'd have to read back too far to answer that :)
how would we measure each cut?
the quill still works
as does the depth stop
not sure about the z stop
so you make a jig that holds the parts in the right spot so you always use the same depth on the depth stop
but x is a little worn
and i never use y
usually i have to blindly find the depth of the tooth
and that seems strange, because this is supposed to be very precise
i pretty much cut about 1 turn from the top of the faced off jaw
thats idk about 1/8th maybe less
then i just keep going down until i see a little lip form in the right side of the cut
has something to do with the carbide cutter
4 of them, kindof square shaped
rotate them with a torqs wrench
sorry, i dont know the names of any of this stuff
all i can do is describe :<
that block idea is nice
but eh, my boss is a simple man
ok just stop before you offend the machinists
and a big idea like that
will scare him
how heavy is the workpiece?
the thing you're making
eh only 2 employees
idk about 35-50 lbs
i think this is a picture of what he's working with http://myconstructionphotos.smugmug.com/gallery/477972_FXsXm/1/18213879_kC5H7/Large
the round pads
you could get a "dividing head", or other form of rotary table
thats not at all
anyways, i cant find any pictures, im not sure what it would be called
but i can describe, square shaped
about 10 inches long
with two rows of teeth on either side
each row has 11 teeth
about 3 qtr of an inch from the last
they are cut square with the jaw
and all of that
if you cant gather a good idea of what im talking about, i will just take the digi cam in tomorrow and snap a few
the head is on a 45 right now actually
is it a bad idea to leave it like that overnight?
stress on the bolts?
it's bad to set it to 90 if it's not really 90
because then someone comes along and things the head is set correctly
it can be bad depending on which way you turn it, I think the oil can leak out
to the right
twice a day
sometimes more ;p
there - added the testpoint http://imagebin.ca/img/nZQXnB.png
(and top pour for the lower current sense resistor connection to the BA cap)
what and the hell?
see, i can program
decent with php
but that... is eh out there
software that you make chips with?
im going afk :<
omg skunkworks U R so 1773
im just... really trying to not feel retarted
I am amazed that Fox News is allowed on the air
I didn't realize it was possible to pack that much disinformation into so little time
even worse than O'Reilley
or whatever his name is
heh - Rachael is the bomb! :)
Rachel Maddow Show
oh her. yep, I've seen her show this trip :)
still no smoke - and the current limit seems to be as symetrical as I can test.
I'm setting up a stepper lathe, is there a way to get axis diameter display to show the same as the Xvalue programed or entered in mdi?
are you using 2.3beta or 2.2.8 or???
ok. 2.3 has G7/G8, which let you program in radius or diameter
otherwise, I think you get radius only
I don't know if I answered a question, or if it was the right one :)
ok I can deal with that, I have the step count so X moves 1/2 command amt to effectively arrive there
careful if you program any arcs
not that you probably would, but still
Ok so the way i'm going about it is going to change the R of an the arc too then ?
you'll get 2:1 ellipses
when EMC thinks it's making circular arcs
come to think of it, I don't know how arcs work with G7/G8, though I do recall some discussion of the issue
What are other emc/lathe users doing with their setups?
no idea :)
one day I'll have an idea - like when I retrofit my HNC
oooooo SkinnYPuP wants hardinge , drool
use 2.3 if you want diameter mode
it works great
Using AXIS, lathe mode, beta1, I noticed that when a tool length offset (X and/or Z) has been set (to a value other than zero) and is active (G43H...), that the "Machine Position" view takes the tool offset into account. A G53 move still references the raw machine position. Is this correct?
ummm. have you tried it?
cradek, are I&k incremental in g7?
mshaver: machine coordinates are the master coordinates all other offsets are applied to
I wouldn't think so - doesn't G53 just mean "G54 for this line"?
in 2.3 i/k can be absolute or incremental, according to your g90.1/g91.1 setting
to move in machine coordinates means to disregard every coordinate offset applied
g7/g8 are diameter/readius
or radius/diameter - I don't know which is which
mshaver: sorry, which part of your question is the question? you said a lot of things at once
toastydeath: but not tool length offsets?
tool length offsets are, by definition, offsets
The question is: Should the machine coordinates display be affected by active tool length offsets?
machine coordinates should not be affected, but the absolute coordinates should be.
toastydeath: by absolute, do you mean relative? Speaking here of the two possible display modes in AXIS under the _V_iew menu.
