a stepper motor wire came loose from its screw terminal and shorted to case ground. the smoke came out of one of the 3 stepper drivers on my xylotex board
.. and the other 2 axes don't work either
anybody coming to workshop have a spare 3 axes of stepper drivers to sell me?
jepler uuuh.. bad luck... i dunno the xylotex, but maybe you can fix it quite easily
are there schems and some documentation avail? i guess its just some fets and eventually a driver or parts of the psu if not shortage-proof
roh: xylotex doesn't publish a schematic, but it's built on the a3977, a surface-mount chip which itegrates the step translator and the drivers
.. and it can be had for under ten dollars http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=620-1144-1-ND
man, what a shitty repair policy at xylotex: after a year they won't repair your board, even for a fee. buy a new one or gtfo. http://www.cnczone.com/forums/showthread.php?t=73307 http://www.xylotex.com/Repairs.htm
those things aren't cheap enough for me, geckos are the way to go
the xylotex performed just fine until this -- as far as I know, only g203v is represented as able to survive what I did to kill the xylotex, but that's still about 3x as expensive as xylotex
hm.. seems when something dies, its that chip
unsoldering that chip is probably out of the question?
eric_unterhausen: oh it's probably doable
the question is whether I'd rather spend a lot of money, or a little money and then a lot of time and then maybe a lot of money
is it one board for multiple axes?
3 axes on one board
but the failure is such that none of the axes are working now -- I assume there's now a permanent short inside the dead chip
(when power is applied, the power LED comes on briefly then fades out)
(and a little more smoke comes out of the chip)
yeah it's really made my day
philinmo boards then
hmm.. you guys ever heard of octopart.com?
* roh just stumbled on it and it seems to try solving one of electronic engineers longest and most painfull issues.. sourcing
never heard of it before
* jepler bookmarls
I suppose I should rip the dead chip off the board and see if the other two axes will work then
in general, check for any damages to other chips, or stuff like diodes
jepler, you can cut the pins, correct?
[01:17:45] <eric_unterhausen> http://pminmo.com/forsale/blankpcbs.htm
eric_unterhausen: yeah, I think I can do that after I get the heatsink off
well off to f*** something else up
that's the spirite
I had a wire come off of my servo amp, seems like they now cost about $200 on ebay
and $3000 new
jepler is your system also one a3977 per channel on an extra pcb? if so , just disconnect the control and vcc to check if the others still work
roh: no, it is one board
one ruined board, now
the pins are too close to the package for me to get with my sidecutters, but in the process of trying I destroyed several pads
i had better quit for the night before I sever a part of my body or something
anyone need some help around michigan doing retrofits?
anyone need some help around michigan, indiana, ohio doing retrofits?
Can EMC2 handle this application? Look at the stylus probing the contour: http://video.google.com/videoplay?docid=6002552788152244280
jepler: how did you kill the xylotex?
Jymmm: shorted a motor winding to case ground
jepler: how did ya do that?
Jymmm: not sure -- the wire apparently worked its way out of the screw terminals
can you take a photo?
Valen: there's not much to see
I was thinking of looking up the part numbers to see what the bits are
if its a surface mount chip you can use a hot air rework to get the chip off
the chip top is still intact, two pins on the side are blackened, and I mangled a few pins and a capacitor trying and failing to cut the pins of the chip
jepler: Well.... I talked to MAriss on the phone for a couple hours once... He said that NO DRIVE, not even gecko can survive a connector coming loose.
the main chip is allegro a3977
otherwise you can cut the pins with a knife
no mystery there
jepler: chip clip
is yours the PLCC ot TSSOP package?
ahh kinda sucky
the other thing you can do is hit each pin with a soldering iron then prize it up with a pin
Nah, just cut the chip out, and remove each pin invidivdually
yeah all of the above, but yeah sidecutters aint gonna cut it
you could prolly dremel it out
Valen: You must have some suck cutters =)
maybe another day
your not going to sidesnip the legs of a plcc
when I have the slightest chance to do more good than harm
so yeah i was thinking these for the drive http://secure.oatleyelectronics.com//product_info.php?cPath=96&products_id=189&osCsid=cd54f26f569b45fbe0d22615e18d17fa
for my mill project
with a set of 7I40's for the drive
going the cheap route
jepler: Well, it's a cheap board anyway
jepler: do you have any other driver available?
you could just disable the\at one axis and use something else
per the jumpers on it
Jymmm: no, jumpering doesn't work.
