(Sorry, closed length 60mm)
Considering a roll of shim steel is £40
serial number 195. yours?
hmm where do I find that?
andypugh_: and don't forget the hospital bill
We have social healthcare, its free. You should try it :-)
behind headstock on a plate, kinda low, to right of coolant stuff
haha as if
brb, I'll look
The HNC of the Beast?
look there is another socialized medicine person...
Screw it, I'd take government medical if it came with full dental =)
let's not talk about this, it hurts
This is getting horribly off topic, but I have read that you pay twice as much per-capita for healthcare than us, and don't live quite as long.
I am with cradek.. Also embarasing..
andypugh: But we have fast food every 2 miles, and starbucks that are only 100 yards from each other.
okay, so I may concievably, in a fit of OCD or something, replace the balls in this leadscrew and regain accuracy. is this pricey? Healthcare is free if you dont have any. lets talk about machines guys.
That was mentioned as a confounding factor, yes :-)
confound it. :)
I wouldn't worry about 2 thou on the Z if the head is heavy enough.
Mire: Great, I'll find you a leadscrew and you find me free dental!
the balls are not too expensive. it is difficult to guess the right size in one try.
I wont. but it's my hotrod.
andypugh_: it's a lathe, a little different story
Mire: where are you?
Mire: Feel like hitting LA ?
Even less of a bother with a lathe? Conventional lathes have 50 thou and work perfectly.
I'm looking at little tins of balls marked .1294, .1319, .1329, and there are still more of them :-)
Perhaps the ones you didn't use will work for him?
Mire: Feel like hitting LA ?
that is possible but I doubt it. iirc, they were mostly all too big
also one marked "bad" - the ones with the chrome crunched off
I know w place that you can take in your leadscrews and hand laod them with the perfect size balls for like $20/ea
I can't remember ever back-turning on a lathe. I think I would be happy enough with backlash compensation in that case.
jymm: yah possibly. I've been needing a reason to go that way. email@example.com is me.
I am practically certain they aren't plated.
in Mire's (and my) case I think the loose Z is what lets the velocity loop oscillate. I ended up turning the gain down on mine.
andypugh_: the ones that came in the screw were definitely two layers, smooth chrome steel over something smaller and rough
I don't know if when I order new "chrome steel" balls they are also two-layer
(but that's what I put in both machines)
The owner is INCREDABLY nice and VERY knowledgable. http://www.precisionballs.com/
jymm thank you.
yes that's where I always order balls
heh heh precisionballs.com heh heh heh
now they have online pricing! it used to be a real crap shoot
Well they have kits, but far better just to go in and preload your nuts with their balls!
nutt huhuh. balls.
I reckon that was spalling fatigue.
[00:14:45] <andypugh_> http://www.epi-eng.com/rotorway_helicopter/rotor_drive_system/secondary_shaft_bearing_issues.htm
Mire: if you decide you need to tweak your velocity loop gain, I have instructions. but you should be very sure before you take soldering iron to the amp!
"chrome steel" just means that they are alloy steel. (trust me, I am a metallurgist)
Wow, I'd better have a look at mine. how many HNCs do you suppose they manufactured?
I never trust anyone who says "trust me"
No, really.... I don't.
I believe that the new ones are solid if you say so - but clearly they all aren't solid because I have counterexamples in my hand
... trust me!
if you intend to use it a lot, I'd recommend checking to see if your balls are crunched up, even if you don't care about the backlash
you might be able to see crunch evidence by just removing the cover and analyzing what crap pours out
if it's clean oil you're probably ok (but I bet it won't be)
RotorWay apparently tears one of the seals out of each new bearing, and by hand-holding the outer race of the bearings to a bench grinder, they cut a crude notch in the outer race so grease can be pumped into the back-to-back pair. (This might not be unusual in the manufacture of hay balers and the like, but certainly is unusual as an aircraft practice.)
what i would say is puzzling is how did water get in there in the first place
If you open a ball cage BE SURE TO PUT DOWN A BIG SOFT TERRY CLOTH TOWEL DOWN ON THE BENCH
a black one
bunched up a bt too
Do your balls look anything like the one in the link?
good question since these lathes aren't supposed to be used with water-based coolant
cradek: I'll do everything else first before soldering. I would of course be interested in those instructions anyway. I'm hoping to document the machine very thouroughly.
dont you use cutting oil as coolant ?