I can see an argument for either way
the arugment for the current behavior is "tool offset moves the tool tip, relative coordinates move the program"
mshaver: i don't know how emc labels it's coordinate systems
cradek: I just don't remember the mill version acting like this.
but the standard is absolute being the current work offset + tool offset
relative is (usually) a DRO available to the operator
mshaver: do lathe and mill act differently from each other for you?
or just "not like you remember" :)
sounds easy to test - I'm trying it
toastydeath: OK, you're talking about what we call relative. I've seen it referred to as absolute as well with other controls.
what other controls?
i see what you are saying
SWPLinux: my memory is always challengable...
mshaver: I don't see any difference between how lathe and mill configs work
toastdeath: non EMC controls...
if I have TLO but not G54 offset, # does not change anything in the view or dro
well, I remember wrongly, or so far back it doesn't matter! ;)
other guis may very well do it differently
I might try it in TkEMC
there are just so many offsets and coordinate systems and units and ... and ...
and guis! :-)
hm, here's a surprise for me -- g43.1 i1 / g43.1 k1
written by different people over different decades
you end up with a tool offset in X and Z after those
I kind of expected the second command to reset the X offset since it doesn't specify it
were there two questions originally? (how is G53 displayed, vs. where does the machine go, using G53)
swplinux: yes, there were
jepler: compare g1 x1 / g1 z1
jepler: or at least I think that's what I was thinking at the time
a g53 command moves to a location referenced to "machine zero", which is usually "home".
* cradek shrugs and points at the reference implementation
right - it's "g54 for this line, please"
SWPLinux: no, not g54
no, G54 can be offset
explicitly means not g54 (or any other g5x system)
oh right, machine, like without G92
and I thought it meant and no tool offsets either.
there are so many offsets ... :)
there are only a few offsets!
mshaver: yes g53 DOES mean that
oh tea, your right, sorry...
not tea, yea
I'll have coffee, thanks
I'll take either, but hot
mshaver: are you coming to fest in may?
* SWPLinux thinks it depends on motel prices :)
And another thing! I noticed that doing a touch off for a tool also causes an automatic G43H(tool#).
I see what you mean about how AXIS says "Machine" coordinates, but includes TLO. If you call G53 "move in Machine coords" you've got some confusion there
mshaver: yes, that's a feature (auto g43)
I thought so, not a bad idea!
mshaver: otherwise the dro won't change and you'll be baffled (IMO)
what happens if you're touching a tool off at multiple offsets
and after you touch off, you ought to be able to go ahead and use the tool
toastydeath: mshaver's talking about setting the tool length, not coordinate offsets
i am talking about tool length
except I changed to View/machine coords and that didn't change either, and then I was baffled
emc2 has only one length offset
mshaver: I follow you now
swplinux: no, it has one for each tool at least
what if i want tool one to have two different H values?
sure, but each tool has only one
we have programs where some tools have three heights
toastydeath: you'd have to load T1 and T10 and touch them off
Actually, lathe mode has both X and Z lengths
hm, so as long as the T code is the offset you want, it works?
toastydeath: tool touch off uses the loaded tool number, not the active tool offset index
can't you also use a specified length? (vs. from another table entry)
both of which seem to be turned on by a single G43Hx command
cradek: so how do i touch off one tool with two heights?
toastydeath: N.B. not the prepped tool -- the LOADED tool
toastydeath: if you can't load "both tools" with t1m6, t2m6, you need to edit the tool table manually
In my case it has to be the loaded too as the machine has a programmable tool post.
for instance the turret on my lathe is modulo 8, so I can load T1M6, T9M6
cradek: I did this with the lathe here too, for gang tooling reasons
it just turned out that way, and I like it
as we speak I have two drills mounted on one turret position and they are T1, T9
doesn't your toolchanger try and do something
that's a perfect example of the utility of this
when you go to t9
it goes to t(9 mod 8)
actually I use g43h9 in the program - but all it would is move to tool change position and back
it would not even lift the turret
I'm assuming that HAL/CL is doing the modulo for you, since the turret controller shouldn't be able to go past 8
yes modulo in CL
very easy to do
why is it done that way
because he chose to do it that way
no, i meant the tool touchoff thingy
as opposed to which other behavior?
touching off at whatever tool pocket you specify
hey speaking of configs; Can I (once I remember how) make a configs/smithy directory in CVS?
I thought there was one already
well, by pocket i mean offset
nope, there isn't one
toastydeath: there are several possible ways I see - use the loaded tool, or use the active offset index - I picked #1
I was going to also list 'use the prepped tool' but that would be nutty
or use the one the user enters?
default to the active tool pocket?
i guess i'm used to the standard, where you have online tool offset editing
sure that would be another way, it would require changing the touch off dialog
you can always edit the full tool table manually! I didn't take that away or anything.
