hmmm. so what kind of cutter would anyone recommend for cutting ABS plastic on a Bridgeport (= top speed 4000 RPM)
the cuts are ~1/2" wide connector holes in case sides that are ~0.1" thick
since I never got around to CNCing my mill (ahem), I'd like to lock the quill at depth and just cut straight through each slot
or I can set the depth stop and bring the quill down
i'm about ready to disconnect my bandit controller and hook up the new drivers and my pc with emc2 to my shizuoka mill for some testing
i'm using a 5i20 with a breakout board
i got the step and dir pulses figured out and ready to go, but i'm unsure how to configure it for external components like buttons, switches, relays, ect
i've read a bunch, but since i'm horrible with computer lingo, it doesnt make much sense... is there anything out there in layman's terms that might explain it?
that will depend a lot on what you want the external buttons, switches, relays, etc. to do
well, i got a vfd for the spindle and a toolchanger for starters
heh. the VFD should be pretty easy (assuming it doesn't need an isolated control voltage)
plus i might score a control pannel with a jog wheel, estop, xyz switches, ect
a toolchanger is likely to be a much harder bit of setup
yea, i know it wont be easy
unless you're good with ladder diagrams
well i can trace out the wires to figure out what does what
and i dont know what a ladder diagram is lol
in that case it'll probably be pretty hard ;)
yea, i figured.... but i'll get it eventually. i dont need the tool changer working right away anyways, so i got time on that
i googled ladder diagram, looks fairly learnable
what i was wondering is, do i modify the config file that has the stepgen stuff on it for the other i/o stuff, or does that have it's own config file?
[00:19:25] <WalterN> http://www.youtube.com/watch?v=2jm4_HikMqk
Guest516, if there's a 5i20 config that you started with, you can certainly modify it to add all this other stuff
yea, i started with the one included in emc2
then starting with that one (after making a backup :) ) is probably a good idea, if it already can turn your motors and stuff
you can have as many configs as disk space allows, so don't worry about making new ones
I wonder how much that machine would cost...
(unless they'll confuse you later)
where's a good place to find info on editing the config file for someone completely new to this stuff?
the manuals are good
i havent tried it with the motors yet because i have to disconnect the existing (working) controller to do so, but i am getting step and dir pulses from my breakout board, and the settings seem to jive with the requirements and settings for my drives
you may need to tune the acceleration and velocity limits, but other than that it sounds good
it looked like the generic settings in the hstmot2 file were pretty slow compared to what my drivers could do, i assumed starting slow wouldnt hurt anything, is that right to assume?
it's gotta be a ton faster than my current drives in the machine, that's for sure
those things are antiques
but you'll be reusing the motors?
very beefy motors
the big ones with the fins?
1300oz in 9.2a
na, standard nema 42 sized
what drives are you going to use?
little undersized, but i'm going to put some fans a a good sized heat sink on them
plus i wont be pushing it hard
are those the Gecko clones, or the new design?
they are superior electric m112-fj-327 motors
new design i believe
ok, they may have a high enough current/voltage then
(I don't know the motor specs, but I do know that 9.2A is past a gecko or a clone :) )
7.8a max, but the existing ones in the machine are only rated for 8a max
and they lasted for 30 years
30 plus years actually lol
i figured with a 3hp spindle motor i'd max out the spindle before i maxed out the steppers
could well be
but then again everything i know about steppers and drives i learned in the past week or 2, so it's likely that i'm wrong
though larger motors tend to lose torque much faster as you increase speed
oops. gotta finish this order in the next 20 minutes. bbiab
thanks for the help, seeya
SWPadnos: ABS needs to be cut with a SHARP cutter, two flute would be better than four, take a bite several mils thick, don't rub or scrub, not work hardnening (obviously) but rather melting plastic is your enemy
thanks. any coatings or anything?
cobalt is about as cheap as HSS, so I may get that for future needs
never found any lube that helped, so I cut it dry
I think McMaster says that TiCn or TiN are better for plastics (less sticking I guess)
amazingly ABS is abrasive, so it will wear cutters quickly
dry is good. cleaning pebbled beige cases is a bitch
ok, so maybe cobalt is a good thing
maybe it is the coloring agent that is abrasive
backing the workpiece to support it is helpful
yeah. I noticed that on a test cut
did a couple of thousand connector notches in ABS using a horizontal mill, only machine I have that ever actually earned its keep, but then I bought it for $25
paid more to have some old cutters professionally sharpened than I did for the machine
what about shipping? :)
well, the pickup was already there to haul my butt home, so the incremental cost of gas was small
steve_stallings: didn't you get a bigass lathe that you were going to retrofit?