SWPadnos: good point
yes, oil on these things
felt is also good, like jewellers use
AchiestDragon, but not everyone uses oil all the time, which could cause water to pool in the ball race
the rusty water was very surprising to me too...
then again, condensation can also do it over the years I bet
Ball bearings can fail with sub-surface crack growth, then the top layer flakes off and leaves the fatigue-cracked (rough) surface underneath.
interesting - that would explain it too
Mire: wish I had a good sheet-feed scanner, but I don't
I should get this stuff online - people who charge for it piss me off
If I had one you'd get electrical diagrams in return. I may try a good digital camera.
As for the Rotorway thing, if they ground the notch _then_ removed the oil seal I reckon there would be no contamination from the rocess.
condensation would i guess long term if its without oil , altough if the lead screw is stainless and ball berings are usualy crome i guess the rust itself is from another component on the ballscrew like the housing
I'd be surprised if the screw were stainless
may actualy be better packing the ball screw with grese rather than oil
machine is >30 years old, hard to say why it's in any particular condition today
we just have to explore and make the best of it
true ,, did not think it was that old
AchiestDragon: you should see the control box!
paper tape and pdp11 ?
paper tape, no cpu
[00:29:19] <skunkworks> http://electronicsam.com/images/KandT/conversion/rearcontrller.JPG
should not be too hard to convert if you have the schematics
We threw out the last PDP11 at work a few months back. A sad day, they had been running dynamometers perfectly well for 30 years.
sadly, you could replace the PDP-11 with an AVR microcontroller these days
i got shut of my pdp8 and pdp11/23 over 10 years back
hnc is new enough to be all ttl ICs, I think skunkworks's is older
But they don't, they replace it with a WIndows PC and a lot of expensive interface electrics.
We just made a ritual of deconstructing the control cabinet so we could dance around it... I havn't seen anything built this ridiculously tough since the Navy. Our garage is so much bigger it's like a different room without it.
We just made a ritual of deconstructing the control cabinet so we could dance around it... I havn't seen anything built this ridiculously tough since the Navy. Our garage is so much bigger it's like a different room without it.
"THIS SIGNAL CONTAINS I & K INTEGRAND DATA ADVANCED BY 3 DIGITS. D6 DATA OCCURS HERE DURING D3 TIME." (arrow pointing to flip-flop 319-5
As for the ballscrews I think you can get stainless ones, but most are carbon-chrome steel, which is fairly corrosion resistant but not technically stainless.
anyway have a emc question ,,, ok so got both linux and windows pc's here ,, but the problem i have is i use suse in a desktop config for my machine ,, i dont want to have to reboot in order to run emc with a dedicated kernel
I'm keeping the checktape as a souvenir...
"LOGIC 0 for 2 OR 27 PULSES OUT OF 127 INPUT PULSES"
ahhh anyone else recognize 127?
is there a way to patch the suse kernel and install emc onto that
Yes, but it is work
AchiestDragon: you can run emc in simulator mode without a realtime kernel if you don't need to control machinery
[00:35:11] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Preparing_other_versions_of_Linux_to_compile_emc2
SWPadnos: but unfrotunatly it does need to control the machine
then yes, it is a lot of work, but possible
that's what those links are for
Say, someone suggested to me that the 5V 50 amp supply we lunked out of this cabinet mite be good for electroplating. would it work?
should be good for annodising also
Great. one more hulk of ruin to hang onto...
and electrolytic rust removal
thought it was too low v for anodizing
jeez, im keepn it for sure
12-20V would probably be better, but 5V may do
generally, the process runs at 1-3V, and the voltage goes up when the anodized layer is thick enough
(that's how the controller knows to stop)
[00:41:05] <andypugh_> http://forum.caswellplating.com/anodizing-questions/10346-anodizing-voltage-question.html
That says 15V minimum
hah! cabinet came with +-12v@15amp supply as well! Anybody need some NIXIE tubes?(forgive spelling)How about a 100lb transformer with about 30 taps?
np you would need another 2 of them also then
Nixie tubes sell well on eBay, popular for retro-clocks
they aren't nixies
they're numitrons (incandescent)
drat. i didn't look that close.
not worth anything because they burn out quick
good. I was feeling great about putting garage on a diet!
well they might sell - I dunno - just not to me :-)
oh ya yr the time guy... do u make those things?
Who? and what things?
clocks w nixie tube displays. saw them on bb.