I have lined up a board to borrow temporariliy
Be sure to use extra fans on it too
oh well off to get some foooods
jepler: FWIW... I worked in a shop where we fabricated UV Curing systems for large (dragon) printing presses. One time I was wiring up the 120VAC logic and one single strand from an 18g wire touched the back panel. the 5000VA 440-to-120 xfmr blew.... the resin stunk up the shop for two weeks
that's more fun than I have a in a year
jepler: These were 600W/in UV lamps, 6 foot long. Fun Stuff
Jymmm: I have to content myself with the little 'pop' that first told me something was wrong
perhaps some day I'll work up to the big stuff
need more smoke?
I'm trying to quit
oh, I see
hm, is it incorrect usage to refer to a solid state device as the computer's "hard drive"?
I'll go for calling it only as erroneous as "floppy" for the ones that are about 3.5 inches on a side
solid is close to hard, so I guess it's OK
the disk is still floppy
it's the case that's rigid
the platter is solid, it's just the spinning that makes it unsolid
not with a floppy, even a 3.5" one
they're definitely flexible. trust me on this :)
though it is funny to think about the fact that CDR and possibly DVD-R media are now less expensive than floppies ever were
SWPLinux: did you have a chance to look at the video that describes what we were discussing a couple months ago with Dan Falck?
or I did but I don't remember :)
hmmm. ok, the optics shop
Do you want the link to it?
I never watched a video though
not right now, thanks. I'm on the road
some time in June would be better :)
I am just very curious about it
I think I'd need to think about it more than I can at the moment
big project in the works
Don't forget to let me know when you visit your brother
(no plans at the moment though)
Looking forward to it...
now this is strange
I think I ate too much sushi
that wasn't sushi
jepler: Oh, I thought you were working on something new, not troubleshooting
hey guys you know those indicators with the little arm on a pivot
like to tram your spindle or see how flat/level your table is
thats an arc measurement, right?
[06:33:23] <renesis> http://en.wikipedia.org/wiki/Dial_indicator#Dial_test_indicator
k big wiki knows
can anyone help me with phase driving some steppers to test emc2 before i pay to buy step and direction drivers
what do you mean
what sort of test
rather that driving the steppers by using step and direction drivers I want to step them by energising each coil
i belive it is called phase drive, driving each phase
thats what the driver does
i know but it is possible to use a darlington array and energise each phase to step the motor
i think this guy has done it but can't find any more information: http://video.google.com/videoplay?docid=5506285167473738661
i have done this before using software called majosoft from majosoft.com and using uln2803's
you either convert step dir in logic or a hal component and use extra pins on the par port
I'm new to emc2 and only just getting to grips with HAL i was hoping someone could send me an example :)
it may be in the distro or source already
i think i have just seen something under stepgen here: http://www.linuxcnc.org/docview/html//man/man9/stepgen.9.html
types 5 &6
archivist: just for reference
stepgen has about 15 stepforms
step/dir is only the most commonly used
I saw th page he found
also here: http://www.linuxcnc.org/docview/html//hal_rtcomps.html#sub:Stepgen-Step-Types
relying on google for search may not be the best idea
archivist:why is that ?
lands on on the wrong stuff
[09:45:18] <Valen> http://www.youtube.com/watch?v=tn7A9PqNftY&eurl=http%3A%2F%2Fwww.machinistblog.com%2F%3Fp%3D91&feature=player_embedded
heh and heres me using a bit of wood like a chump
get it wrong and the part flies out
the same could be said of many operations ;->
so yeah I'm gonna try the cheap motors, see how it goes
seb_ is now known as sebjames
seb_ is now known as sebjames
seb_ is now known as sebjames
hi ho, hi ho, it's off to work I go :)
well - stuarts first motel choice still has openings.
steves_logging is now known as steve_stallings
Hi skunkworks. Missed your comment to me yesterday. Great news on the servo. Bringing it to the Fest?