I just wanted to make the very commonest thing easy
no, i meant the control itself has a screen for tool offsets
toastydeath: what is this "standard" of which you speak?
mshaver: every control i've used
this control has a text editor, which you can call a tool offset editor if it makes you happy :)
fanuc, haas, mitsubishi
sure, same difference
"you edit the numbers in the tool table" is the operation
swplinux: a text editor isn't interfaced into the machine
well, not on a Haas/fanuc/mitsubishi
IIRC on a BOSS8 Bridgeport it's, T1<EXECUTE> then SET TLO=Z <EXECUTE>.
I understand what you're saying though - there's no "integrated" thingy that lets you edit tool data on-the-fly
who was it that submitted the "integrated thingy" in tcl a week ago or so?
mshaver: yep, or you can painstakingly edit the numbers manually by stabbing the TOOL button over and over
fenn: it was still a separate editor program - it was dewey garrett
it's amazing what sticks in your mind and what doesn't...
mshaver: painful memories last longer
is there a way to cause emc to reload the tool table in the middle of a program?
but on a fanuc-controlled lathe, for instance
would it be difficult to add?
i can hit machine hold
but you can edit tool offsets in G-code
mshaver: actually I think TLO=Z is a pretty good setup
G10 L1 ... (??)
fenn: not sure...
and type "w+.005" into the tool offset screen
and it'll add five thou to the Z offset of the tool, and that change will be picked up the very next line
cradek: it's more or less what you have with touch off
W isn't tool offset in EMC2
mshaver: yeah, that was one of the things I had in mind when designing our new touch off
swplinux: w is incremental on lathes.
not in emc2, AFAIK
mshaver: except on ours you can touch off to 0.250 if that's the gauge you're using - much better
swplinux: i'm not talking about emc
i'm giving an example of what can be done on other controls with online tool table editing
is W defined differently for lathes, or is it just implemented that way?
what do you mean by defined vs implemented
the RS274NGC spec says that UVW are linear axes parallel to XYZ
cradek: yep, on a BOSS8 I would use a .1 jo block, SET TLO=Z, than jog off the part move down .1, and SET TLO=Z again
fenn: I'm tempted to say you could make it reread at each g41/42 entry - not sure what it would take
on lathes, uvw are parallel, incremental axes on top of xyz
mshaver: bleh :-P
W parallel to Z is not equivalent to "relative in Z, and by the way we'll just add them together to get the final joint position"
cradek: bleh indeed!
U and X move the cross slide, and Z and W move the saddle
toastydeath: and on some mills too
Hi all, have my cnc'd mini lathe all oiled up, last parts for the power supply ordered and brought home a computer to dedicate to the project. I have 2 ubuntu disks, one I got from the link on linuxcnc homepage and the other from link in here the other day. I can get both to boot the image and run live, but neither one will instal. One gets to step 7 of the install and has an input output error end exits...the other doesn't do
so G1 F.005 X1.00 U.005
will add .005 to the X value
or U by itself will just move up five thou
that isn't consistent with the spec, IMO
willburrrr2003: run the media check on both of them. I bet you have bad burns or a bad cdrom drive.
swplinux: it's consistent with almost every control made in the past twenty years, though
toastydeath: good for taking out taper in a part
danielfalck1: heck yes it is
ok I will try that and see, brb
SWPLinux: ngc was not a lathe spec at all
well, if they're programmed with g-code ...
but I understand that there are differences
lathe g-code and mill g-code are different for very good reasons
esp. the tool calls
or bad reasons - but either way it's definitely common
ok, I just haven't seen any documentation - only "I've seen it done this way"
even if it's done that way basically everywhere
that's because most manufacturers follow what fanuc does
except mazak and okuma, obv
"Freedom to Re-Innovate!"
pretty big exceptions if one wanted to argue against it
I'm not convinced there's any rational reason for the T1M6G43H1 / T0101 difference, but it's there anyway
well, on a mill, the T word is a call to the magazine
on a mill
there's no tool magazine on most lathes
toastydeath: is there a book or guide or specification of the "fanuc way"?
other than the fanuc programming manuals, no
uh - isn't a tool turret pretty equivalent to a tool magazine?
no, not at all
SWPLinux: sometimes the magazine can prep
but ... not all mills prep, not all lathes have turrets, ... ...
one day I'll remember half of this
right, and we have a mill or two that runs on lathe type tool offsets
and then you'll be sorry. or happy.