[00:49:19] <andypugh_> http://en.wikipedia.org/wiki/Nixie_tube
Has some clues
[00:49:42] <skunkworks> http://timeguy.com/cradek/clocks
[00:49:51] <andypugh_> http://www.electricstuff.co.uk/nixclock.html
I decided to make a clock about 18 months ago. Then I decided I needed a CNC lathe and mill to make the parts. I nearly have the lathe and mill now...
whoever hasn't made a clock, raise your hand
* skunkworks raises his hand...
okay enough fun. now to lathe for more workfun. You guys are inspiring. I'll let you know what I discover. oh, and hand down i'v made me a clock or two.
cradek: I love the crt clock.
made a led clock some years back 184 leds and ttl
I made one with laser line generators (coaxially in glass tubes) to project red and green hands on the whole wall. Then decided I didn't like it.
seen someone doing plans for a wooden clock that could be done with a cnc ,, maybe something i may do if i get this new cnc working
But it was too much work to throw away. If anyone wants to pick the project up....
[00:56:41] <andypugh_> http://www.atp.pwp.blueyonder.co.uk/BigHand.jpg
[00:57:15] <andypugh_> http://www.atp.pwp.blueyonder.co.uk/CanYouTellWhatItIsYet.jpg
If you want it, you can have it.
* skunkworks has way too many projects...
It needs a stronger drive, probably a synchronous motor
And it also needs to not be mounted in a frying pan, but it was late and Tesco was the only place open
And it is late again. Time I was elsewhere.
smog and tags complete! Ya just gotta love the interwebs!
any opinions on how well 'lash comp works in EMC?
hey good news... seems I mislead myself with harbor freight dial indicator. Careful checking shows backlash under a thou in Z, one or two tenths in X.
turning FERROR up around .005 i was able to run both axes up to 100ipm, which is all I want.
Now I'm playing with tuning; overshoots are my limiting factor as I run up P with I and D zeroed. I'm hoping to be able to turn following err sensitivity back up, rite? what's my next step?
Sent here from the RepRap guys.
Here's my delima. Trying to convert a Seiko D-TRAN XM 5000 Cartesian into a CNC/3D Printer.
[02:48:07] <keebler> http://gallery.me.com/keebler64#100035
It has 3-DC Servo Motors, (2-200w for Y and Z) and a 500w for the X.
I'm sure emc can run it. As far as I know though emc only talks "gcode"
EMC2 doesn't have any servo code?
it controls the servos, but the command interpreter takes gcode.
note: I'm pretty much a novice on this large of robotic equipment.
Looks like the makins of a Mighty bot indeed. RepRap guys referred you here?
gotta scoot for a few.. bbl
Mire: can you show a halscope plot of ferror vs velocity for a move that's long enough to get up to full speed and cruise for a bit? I can maybe guess what you should adjust then.
cradek: I hate to have to admit this but I haven't gained much ability with Halscope yet. That may be the advice I need for now, I'll go play with making plots.
ah I'm surprised you got it tuned at all then!
you want to plot axis.0.f-error and axis.0.velocity-command or however it's spelled
then get it to trigger at the right time (either trigger on velocity or position)
I like to make a gcode program that rapids back and forth with pauses in between. then once halscope is set up to trigger, it will give you a constantly updated display while you tune
the rapid should be long enough that the axis cruises but not much longer so you can capture the whole thing
awesome. this may take a while dinner is soon. thanks.
ah right, you're in the past, around dinnertime
maybe I'll never get used to time zones
sometimes I talk to people from the future too
cradek: QUICK! GIMME THE DAMN LOTTO NUMBERS FUTURE BOY!
machinists keep odd hours anyway.
jymm: 1, 2, 3, 4, 5
Valen1 is now known as Valen
Valen1 is now known as Valen
MarkusBec_away is now known as MarkusBec
Mire: But yeah, mainly because due to the power requirements, and lack of servo code/support in the RepRap software, they thought I'd be better suited looking into EMC2.
Granted this was after I placed an order for 3-RepRap motherboards and 6-Extruder Controllers
Valen1 is now known as Valen
Hi Keebler, I'm new to emc2 and had been reading about it for quite sometime now and now that i have the hardware in place, might be the right time to start.
I have an motorized IDC XY stage with encoders and limit switches and although another motorized stage suitable for the Z axis, it's not like the XY which has an encoder and limit switches.
new that is
These are used stages from a scrapped inspection station and I'm thinking of a PCB router or plain router. Micges, you're welcome to join the discussion.