Matt and I will be at the Western Holiday. Between that and the Air Trans flight it should be an adventure.
heh - yah - I just booked the western holiday also.
I emptied and filled the various tanks on the bus yesterday...
steve_stallings: I am planning on bringing the h-bridge. (no supply or servo though)
hope I got them right
cradek: for everyone's sake
and all this time I thought by "the bus" he was referring to Greyhound
skunkworks: will your driver work on 12 VDC? maybe I could bring a servo about 1.5" x 4" to play with
a real bus for mobile workshop and to carry more?
steve_stallings: I think it should
Was thinking if we get our linear motors working, Do you think an aluminium framed mill would sell?
decent size one, not a baby mill
300x700x400mm working volume kind of thing
skunkworks: ok, I will plan on bringing it, perhaps a 24 VDC supply is I can find a small one, already shipped my 50 pound "kit" to Kansas so the servo will have to go as luggage
I fear that your FET drivers may have an under voltage lockout, thus the 24 VDC supply
24v should work great for sure - could always lower the max duty cycle also
I know I have run it on 18v
yes - 12 might be pushing it (for the bootstrap to charge and such)
steve_stallings: they rate it Logic supply + 10v.
robin_sz: Riding out the downturn?
robin_sz: greetings stranger
in the best possible way :)
still can't spell :D
as for the engineering company ...
I closed it
Wow - closed or sold?
made the stafff redundant and called in the receiver
just got rid of
engineering in the UK is stuffed
50% drop whereever yoyu look
steel stockholders down 50% as well on delivery .. so its pretty universal
yikes - Good luck. (is this retirement?)
gone back to software :)
retired? at 46? I wish
one word - plastics ;)
my brother works in the injection moulding industry, monitoring software
thats even more gutted than sheet metal
(that was a bad quote from the graduate)
i might set up a 3d rapid prototyping thing in 18 months or so
laser sintering of plastic powder
whne the recovery starts
skunkworks: thanks, I will see how sluggish the motor is on 12 volts
hows it going Alx?
alex_joni, hows it going? still selling bots?
robin_sz: yeah, still
glad someone is doing OK
err.. kinda ok
I still have some big contracts going (they take 6-8 months to complete)
afterwards I'm not so sure
things are also at ~50-60% compared to last year
some domains start to get started again it seems
construction and such
id agree with the %ge ..
I think it will be bad until hmm ...
for output, might see it start again in the autumn
yeah, that's what I hope too
certainly feb 2010
but new projects could be another year
people still feeling the pain of the recession
over here it wasn't such a pain
more that people were very cautious
thats why I got out, I knew it wasn't going to be just 3 months ...
even if they have money and planned investments, they decided to hold off
yep, here it was dramatic
[14:31:45] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=81210
IT work in australia seems largley unaffected ;->
robin_sz were you planning on buying or making the 3d printer?
Just got off the phone with mesa, ordering bits for the mill
cradek_ is now known as cradek
steve_stallings is now known as steves_logging
Goog evening at all.I have retrofit the 5 axis maschine .This maschine have more I/O.Mr Peter of mesa speak me the final project for spi interface.Some people have notice for this project?
[17:57:00] <skunkworks> http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,30/id,311/lang,en/#311
I found something interesting. The Feed in the GCode and the manual jog speed seems to be overriding the max_velocity limit.
I know I've set the max velocity for x and y axis using stepconf to be 20mm/s and 30mm/s, but I can jog the x axis in axis at 1800mm/min (30mm/s * 60s/min).
and in GCode, the rapid feed (made by mistake) to be 1500mm/min, and the machine will try to do it.