SWPadnos: you need to work with plastics?
tomp is now known as tomp3
JymmmEMC, yes, I have some boards that need to go into ABS cases, with connectors poking through
cradek: yes, still have it in the warehouse at the office, shop at home is hostage to economy now, so I don't know when it will move ahead
actually before I retrofit it, I will have a go at getting the Fanuc T6 control running
SWPadnos: are you drilling or cnc milling the holes?
I have almost a full set of spare boards for it, all unknown condition
they've got to be milled, since the cutouts are rectangular
SWPadnos: .125" radius ok?
and that's just a maybe
and I don't have a hot wire cutter either ;)
that is not going to be fun in ABS, not much room for the chips to clear the cutter
*SMACK* I mean is the smallest radius being .125" ok
yes, I knew what you meant, nad hte answer is no
you might consider one of those cold air guns to cool the plastic and blow the chips away
the corners need to be pretty darned square
SWPadnos: Well unless you find a square mill it aint gonna happen
punches can work well in ABS
the wall is only 0.10 thick, so if I plunge into it, I'I should be OK (stiffness aside ;) )
yeah, a punch would be great
this is only 3 sides, I cleverly placed the connectors such that one edge is along the case parting line
so a nibble could do it, if it can do the thickness
that is the same thing I did, such that I could use a horizontal cutter with a square corner
horizontal cutter also allows better control of chip load
I was also thinking of plunging into it, but I'm not sure about that
it seems easy, but I dunno
nice thing about horizontal cutter is all parts of the cutter edge are moving at the same speed with respect to the workpiect
no spot where it is more rubbing than cutting
a friend of mine cut a small batch of 0.050" thick flat ABS panels for connector cutouts using 1/16" endmill, yield was only about 50% due to ABS melting
I'll get freeze spray ;)
alcohol mist might be cheaper
I have several air cans I can hold upside down
steve_stallings is now known as steves_logging
wood block inside box (must fill wall to wall), wood block outside, clamp in vise, mill slots, get 3 sided openings with sqr corners
yep, that's the plan
[02:25:24] <JymmmEMC> http://grizzly.com/products/G9701/
wow a very fancy small wood working machine (example violin body it carved) http://www.carvewright.com/gallery/?g2_itemId=2570
buy a set of jewelers files and an exacto saw blade first, those corner punches are crap,
no their not, maybe that brand is.
laser the connector openings
I'll laser you!
where's teh sharks?
what context? There is a hockey team around here by that name
sharks, with frikkin' laser beams on their heads
hey archivist check out http://www.thedesignblog.org/entry/the-100-year-alarm-clock-lives-every-second/
you like clocks right ;->
haha .. the gap is the end of time ?
i guess its the "alarm"
when it falls off 100 years has passed
i didn't see if they said the rotation
Valen1, the drive belts will fall apart well before the end of time
yeah but still
only needs to last 100 years though
(gears would be better)
I bet that within 100 years there would probably be a power failure or 3 as well
I service clocks that are older :)
micges: after the problem occurs, does the AXIS DRO display the correct position from gcode, or does it display a different position?
another way to determine if you've lost position is to note the difference between axis.2.joint-pos-cmd and axis.2.motor-pos-cmd, then home, then note the new difference.
If it changes by a lot, then you lost position
so question is: where can I add some more logging, "moved to XYZ" location, "CHANGED smth on/off"
micges: I don't understand the failure yet, can you say more precisely what happens?
it is only one bad move and it doesn't lost position becouse anything after bad move is correctly
"machine > set debug level"
there are sended messages not executed
cradek: I have program with say 10 holes
5 is correct
6 is hole with some damage(z not up correcly)
7-10 is correct
what is the incorrect motion at hole 6?