I have a few questions to start, if the single parallel port is not enough to accomodate the XY stepper motors' step/direction signals and also the encoder signals, etc. will another parallel port be enough or one of those those MESA boards be enough?
mesa gives you more options
If I go with a stepper servo setup (with encoders) and also spindle control for up to 5 axis (currently with 3 but would want to expand later) and maybe other I/Os I haven't thought of, what MESA board would you recommend?
I dont have a mesa card yet, others do, im still struggling with not enough pins on the parallel port
the usb thing?
[08:10:51] <oPless> http://www.automation.com/content/mesa-releases-7i43-fpga-io-card
I'm currently checking MESA right now and cost maybe an issue, another parallel port maybe? How many I/O can the parallel port handle for emc2? I don't know if I'm typing a lot here, just my first time to join a chat forum.
* oPless goes to work
Thanks oPless, i'll start saving for the board.
By the way, has anyone tried this board with EMC2?
On the software side, what freeware software (CAM, gerber support) works with Linux?
[08:21:44] <celeron55_> http://www.linuxcnc.org/docview/html//man/man9/hm2_7i43.9.html
so it's supported
usb connection is not, but parallel port is
at least that's what that page says
Another question: Is there a way for emc2 to change the motor step frequency on the fly?
in what way, Axis has a slider which can vary the machine speed
in precentage of the speed set by the program or the user (in manual control)
hmm $59 in hundereds off ,,, $5900 is a far bit to save
it's about $100 a piece :P
so about £75 gbp exluding shiping and tax
expencive for a 200kgate board
[08:50:52] <AchiestDragon> http://www.enterpoint.co.uk/moelbryn/darnaw1.html
although its 3.3v logic on the i/o its not a real problem to accomodate and a 1200k gate device with ram also
same company do some 200k gate 5v tolarant modules for about £35 although minus usb
there should be enough gates in the 400k versions to allow for you do a far ammount of stepper control in the vhdl ,,, ie rather than the pc sending the individual step pulses to it , you could give it a value of steps ie move +20 X +2 Y and +5 Z and let the fpga actualy sort out the timeing of each
the mesa boards all have the stepgenerators in the VHDL
even the 200k gate versions
should be posible to accomodate a up to a 60mhz step rate
the loop is in emc not the external hardware
the nice thing about the current VHDL is that you can mix&match servo's and steppers
e.g. stepgens & pwmgens
but to use a new board some lucky person has to do the fpga
and the emc2 driver
archivist yea i know so the step rate max is limited to the pc i/o rate about 1mhz
so that's probably a 2 month project right there
I think the mesa stepgens work up to 30MHz or so
but it's not really usefull
emc2 feeds velocities to the stepgen
once per servo cycle (1 msec)
keep running at 200 units/second
then the stepgen in the VHDL knows that it needs to output 153.6KHz pulses to reach that speed
only problem i have found over using fpgas is the fact that only the smaller capacaty devices come in PQFP style packages ,, even the small ones require the use of a 4 layer pcb ,, sort of precludes home construction dew to pcb costs
so you endup stuck using prebuilt modules , they only seem to have a market life of 3 years at most as there "development/evaluation boards" not consumer products and again are rather pricy
Looks like the 7I43 is not yet complete in itself to interface to the motor drivers and encoders, switches. If I use an opto-isolator board be it MESA or other makers, would this complete my hardware (PC side)?
there is a sloution but will take 6 to 8 months ,, i could develop a fpga pcb for use with emc , and probablay other such i/o apps and make that gpl also
with provison for opto isolators and the 5V level converters needed to interface with most motor/stepper drivers
a cordless drill servo 8) http://celer.oni.biz/~celeron55/random/2009-10/totallyrandom/01102009.jpg
quite useless for anything serious but fun
you are my new hero
my current cnc project is sat on the floor in bits atm ,, broke a finger yesterday so its going to be staying in bits there for a while
that enclosure looks like it was made for servo drive electronics
i happen to have a couple of those lying around
[09:33:21] <AchiestDragon> http://www.whipy.demon.co.uk/dscf2323.jpg
< a 3 X 15A Hbridge module prototype
realy wanting to get the new machine finished so i can route out pcbs save me having to get them made for prototypes
that servo control board was the first pcb i've ever made. I made it with UV+etching... easy and precise if you have the equipment
i'm now in university and they have a nice workshop in there 8)
Good night guys.