NewType: please pastebin your ini file and say what version you're using
here it is: http://pastebin.com/m49ad0c9c
I get warm fuzzies from emails like len's :)
in fact, G0 is going very very fast, and my steppers are missing steps!
I am using 2.3.0
skunkworks, I have been talking to some mach users... I see conversions in the future
what does the velocity display in AXIS say during a G0?
let me see again...
is this machine trivkins?
that is 17mm/sec which is less than all your axis velocities
nope. this is 5axiskins. but I am running 3 axis only.
yeah, but my MAX_LINEAR_VELOCITY = 10.0
so which wins?
I think that sets the top of the jog speed slider only
you might mean to set [TRAJ]MAX_VELOCITY
for that reason, MAX_LINEAR_VELOCITY has been moved to the DISPLAY section (but still accepted from TRAJ for compatibility, which might be a mistake to do)
cradek: yep. but that's not in the *.ini generated by stepconf.
so can I declar MAX_VELOCITY in there?
here's the doc for TRAJ section: http://www.linuxcnc.org/docs/devel/html/config_ini_config.html#sub:%5BTRAJ%5D-section
I have MAX_LINEAR_VELOCITY = 10.0 but it seems that's not what I should set.
skunkworks, one of the UK beta testers for the chinese lathes is converting
is that normal to have a XYZBC machine to still have the 3 joint or A axis to show up on AXIS?
somehow I need to keep this in to get it to Home correctly.
didn't you start from the 5axis sample config? It has X Y Z B C W. you can have any combination.
Bingo. The [AXIS_0] MAX_VELOCITY is working. :)
yep, I started from the sample config.
but it does not have an A axis
I am still setting up the *.ini. I set GEOMETRY=XYZBC, and COORDINATES = X Y Z B C, but I wasn't able to home the C axis without putting the 3 axis in.
I think I better restart the *.ini file to see where I did somethign stupid.
anyway, the bigger problem just got solved. x axis is not over the speed limit anymore.
you're in very new territory - if something is odd, it's ok to first assume it's something you did, but don't be too surprised if it isn't :-)
thanks, cradek! that's make me feel very good! :)
what's puzzling is the performance of the machine is different when I use stepconf to test it and when I use the axis to run an actual GCode
I set the x axis to have max velocity of 20mm/s, and acceleration of 50m/s^2.
during the test of +/- 100mm, the screws will perform fine (I have 2 screws to drive the x)
but when I run it in GCode, even at 10mm/s velocity, the steppers will still miss steps.
I've heard similar reports before, but I don't think anyone has studied it
I suspect there is a RT Timing issue. may be.
one relevant difference I can think of is that stepgen uses a different BASE_PERIOD when doing "test axis" than when it writes the file
I am testing the machine using stepconf now. 15mm/s velocity, 50mm/s^2 accelecration, running +/-150mm test area.
this is because during "test axis" it chooses a BASE_PERIOD that allows the axis speed to be increased during the test
when writing an inifile, it writes one that's "no smaller than necessary"
that suggests something you can do to test this theory: while running "test axis" do "halcmd show thread" in a terminal, record the "period" shown for the fast thread, and put that value in your inifile as BASE_PERIOD
my experience so far is that when there is a suspected delay event that causes a jurk, the x axis will miss step.
yes, those kill stepper systems
if you're getting that error, then your machine is misconfigured or incompatible with realtime
I am not getting a real time delay error.
well, actually once or twice.
what are you getting?
the error is only shown once per run. it may be happening often.
but I found that when I have the USB thumbdrive plug in, it makes things worst.
the error is more of a random event.
you should never see that error - if you do, you will have pain until you fix it
jepler: I'll try to find a better base_period to see if it will work.
cradek: I am retesting the latency now.
I think there is something about the network card, the USB, and other stuff that's causing me this pain!
most frequent cause is video
I am using a rather old machine, using its onboard video.