[13:31:56] <micges> http://imgpaste.net/tmp/bpluj.jpg
some sort of that
so I'm thinking of improve debug to find that thing/bug/
jepler: and for what I saw bad move isn't in preview
this machine is 24/7 running so I can't play with it, I can only enable logging everything I know and then search
micges, is the gouge on the entry to the hole or the exit from the hole?
I wonder about making a tube amp for a surround sound amplifier
just for the "oooh pretty" of it ;->
if you run the same part again, does it run the same way?
if it does that without losing position, the bad path would be shown in the backplot
yes, if it appears in the commanded position it should appear in the backplot
on the other hand, if it's a following problem (like a loose encoder coupling), it would not show in the backplot, and the machine would not later recover the correct position
if sticky slide then it could recover
so I don't feel like this problem makes sense
archivist: if it's sticky he'd get an ferror trip
if not getting an ferror, you know commanded and feedback match, so either you have an incorrect command, or you have feedback not matching the actual machine position
neither of these sounds like the described problem, yet I think they are the only two possibilities
head loose on carriage
tool getting pulled out and then pushing back in at the next drill?
for a 24/7 machine then a service perhaps
yeah, I was just thinking about that
examining the ruined part very carefully might tell you something
I ruin circuit boards like that, but it's never pushed the tool back in afterwords
(the first part of this conversation was logged http://linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-05-12.txt
starting at around 13:11)
i'm noticing things to check when next time will happen, thanks
IMO the backplot will tell you with almost certainty whether it's a software or hardware problem.
oh I misread earlier: you said the problem is not seen in PREVIEW, you did not say not seen in BACKPLOT
micges: what gcode is the program using to drill the holes?
ok, not a drill cycle
and small g3 after that
full circle g3 then g0 up?
using tolerance mode? I remember there was a fix recently about this.
the full-circle fix?
yes using latest 2.3 branch
full circle problem I have on 2d laser, never on mill
jepler: cradek: I'll use your advices for tracking it, thnak you
hope you find it!
micges: if you find that the bad path is shown in backplot, keep a screen shot and the gcode.
nice enco special deal today only: "Take 10% off your order and get Free UPS Shipping"
er, and tomorrow I think
hmm, what do I need from them?
it's such a great deal I should spend money immediately
Apply Promo Code WBMYC9, then apply WBMYDD.
(actually I'm trying *not* to buy anything more for electronics or cnc until after fest .. I spent more than I'd like lately, between my unexpected computer purchase, the beagleboard, and the mesa stuff..)
I'm doing that too
another 8 rows of shingles up last night. so far so good. (supposed to get windy though)
the last ridge cap is the best one... you get to glue it on :)
skunkworks: shingles?! Dude you're suppose to go green and plant grass up there!
jepler: what computer did ya get?
JymmmEMC: one with a cooler like this: http://www.theregister.co.uk/2009/05/11/koolance/
JymmmEMC: I got a system based on Phenom 9600 to replace an older AMD X2 that failed unexpectedly
jepler: ah, ok
biostar ta790gx xe + phenom 9600 agena 95W + antec earthwatts 430W + existing case and ram
Heh, the name "biostar" just sounds cheap to me every time I see it. No idea why though.
the motherboard got decent reviews on newegg, and I was able to establish the linux compatability of most of the onboard stuff before buying.
jepler: Oh no, it's just how the name sounds to me is all. =)
the failed motherbard was gigabyte brand, the system *that* one had replaced is biostar brand (circa 2005) and is still going strong
Aint this the shits... Thinkpad t60 Core 2 Duo Supports up to 4GB ram. But the chipset 945GM and 945PM only support up to 3GB. Right on IBM's website too.
Is it me but when you say that it supports 4GB and you know it doesn't that's false advertising.
jepler: See MEMORY section: http://www-307.ibm.com/pc/support/site.wss/document.do?sitestyle=lenovo&lndocid=MIGR-62487
JymmmEMC: that page suggest you can get a T60 with a different chipset
Hello alex.ok for Yemc.One question: Yemc can control circular axis.I prove at modification in ini file ,the axis r, p or w at circular ,but have the error. because?
alex_joni: where do you see that?
motioncontrol: because the axfis names are XYZ ABC UVW. RPW are not axis names.