made a few in the past that way ,, but not practical to have etching chemicals arround
after time because the feric cloride works better warm it tends to evaporate when i use ,,, you then find that any metalwork in the same room as starts to get surface corrosion including stainless that it will give it a perminent stain
and you still have to drill the boards after eather cnc or hand
can't have chemiclals arround here with kidlet realy
a machine you can stop or disable from beeing a porblem
650mph collison with a geo metro... http://www.youtube.com/watch?v=MWJU6sbf8Ng
[10:28:02] <AchiestDragon> http://www.whipy.demon.co.uk/dscf2326.jpg
< the machine i'm building atm
ok so 19" rack sliders are probalbaly not the best thing to use but there rated at 65kg
fenn: Kinda slow
yeah i guess they should have used a railgun
I just want to know where the hell the engine block is
I figure tht would have been the biggest 'chunk' of material
that's interesting, indeed
shatted inot tiny bits probabaly
There were no bits. I suspect they stripped that car pretty well. Not even a driveshaft or steering wheel.
Who was it here that made a "pallet buster" ?
OH geeze, that's WAY TOO MUCH work to break up just one pallet, even with a pallet buster http://www.youtube.com/watch?v=ZotsRhNhQLU&feature=channel_page
why not use them if they aren't broken :P
for the lumber
most are rough cut white oak
This is even worse... http://www.youtube.com/watch?v=ZtPmUZ-k3aI&NR=1
making pallets http://www.youtube.com/watch?v=DH1WyFBecdg&feature=related
tomp is now known as tom3p
[Global Notice] Hi all, unfortunately one of our former developers has left behind a memory leak in our ircd software which means we'll need to restart several ircds over the next few days. We're going to stagger it to reduce disruption, and the first round will be happening very shortly. Affected users for now will be about 1300. Apologies for the inconvenience!
archivist: can you test a threading improvement?
what do I need to do
do you already know how to get a git checkout and build master?
never touched git yet
I touched it atleast twice...
looking for instructions...
wait - what are we talking about? ;)
[14:04:08] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Getting_the_source_with_git
The few times I have git'ed - it worked smootly.
[14:07:28] <cradek> http://timeguy.com/cradek-files/emc/0001-Improve-initial-threading-synchronization.patch
cradek: what did you do?
used feedback over more servo cycles to guesstimate the initial velocity that should be matched
ah - neat
previously it used just one servo cycle, now it uses all of them while the axis is accelerating
it's another assumption - that the spindle is constant during this phase
heh, this is a slow start spindle
fenn: I made a single coil rail gun in highschool. (charged a bank of caps to fire it) shot a BB quite well. (went thru a few sheets of papar)
always wanted to add more stages.
ok download my patch and apply it to your git repo with "git am [file]"
then compile and test
tomp is now known as tom3p
just getting the build-essentials etc
apt-get build-dep emc2
probably the easiest way
configure: error: version.h not found - Is the kernel headers package installed ?
did you link kernel headers directory to /usr/src/linux?
no don't mess with making symlinks
you just don't have the rigth packages
did you apt-get build-dep emc2 like SWPadnos said above?
that's 2.1.2 on that page
you kind of have to jump around to get all the instructions
Bunny Wabbit Installation?!
which dir should I be in when I do git am 0001-Improve-initial-threading-synchronization.patch
just inside the git repository
Valen1 is now known as Valen
libnml/cms/cms.cc:319: internal compiler error: in gimplify_expr, at gimplify.c:4217
this being an old dapper box
mine is dapper too, and it builds fine
if you try it a few more times do you get the same error?
g++: Internal error: Segmentation fault (program as)
different every time?
that means your hardware is bogus, and usually means bad ram
yup and I can start up again
you should probably run memtest for a while, sometime
I wonder if my unclean shutdown of emc is in the way
if you dont click a dialog for a toolchange you can stop axis and get an orphaan dialog box you cannot get rid of
Is there a way to get axis/emc to save the encoder positions when it quits?
* archivist fixes by the windows method...reboot
Regarding G code parameters.... Are some of the parameters automatically loaded with system numbers such as X axis position, etc Or are they basically empty space until values are loaded?
If they are all empty - how can I get system values such as X axis position into a # parameter. I know about the hal interface and pins - so I can get to those ok. If this should be obvious - tell me where to go to read about this please ....
a lot of undefined reference to `CMS_SERVER_xxxxxx errors
archivist: that is baffling
it should very easily build
tomp is now known as tom3p
just done a make clean and retrying
(I don't trust your computer)
archivist, and also make install / make setuid?