it is what it is
just issue a t word and the whole thing flips around
cradek: T1M26 right?
it actually is a lathe turret, i think, just modified for mill use
mshaver: yes it has M6 and M26
and on mill-turn controls you call tools up like a mill
mshaver: and we have something like that now! (TOOL_CHANGE_AT_G30)
which kicks ass for the lathe
you can set G30 just where the turret can spin without hitting anything, no wasted motion
that's another good idea
* cradek bows
there's a G28.1 / G30.1, right (set some parameters to the current position)
yes that's the easy way to set them
do those also set the G28 or G30 "home" positions?
yes that's what they do ("reference points")
oh, or is that all they do, and those positions happen to be in parameters also?
hmmm says no errors found....must be the drive :(
or a bad hard disk?
or controller, or cable
unfortunately "I/O error" doesn't sound very specific
using an 80-pin IDE cable with a non-UDMA drive, or a 40-pin cable with a UDMA drive, could possibly cause problems
err - 80-conductor. they're all 40-pin
don't think so, been a stable laptop for long time ... but have had issues with dvd's in it in the past
(damned if I can figure out why the 80-pin ones are better)
It may just be dirty too
I say give up now, unless you've run the latency test for a long time and not had any problems
eek, laptops are a curse for realtime machine control
but the live version runs great on it that's the pisser
oh, cool :)
I have and have run my lathe on it quite a bit
neat, you are lucky then - very few laptops work ok
(mostly running it th rough test paces and test cuts)
I have one, after much looking
what brand/kind of laptop? (just curious)
mine's an old piece of crap gateway P3
toshiba sattalite pro 4600
the only thing good about it is the realtime performance :-)
Built like a tank
Have been running various versions of linux on it for years
It's 10 yrs old and the linux drivers seem to be well established for the hardware in it, never had an issue getting anything to work on it
willburrrr2003: how much ram in it?
gig, and 1gig cpu
is there a way to do a network install?
I just ran into something strange. Can someone take a look at this axis window and see what went wrong? http://www.wildrice.com/Screenshot.png
SWPLinux: What problems (40 conductor vs. 80)?
willburrrr2003, PXE install if your net card supports PXE
mshaver: I think the drivers/chipsets can misdetect the correct data rates if you use the wrong type of cable
willburrrr2003: simulator, or controlling a machine?
errr wildrice: ^^
Skinny, you have my complete attention... what is PXE?
I am controlling a cnc'sd mini lathe
network boot to a tftp server on your lan
wildrice: the preview might briefly stop updating if the computer is very busy, but the controlled machine should still move properly since it is realtime
I did not have the driver powered up.
ok, well will try one more cd and see how that goes as both the ones I have fail in diferent ways but the same way each time
I was just testing. The machine sould not be
burn slowly, like at 4x
SWPLinux: Well, you learn something new every day! I thought it was just additional interleaved grounds.
should not be too busy, though I did have a VNC session to it.
wildrice: it's not cause for much worry, but would be improved by raising your BASE_PERIOD if possible
ok, will do...last one was 24x I think
it is, but I guess they make a difference :)
wildrice: (the machine itself will still move correctly, as-is)
willburrrr2003 You'll have to see if there are any pxe boot options in your bios, other than that google pxe ubuntu there's plenty out there
If it is just cosmetic, then that is ok. Thanks.
I am looking forward to getting my cnc setup going, already have $ job lined up for it
thanks skinny, will check that out
so "festering in Wichita" Is May 21-27-ish?
when I get it up and running I will share some vid with you all
cradek: there's no emc2-trunk/configs/smithy directory. It would fit fairly well just between sim and stepper.
I have to say I am loving EMC2 , being a system integrator for a living I have no problem with understanding the "wiring of the components" and looking forward to classic_ladder as I program many various PLC's at work
mshaver: feel free to add it if you have good configs for particular machines they will sell - we have those for several companies
add them on trunk - you should probably talk to jepler if you want them in 2.3
Have any of you tried the Modbus features in the latest Classic_Ladder yet?
cradek: will do
mshaver: glad they are coming with a maintainer :-)
hmmm.... well I will let you know how it goes when I try it :) It doesn't look much different in the manual than from What I learned in an Idec PLC programming class last week
yea... even a paid maintainer!
heh. all the companies have to do is donate a machine or two :)
SWPLinux: Actually, I need a machine in Wichita. Maybe they can send one down there.
fun - what's the project?
Stuart has several ;)
but none that - err - big
cradek: just a machine to test with and screw around with. I really want to work with Seb on hm2 stuff and having a machine to move would help.