It is a dual PIII machine! :)
in that case, often, plugging in a pci video card fixes it
I'll see if I can find a cheap video card around. I don't want to buy hardware for this guy.
these parts should be easy enough to find around school.
a nice pci ati rage card or maybe a matrox :)
will the PWM generator cause more delay? I can slow the PWM period down more. it is only for the spindle speed control
acutally, it is already at 100Hz. I think.
what numbers are you getting from your latency test?
make sure you run glxgears also (from a terminal)
oh? glxgears. let me try
humm. big number. different from what I put in the *.ini
base thread, Max Jitter = 935189
somehow I put BASE_PERIOD = 100000 in my *.ini file
I must have been drunk
max jitter is way too big
for a stepper machine.
ok, you found your stalling problem
darn. dual PIII is really really powerful. why doesn't it work!?!!?!??!?!
did it happen when you ran glxgears?
and x-chat, and network/
try a video card as cradek said.
acutally, I just peak now, and I do have a PCI video card in the machine.
not sure which one.
bad latency when you start glxgears is a dead giveaway
darn. this is so sad.
woo! I have a onboard ethernet and a PCI ethernet.
humm. I can't tell what video I have.
Still trying to get used to the device manager thing in Linux.
lspci -v in a terminal
archivist: how can you tell what driver is being used?
* archivist just did it to find a new card
VGA compatible controller: ATI Technologies Inc Rage 128 RF/SG AGP (prog-if 00 [VGA])
that oughta work
boy - I would think that would work.
if it uses shared memory it will cause latency problems
that's set in the BIOS, right?
Can you pastebin the output of your lspci -v
yeah usually you can disable the onboard video in the bios.
wondering if it is an smi issue..
not on P3s
* archivist has to learn to drive second parport now
here it is: http://pastebin.com/d6fdb46ce
oh. you need a 2nd parport? what are you doing?
I only see one video card there
I have used the ati rage 128 also with no issues.
oh, I don't have onboard video. that's a serial port. ;p
NewType, 5 axis
NewType: you don't have any usb memory sticks plugged in - do you?
yeah, I have a USB thumbdrive in. and that usually makes things worst. I knotice that too.
hey, in x-chat, when your name is in bright yellow, is that a PM? like in mIRC?
no just highlighted probably
Re-run the latency test without the memory stick plugged in. (with glxgears and such)
ok. one sec.
oh, here we go.
Base thread Max Jitter = 10422
let it run
<20000 is pretty decent
I have had PIII vintage motherboards not play well with usb and realtime. (mostly memory sticks)
I am learning to use CNC toolkit to generate GCode for a hemisphere... so I am fighting on 2 front: the machine and the CAM software.
so USB is BAAAADDDDDD
great. another problem solved today. today is productive!
yes - on some computers
I was thinking I had to upgrade to run a XYZBC configuration. I already gut out a 2nd PIII CPU for this box. I don't want to upgrade.
the max jitter remains at 10422.
I will lable the box with NO USB on it.
let's test it with axis now. :)
this is sad. I have a 5axis machie, but don't yet have a way to generate a 5axis GCode to test this guy.
so hand-write it
I am learning that as well. :)
the 5axiskins is working.
actually, I haven't build the tool table yet.
but I will
this is nice. since I can configure max velocity on each axis, and my y axis is the faster axis, I can code all my roughing using the y axis. :)
NICE! The jurking stops!
yay for systematic troubleshooting :-)
I am going to put the config file up when this is all done.
and attach a hemisphere test code.
if I finish coding it. :)
interesting. x and y keeps the same max velocity.
you can do a decent hemisphere with something like g1 b90 c3600 (ten times around in a spiraly fashion)
well, I also need to make sure the cables on the head won't tango up and the +/- 180 limit is set right, etc.
I am learning. just give me time. :)
with the 5axiskins, I only need to specify the tangent surface, right?