R, P and W are used for other purposes in emc gcode -- for instance, R is used to specify arcs
well, W is an axis name, but it's not part of the trio RPW
oK the solution is add the a or b or C axis in Yemc? I want retrofit one maschine with 5 axis xyz and a and B axis. Is possible whith Yemc?
why do you think you want to use yemc for this?
becAUSE YEMC CAN CONTROL6 AXIS , XEMC ONLY 3 axis.
JymmmEMC: that's my interpretation of that paragraph
xemc and yemc have not been maintained for almost ten years. there are two other guis that are a better choice, tkemc and AXIS.
both support all nine axes and nontrivial kinematics machines.
alex_joni: More than likely (depending on specific model) you'll get one of those chipsets listed
alex_joni: And of course their website just went down.
yes i thing use the axis if no possible use circular axis on yemc.
in any case, decompose the problem into two smaller parts: getting a working conifuration with B and C axes using tkemc or axis; second, get the gui of your choice working, if that is important.
Ok i want use xemc or yemc because my client normaly are habitual at work with simple interface.Now i have add at ini file U axis and have the error :setaxes faling: axis=4 axismask=47.
review the description of "COORDINATES" and "AXES", and see whether you can determine the proper setting (you probably have AXES set wrong): http://www.linuxcnc.org/docview/html/config_ini_config.html#sub:[TRAJ]-section
Yes thenks for help.I have add in coordinate xyza and home 0 0 0 0 and set the axis x y and z linear and a circular . Yem run but not visulalized the a axis in gui.
If i write in coordinates xyzrp, and set the r and p axis 0 linear the yemc run ok and visualized ok the gui.If i change the r and p in circular have the error INIFILE: ERR_CONVERSION, section=AXIS_4, tag=TYPE, num=1, lineNo=219
THERE ARE NO R AND P AXES
the rotational axes are called A B and C
Yes i undertsant.But the a b and c axis not visualized in yem, because is old project ?
<cradek> xemc and yemc have not been maintained for almost ten years.
ok thanks in use axis or tkemc for maschine
one question above indicized site. I write on emc chat whith old nick name.Is possible delete at google indicized ?
no, the logs are never deleted or modified.
ok thanks for your patience jepler
I've just had my new business idea: a by-the-hour machine shop with espresso bar. it'll be called Precision Grounds.
I went machine collecting today, fetch the bosses new toy
ooh anything cool?
a Theil milling machine
been sitting outside for a while but seems ok so far
like that table
rotates a few ways
So suppose that I wanted to replace a PLC with a computer running EMC/RTAI so that I can get good real time performance.
Is there a good way to coverup, replace, or modify AXIS to have a generic HMI?
what does the machine and PLC do? sounds like it's not like a mill or lathe.
No it wouldn't even run g code, however it would need to control two axis of motion.
maybe you need HAL more than you need EMC
it would just be processing some wire.
EMC pretty much runs gcode, and AXIS is the interface for that
your "UI" could either be made using pyvcp, or by a program in any language you like that uses HAL APIs directly (C, C++ and Python being the easiest since there are existing bindings)
geo01005, there are wire bending machines using EMC iirc
"halrun" lets you use the hal environment without loading any of the gcode specific stuff
if you're replacing a PLC, maybe you solution would be part or mostly classicladder
Hmm that sound like what I would like to do.
I would like to avoid writing the code in ladder logic.
However, some of the program would require something like classicaldder, so maybe I would use a combination of hal and classic ladder.
So it would be possible to make a really nice looking interface with c++ and QT?
that depends whether you think it's possible to make a really nice looking interface with c++ and QT
Well I like the look of QT apps :)
I just have to say I'm 100% behind jepler's answer there
I don't know of any reason a qt program can't also be a hal userspace component, which would communicate with the realtime parts of your system through hal signals
By the way the last version of this machine we built used National Instruments hardware and Labview.
in your place I'd consider whether I could use all real controls instead of a gui
depends on the problem domain and complexity of course
jepler: that is mostly was axis is anyhow right?
geo01005: no, axis uses the nml interface to talk to task (it's very gcode centric)
ahh I see.