Did I cause the net split? Am I bad luck?
I can live with that.
you are the proverbial straw (as in camel)
So, are you good with my terms for going to Kansas?
i forgot the terms
Y=mX + B
or is that y = Mx + b
was it, "I drive the first half, then I also drive the second half"?
or missed due to the split, " OK, you drive ALL the way and get Smithy to pay for the hotel and I will come along as baggage."
Given my ability to code for EMC, baggage is a good description.
hmm, interesting... I did post on the e-mail list that people were to badger you about this, but so far they've been slacking.
I'm away from home, and therefore only have webmail access
(that's my excuse anyway)
webmail, and IRC
web is my only mail (for outgoing)
I've got to go to fest to talk to Seb
I hate the measures ISPs take against spammers
which also prevent me from sending mail when I'm not on my home network (except via webmail)
SWPLinux: Use port 465
or run your own mail server...
I can try that. why should it work?
(port 465 that is)
SWPLinux: Many use that to bypass hotel's blocking outbound port 25
it's not the hotel, it's that my ISP won't relay from foreign networks
for a while I was sending from SMTP on my desktop machine. Some isps would block my mail, but most wouldn't.
SWPLinux: Did you authernticate?
yes, I'm asked for a password
but it's not like I'm setting up a VPN or SSH tunnel
just normal pop/imap password stuff
SWPLinux: I'd check with the ISP, MANY ppl send mail from remote locations
consider iPhones and BB's
I know. there was a time when it was recommended to stop accepting relayed mail, and my ISP did
Relayed mail? Huh?
No no, direct connection to their MTA
well, I think it's considered a relay when the mail comes from somewhere else
telnet xxx:25, HELO / RCPT TO xxx ...
try 465 SSL or TLS
they have the mail server firewalled against outside networks
They can't, or they would never receive any mail at all
I kinow they set up this filtering on purpose - I don't know exactly what they're doing
SWPLinux: Ok, they are doing fwd/rev dns lookups
setup timeout to be 91 seconds
Some clients have a timeout option
that was odd
hey, maybe TLS does work remotely. thanks! :)
yay (sort of)
since I still end up with no record of sent messages at home, when I send them on the road
need to get IMAP going at home
steve_stallings is now known as steves_logging
ok, so I guess the network cable right by my foot is a little flaky
Foz: if you build a fixture sat 45deg, you loose this goodie....
you cant just crank the knee up to adjust for the wear on the part now in fixture...
becuz the x dimensions move as yo move the Z... the V cut goes sideways, not down.
maybe make a jig that lets you accurately tip head is handy (just a big 45 to feel the quill against) should be close enuf for pile driver heads
oh, look into steel for hammer dies, check uddelhom
you can flood weld that stuff
tool 0 on some systems means... use whatevers in the tool holder now
hows emc running on 64bit systems ?
AnUnderling is now known as SkinnYPup
Whats the name of the package with the opengl headers?
Got it , libgl1-mesa-dev and libglu...
my drives use scsi2 connectors ( Yaskawa and Panasonic)
i need to make a breakout like board
the scsi2 is not nearly fully populated,
will db25's work ( twisted pairs and shielded?) for encode, enable, limits, VelocityCommand...
using scsi2 on the breakout is not desirable ( too hard to push past penny pinching eng staff )
and i dotn like soldering the scsi2 anyways
why wouldnt using parport work ? :P
speed, different animal
scsi2 and parport are similar evolution branches, but parport is still living up in the branches ;) and scsi2 is driving a benz
oh well, off to cad land
this is all becuz the dufuss who made the cnc used db15s for the 1st 4 axis, then a db37,db25,and 2db15s all crisscrossed for 2nd set of 4 axis
he doesnt want to stock the modified boards i made up ( changed outptus to db15s) , so i'm stuck fixing it off board
oh well, off to cad land ( stop bitchin, get to woik! )
oooh ooh, i already have an orcad for this, the schema is done, i just need the pcb layout... gonna go home and put it up on the big screen tv :)
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-04-02.txt
let the fun begin... the V-F board is on the way
voltage to frequency?
whats the voltage represent, what does the frequency control?
the input is either 0-10 or -5 +5
the frequency output drives an encoder
is the voltage from a pot or some sensing?
using the velocity output from the encoder I'll know if the voltage changes
it is from a 0-300vdc signal from my plasma torch
scaled down to 0-10 with a voltage divider circuit
ah, a real sense, and the output simulates rpm? (is there a real encoder?)