I don't understand the question
well, this goes back to who is doing the math.
the CAM program asks me questions about the parameters and geometry of the machine.
so it will do the tool compensation, etc.
oh right, you have to agree on who compensates for the pivot length
but with the 5axiskins, the EMC knows how to position the tool.
yes EMC's kinematics do the compensation for pivot length (and tool length)
so all I need to tell EMC is the the location of the tangent surface. Yep yep.
given that 0x378 is the onboard parport what addr do I use for the plug in card http://pastebin.ca/1427757
as lspci does not report the base addr
XYZ are the tool tip point on the workpiece coordinate system, BC are the tool normal angle
archivist_emc: you just have to try them all :-/
isn't there a generic 2nd parallel port address?
I prefer logic and a right number first time
NewType: not for pci
archivist_emc: there's only six to try, you big whiner!
cycle time to try is a pain!
I suppose guessing is not in archivist_emc's nature. :)
its not the right way TM
I am running late for class. I'll report back when I get to the next step. :)
NewType is now known as NewType_away
will the keystick interface run 5axis?
or is it limited to 3?
your options are AXIS and tkemc
keystick is nice for an ultra minimal setup
it seems nice at first, but after I used it for a while I wanted simple things like a file chooser
File to open: _
uhhhh what was it again?
(it was long before the days of AXIS when I was using it)
maybe ill just run axis directly in X without a wm
you probably want a WM because some things pop up.
most of the pop ups are fine
its just the halscope stuff that has no exit button
doesn't halscope have file>quit in 2.3?
most wms should have a way to make the main axis window full-screen, no decorations, and always "under" everything else..
there has to be a better way of finding the extra parport addr hours of edit scope and button press!
can't you just scan the bus lspci
Jymmm, lspci gives 6 addresses...which one.... tried them all no output
I would say something is wrong then... ;)
but what? brand new card
* skunkworks was just being a D*ck
I have never played with more than one printer port
actually - I don't think I have played with anything but the onboard port
a few hours wasted is getting me very frustrated, docs are not written in a manner to diagnose a problem
archivist hi, what card is it?
given that 0x378 is the onboard parport what addr do I use for the plug in card http://pastebin.ca/1427757
as lspci does not report the base addr
[22:34:37] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?NetMos
From experimentation, I've found the first port (the on-card port) uses the third address listed (c000), and the second port (the one that attaches with a ribbon cable) uses the first address listed (b800).
is worht a look, it uses the netmos card
pjm__, I saw that
i had some bloody problems with a dual port card
i ended up having to initialise it with modprobe parport_pc
spotted one mistake the hardware drivers part of the manual needs improving on
then rmmoding the parport_pc and then running emc
and it fixed the problem completly
i added that stuff at the bottom which is what i have to do here to make the card work
if u lspci -vvv does it show any additional info for that particla
archivist_emc what happens if u use port 0x9400
i can look at my config tommorow with the cards, i have 5 ports running in my machine
9400 has been as deaf as any other port
ok try this ;
see if parport_pc reports other serial ports being found
if it does, it'll give u the proper base addresses
obviously u have to do ^^^ without EMC loaded/running
I see 9400 and 9c00 after that
in dmesg output?
u should see your 0x378 or whatever
for the onboard
so add in the 0x9400 and 0x9c00 to the relevant .hal
and try starting emc2
and no out put from the port
can u paste the line from the hal here?
hal file http://pastebin.ca/1427939
and have u checked both connectors pin3 for output
try loadrt hal_parport cfg="0x378 0x9C00 in"
i mean out
at the end
just in case the port is configured as input
in which case pin-03 wouldnt output
as per http://www.linuxcnc.org/docview/html//hal_drivers.html
out is the default
just to be sure can u move it to pin-01 and retry as that will be output in any mode
just to ensure u are chatting to the right port
gees just seen the time, can we continue tmw if u dont get it working over night?
Im getting bored with it now any way
i can check my machine config tmw and be on it to help etc
I bought two of the cards, first stopped the box from booting
are ther ejumpers on it?
is it settable via firmware?
oh, and why dont you just reboot the box and look for the base address during POST
reboot!!! Im not a windows user, lspci should give out the numbers
well, you ca either bitch about it, or just reboot =)
or cat /proc/ioports :)
$20 says he read alk's message 0.00003 seconds AFTER he he the reset button