I'd say machining centric, not necessarely gcode centric
it is (in theory at least) possible to swap out the interpreter in emc2 from gcode to some other language
doing that will probably isolate a _LOT_ of places where gcode specific things have been done in emc/emc2
I just have had a better experience working with hal that I did working with NI software.
anyway, the way axis sends "cycle start" is through nml, not through hal
that's the distinction I was trying to make
the effort is probably bigger than the clean GUI though
geo01005: maybe if you describe the problem you're solving we can be more specific with advices
is there some program involved in that machine?
or is it a predefined state machine?
Well I hate the position I'm in, I'm working on automating a proprietary production processs that I'm not at liberty to discuss but I'll try to explain what I'm working on a little better.
I guess the distinction is: does the process ever change?
is there some sort of "program" to change the process?
or is it a PLC rigid programming
Well the previous machine had pre-defined states.
and if the process changes, the PLC gets reprogrammed
I havn't used a PLC for this process, but there is pressure to move to a less expensive controller than NI hardware/software.
So the machine has an idle stage, a setup stage and a processing stage, as well as an e-stop stage.
durring the idle stage I need to be able to jog two stepper motors.
gotta run.. bbl
sounds like you want four knobs and a jogwheel
(or something like that)
It could be a simple as that.
There is also some process measurement that needs to happen.
and it would be good to be able to display some things like wire feed rate, fluid flowrate, and some other parameters to the user.
I would want to use all tactile buttons and switches, but just display stuff on a monitor.
it should be possible, but I think you'll face a fair amount of time coding the GUI and the realtime part
communicating between the GUI and the realtime part through either HAL signals or RTAPI shared memory regions is possibly unlike anything you've done before
you should really look into how much you need to learn in order to do this
and the time it takes to use a proper plc with gui support for changing parameters and stuff
there are no function calls across the userspace-realtime barrier, just shared memory (with HAL signals being a special case of it)
well so I was thinking of using axis and a pyvcp as a spring board to see if it is doable.
people always forget how much time they spend in order to save pennies on hardware.
by the way, I can't find many plc
if you have some dacs you could display things like rates on real analog meters
's fast enough to do what I want to do.
geo01005: what kind of cycletimes are you talking about?
and how fast scanrate?
cradek, yes that is a good idea.
ananimasu: about .5ms to 1ms.
that's not specially fast
I need to read a adc every scan.
the problem is that 90% of all hardware like that is filtered..
BTW i'm not terribly knowledgeable about PLCs, perhaps if you know of one that you think would do the job that might be a better option for me.
I think mesa has some very new product(s?) with adc
new enough to not have support in emc's driver yet
yes I'm aware of that.
geo01005: sigmatek, siemens, 3s(codesys) running on some random kind of hardware
I dont know alot more, mitsubishi(but they work like fan controller imho)
although the I was thinking that vital systems motenc-Lite boards are supported in hal?
neither one is really cheap, though.
yes there is a driver for those
mitsubishi is, but then they wont do the cycletimes you want :)
as far as I know, everything documented works properly, but I have never owned a motenc card myself: http://www.linuxcnc.org/docview/html/drivers_motenc.html
I understand that the motenc card dosn't have built in hardware stepper support like the mesa boards right?
that's right, it's servo-oriented
what step rates do you require? you might be able to mix smart adc with dumb step generation..
I think that would work fine.
i'm hoping to add spi adc support to the hostmot2 driver at fest
I'm thinking 10k to 15k
seb_kuzminsky:Do you think that spi support can be added by the end of the fest?
not sure, but i hope so
that would be cool :)
which reminds me, i need to pick up a MAX6675 and a K thermocouple to bring with me
* seb_kuzminsky goes to raid the supply closet
so are there any examples of the hal api with userspace applications?
the old halvcp might be closest to what you're looking for, since it's a C program that mediates between HAL pins and GUI elements..
[20:46:44] <jepler> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/hal/user_comps/vcp/?hideattic=0;only_with_tag=v2_2_branch#dirlist
but hey, if I use all switches and analog meters, then I don't really need to worry about that part right?
seb_kuzminsky: do you have any ideas about how SPI will work in hm2, or are you just planning to wing it at the fest?
for writing realtime components: http://linuxcnc.org/docs/html/hal_comp.html
geo01005: no real plans yet ;-)
seb_kuzminsky: good luck :)
free sample on the way - thanks Maxim :-)
If they don't come in time I can send you an extra.
i have a 7i65 to test with, it's got a couple of different spi chips on it
good night all