the velocity output from the encoder (HAL) is used to sense the voltage change
it sounds cool, i need to boot an emc box to see how an encoder ( an output device ) can be driven. but you made a VoltageToVelocity gizmo, and sounds like you have threshold triggers too.
sort of a comparator
yep I think so
hmm, today the carwash had a clumsy opto system theat moved the brushes along the countour of the car profile. just two optos and you could watch it track between the two
one above the body roofline, one below, and arms. the brushes just were slaved to the opto arm position
we used to call that bit-bang or 'drive between the curbs' ;)
BJT-Work: sounds cool enuf to log off and look into it, posted any hal files?
just this http://www.linuxcnc.org/component/option,com_kunena/Itemid,0/func,view/catid,10/id,17/lang,en/
not bit bang, bang-bang
y the pyvcp offset pot is way cool
i wanna see if i can make the pot +/-10V (or 10 whatevers)
you can set the range
wouldnt you really rather have a real pot?
(wouldnt you really rather have a buick was running thru my head)
this works for testing the HAL part. I'll use my lab power supply to test the board part
again thx, gonna reboot out of eeepc linux to ubuntu on lappie
[15:37:04] <jepler> http://www.youtube.com/watch?v=Zz95_VvTxZM
jepler: have you ever seen the mythbusters hot water heater show?
[15:43:21] <skunkworks> http://www.youtube.com/watch?v=68p4ngS-yME
[15:44:34] <skunkworks> http://www.youtube.com/watch?v=pu3FwgIHsQA
somebody answer the limit switch question on the list, he is a mach3 guy I hope he converts http://www.youtube.com/watch?v=FkzqwIjjBhs
I cant experiment at the moment as Im mid gear
Would Ubuntu & emc2 run on any of the newer HMI's that are available on the market?
what is an hmi?
human -machine interface
HMI , Human Machine Interface.... GUI screen used lots with PLC's
Some are really dumb terminals with basic functions, others are full on embeded pc's
only if the device you're talking about can run Linux+RTAI
you'd need a full-on embedded PC
Yes , Makes sense
are they cheap? sounds like a time sink
no, they're generally quite expensive
depending on size they can be pretty pricey....looked at one for my work the other day and it was $1400.00
12.1 color touchscreen with embeded pc
and a lot of those embedded computers are junk
Not a lot of experience with using them after they leave my shop, and the op sys is usually pre-installed so I do not know what it takes to get them up and going from scratch
usually just a CD drive to boot from
units from companies like Allen-Bradley are less likely to be generic, I'd bet
they probably don't cost $1400 either
but a "kiosk PC" should work fine (I have one of those)
no, more like $4000
The one I priced out was an IDEC
Allen-Bradley are more like double the price of IDEC, as they are top shelf industry standard
I will have to get ahold of my distibutors and see what platforms that can be installed on the various brands
HMI = lcd monitor, generic PC, capacitive touchscreen, massive 0.5" thick steel bezel to make you feel like you're getting something "quality"
ok - latest eagle files on the website http://www.electronicsam.com/images/KandT/servostart/schem/latestcurrentlimit/
I think it actually has all the right values.
[17:29:08] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/schem/latestcurrentlimit/latestboard.png
[17:29:15] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/schem/latestcurrentlimit/latestschem.png
skunkworks: let me know when you send units out for betatesting :)
no more IR2xxx ?
or are those IC11-IC12?
I new i would forget somehting
Mainly changed to that package because motor common and logic common are separate. (and allow +/- 5v diff)
and they have an adjustable deadtime
Now the latest eagle files are up ;)
Spam Subject: "I gave up hope finding you". I'm thinking (if it wasn't spam), ok, now that you found me, what makes you think I wanted you to?
you could ask in your reply
subject "well, you found me. Now what?"
I've actually done that. My cousin's gf emailed me not sure if she had the right email address to invite to a surprise bday party.
archivist_attic is now known as archivist_emc
archivist_emc is now known as archivist_attic
SWPLinux: did you see that o'scope?
[18:00:39] <JymmmEMC> http://yveslebrac.blogspot.com/
I'm even thinking that it could be used with HAL scope
but emc no do usb
a heavily modified halscope, maybe
well, you could make a userspace app similar to halsampler, but gets the data from a USB device
then use halscope on the output of that (more or less)
or just replace the scope_rt module with that userspace app
skunkworks: you need separate motor common to do low-side current sense, right?
At that rate, you could incorporate a o'scope directly into a project
sure, if that level of performance is enough for you
SWPLinux: I think for anything that I would have needed to check my motors, encoders, etc would have worked
hmmm. maybe, maybe not
the ADC in the AVRs is pretty slow, like maybe 60 kHz total
(that's total samples per second, across all channels, so divide by the number of channels to get samples/sec/channel)
SWPLinux: even worse than you thought. "65us - 260us conversion time" for attiny25/45/85
how's 15kHz sound
or 7kHz for 2 channels
It may not be perfect, but would at least give some indication that something's up - enough to justify pulling out a real scope
I may have been remembering the 16 / 65 backwards (thought it was 16 microseconds, but I guess it was 65)
again, maybe - maybe not
I like a real scope
even an average PC can have 3 or more BASE_PERIODs in one sample time of that chip, and you need two sample times if you're doing two channels
Fine, say one channel then
so it would be useful for looking at limit switches or direction lines, but not steps, and certainly not step and dir at the same time
it's cool anyway, but not really useful for anything beyond low audio range
at least not as a scope :)
if it had a digital mode it could be much faster
fine, up the chip then
yes, for two bits ;)
(3 actually, there's an unused pin)
different AVRs have different rates -- atmega48P/88P/168P/328P = 76.9kSPS
13-260uS conversion time
none are very fast, though some of the higher end ones might be usable for low speed step monitoring
with analog that is. digital capture is a hell of a lot faster
wouldnt cost much to add a real a-d
in absolute terms, no. in relative terms, it would triple the cost
maybe more. and you'd need to use SPI, but those pins are being used for USB (not sure if that's necessary, I haven't looked at the code)
so you might have to use a bigger uC ...
Im possibly doing someting similar it could do a scope as well
SWPLinux: with the simple software usb stacks I've seen you have to have the usb D+ on an interrupt pin. That appears to be INT0 only
ok, that makes sense
jepler: yes - that is why I switched.
(not sure whether a pin change interrupt is acceptable)
it should be, unless you need multiple pins
then the "arbitration" may take too long
the inital current limit circuit - the common for the driver ic went thru the current sense resistor. (I just didn't like it)
(ie, deciding which pin caused the interrupt)
there are also additional delays for the pin change interrupt
that seems strange
on a slightly related note, I was talking to some of the NXP guys here, and apparently there's a mode in some of the ARM chips that gives you zero jitter for interrupt latency
not zero latency, just always the same number of cycles
in the attiny25/45/85 doc there's a table on page 52 that shows how the PCIF flag is only asserted 4 clock cycles after a PCINT(0) rising edge. There's no similar table for INT0. Not sure why the difference..
you can select only a single pin for PCINT, so that's no worry
hmmm. I'm not looking at it at the moment - is that in a power-save state or fully on?
BJT-Work_ is now known as BJT-Work
alex_joni: are you near your keyboard?
where do you define the kins for a robot?
err.. in emc2?
[19:06:41] <BJT-Work> http://www.linuxcnc.org/component/option,com_kunena/Itemid,0/func,view/catid,10/id,34/lang,en/#37
BJT-Work: basicly he'll have to look at trivkins
and replace it with his own kins
then change the loadrt trivkins in the HAL file
with loadrt mykins or whatever
but the mykins needs to export the kinematicsForward() and Reverse() functions
just reply and I'll fix it later if something ain't 100% right ;)
gotta leave the keyb for a while
Thank you for answering
I would very much like to consult you regarding axis drift...
sarma: just ask and wait
there are a lot of helpfull people in here, but it takes a while to get an answer
I have been using motenc-lite card with incremental encoders. It seems that the card doesen't register all the counts. Thiss causes very serious axis drifting, and i don't know how to solve it.
could be electrical interference...
sarma: can you move the encoder by hand?
without power to the motors
move it as fast as you can and see if the position is consistent
I was asking about encoders (somehow i got disconnected)
22:17 < BJT-Work> could be electrical interference...
22:21 < alex_joni> sarma: can you move the encoder by hand?
22:21 < alex_joni> without power to the motors
22:21 < alex_joni> move it as fast as you can and see if the position is
after trying everything possible, i finally connected only one encoder to single motenc-lite channel, and moved it by hand for 180 degs cw and ccw. Counts differ every turn, and seem to increase every cycle...
sounds like something is not right
I think this can happen if you have differential encoders and the card is set up for regular TTL
(but I'm not familiar with the motenc, so I don't know if there's an option to change this)
[19:36:39] <BJT-Work> http://www.linuxcnc.org/docview/html//hal_drivers.html#r1_6
Guest609: my personal choice would be to contact Vital with the encoder model/manufacturer and ask for help
don't take it wrong, we're happy to have you here, and try to help you out
I've tried that as well (connecting only A+/B+ channels, and than only A-/B-), but the effect os always the same. Signas on the scope look very nice (4.5Vpp)
Thank you all very much. I was just wondering if someone had the same problem...
Guest609: haven't heard about it (with a motenc)
I did hear about a similar issue with a mesa board though, after setting a jumper it was much better
Guest609: the motenc manual says differential encoders
The same type i used. i only tried disconnecting A-/B- channels to see if anything would change, but there was no effect.
Guest609: if you rotate it really slowly 180 degrees and back, does the count come back ok?
motenc is not differential??
Only if rotation is very very slow (approx 0.5rpm).
motenc is indeed differential
dont leave unused differential inputs open
cradek: it is differential
archivist_attic: that was only for testing, to see if it makes a difference
would You suggest to ground them?
Guest609: no, use A/A- and B/B-
they have to be used properly or to a fixed state
sorry, i was disconnected again.
Servo drives with buffered differential encoder output were used. First i tought that the amps are generating noise which is causing false counts, so eventually i removed the amps and left single differential encoder conected directly to motenc brakeout.
Wiring terminating resistors also has no effect.
Guest149: sounds like you're doing things as you should.. not sure what else we can say about it
FoZ is now known as FoZzY
Guest149: the only thing that can be done in such a case is try to get a 2-channel scope picture of the encoder
and see how it looks
did you try a different encoder?
i would dare to say that signals appear perfectly on scope. 4.5Vpp sharp on all channels (A+, A-, B+, B-, I+, I-). No amplitude drop regardless of rpm (0-3000). i also tried connecting different encoders
what is also sometimes important is the timing
that the B transition comes at the half of A
I does (at least from what can be seen on scope)
total of four encoers were connected, but the drift sadly remains...
(scope signal on brakeout board terminals is sharp 4.2 - 4.3 Vpp every time, with 90 deg phase shift between A and B channel)
what cable do you use from the breakout board to the motenc?
JymmmEMC, hey... the ISS was just on 145.800 talking to some scholl... probably in California because I lost the ISS shortly after the school dropped out... next chance is in an hour http://www.issfanclub.com/
21:11UTC is next AOS
JymmmEMC, call is NA1SS
LawrenceG: ah, cool. too bad I don't have anything but HT
you should be able to hear them on a handheld
Very nice idea to suggest. Original cable from manufacturer was used (differentially routed flat cable). btw. i also tried to change it and it had no effect whatsoever.
pulled mobile rig out of car when it was broken into
Could the motherboard possibly have any effect?
just leave the radio on with a loose squelch
Guest149: I think the motenc quadrature counter is very sensitive to noise. I have heard of two other people having trouble with it.
Guest149: do you maybe have high frequency noise at the card that your scope can't see? is all your grounding good?
Guest149, check also where the switching point is, add a pull up/down to improve
I wasn't sure about the noise. In order to elimitate the noise as much as possible, all the drives were disconnected, and only a single encored, connected to PC +5V supply was used. No noise is visible on my 20MHz scope.
sorry for the mistyping, i ment single encoder
btw. there are no counts when encoder stands still
a differential encoder with all A- A+ B- B+ hooked up directly to the card and it still miscounts?
then maybe your card is defective
have you called vital yet?
and also with 300 ohm termination resistor
I will contact them tomorrow
Guest149: I'm not sure it helps you at all, but everything you've tried so far sounds ok
yeah sounds like good troubleshooting to me, too
the only thing I havent tried is different motherboard, or connecting pull up or pull down resistors
or perhaps a lesser termination resistor? 100 ohms perhaps?
What is your experience with mesa boards btw?
they work perfectly, support is first class
the guy who designs them works very closely with us
that doesn't mean the motenc shouldn't work :)
just answering the question :-)
Thank you all guys for your help. Tomorrow i will contact Vital and ask for help.
welcome, I hope you nail it
bye.. good luck
and do come back in here .. even if only to report how you fixed it :)
yeah I'd love to hear the solution
Will do. Thanks for invitation. It really makes me very sad, after so much time wasted on every possible combination and still no results.
If there is a way, i will gladly send you pictures of my setup
no GOOD results - but you know much more than when you started - those are still results
that is true indeed.
Guest149: sure there is
if you don't have a place to upload them
you can temporarily use services like imagebin.org
or you can make a page at wiki.linuxcnc.org and add your info there (even upload images)
I will put some and send you link.
good night all