Dave911: some are filled out with various things. for instance offset values and results of probing are in certain parameters
(for installed and run-in-place, respectively)
(and of course, you should be using run-in-place)
(which will require you to source scripts/emc-environment)
I've seen people use g28.1/g30.1 to read the current location
id do run-in-place on the ./configure line
and then sudo make setuid
but if you want to do something "from here" the simplest way is to use G91 mode. can you describe the problem you're trying to solve?
then . scripts/emc-environment
(that's period space scripts/emc-environment)
the fanuc books have this wonderful sideways square braket to denote a space, like an underscore with little ends. its real easy to see spaces even with non-fixed width fonts
SWPadnos: Is that anthing like .+.com ?
SWPadnos: buıʎɐs ɹnoʎ buıɥʇ ɐ puɐʇsɹǝpun ʇ,uɐɔ ı
you've got to be joking
tom3p: LaTeX has "verbatim*" mode which does that space thing too - can't say I like it though
I have 2.4 pre up
must have had something around screwing the box
I may try and duplicate a "dirty shutdown" bug
there is a bug, if you set the pindle too fast you dont get a following error
>>>> cradek I was having a discussion with another guy about Gcode implementation and he was saying that with a Fanuc controller he can access all of the XYZ coords and also the offsets from the Parameter list but he could not find anything similar in EMC2 - I've looked for a pre-defined parameter list and could not find anything either. Then I thought perhaps there is a mechanism to...
not sure if I can hear/see a change with the patch
...link Halpins to Parameters but I can't find that either.
you can't link parameters to anythin
Dave911: the emc interpreter does not have full introspection in that way.
archivist: hmm. does that mean it works ok, just like the old version worked ok, or that they are both bad?
cradek> Can you elaborate ...? :-)
you cannot access all of the interpreter's internal state from the gcode
but I wonder what problem you are trying to solve - maybe I can help with that. Or if this is purely academic (because another interpreter has this ability) that's ok too.
* AchiestDragon the down side for you if i get emc2 running is the number of feature requests i may file
AchiestDragon: The upside for you is that your contirbutions to the enhancement and devleopment would be appreciated!
OK, but you can load and modify parameters.. via Gcode of course. I and the other guy I was chatting with tend to bend things in directions that are often not considered "standard"
yes you can use parameters for all kinds of flow control and calculations
We can call it academic but I will probably try and implement it.. ;-)
i need to allocate a machine to it .. but did not want to have to tie a pc up as a dedicated cnc machine controler
Well you mentioned G28.1 and G30.1 perhaps I should take a look at those.
Is there anyway to load a parameter via a mechanism outside of Gcode? A script, obscure hal command etc?
guess first feature request would be a rpm package format install also
cradek, perhaps andypugh should try as well as his seems worse than mine
AchiestDragon: we have always offered to host release packages that anyone wants to build and maintain
Dave911: load what into a parameter?
Say I wanted to load the value of some hal pin into #10 ?? Pick on - following error on X axis for instance - I think that is in there someplace ..:-)
I meant pick one - not pick on
bk later , going to recover a sutable machine
cradek, got to go out and fix car while daylight
archivist: have fun, thanks for helping test
I was running g76.ggc with a 4mm pitch as I was the yesterday
OK, I'll study this further... Where are you at Cradek ??
onn g0 move does seem different though will investigate later
Anybody here use any of the Keling stepper drivers?
>>Anybody here use any of the Keling stepper drivers?
I have talked to a lot of people who use them. Most of them are various Leadshine models.. I'd avoid the lower end ones...
>> cradek - I found what I was looking for. Thanks for the help!! :-)
I'm going to buy a KL-6852 driver just wondering how good they are.
They look an awefull lot like the Anaheim Automation drivers, so I was wondering if they are made by the same company.
$270 from Anaheim, and $86 from Keling.
It sort of makes me laugh to see how many products get repackaged under different names.
note that the drive specs are 3.7A RMS, the 5.2A number is a peak
(which is shady practice, in my view)
also, they explicitly say that 75V (the spec sheet link is for the 7552) is the supply voltage, but later on they say that 75V is supply + variation + back EMF
See any problems with that driver driving this motor? http://www.anaheimautomation.com/manuals/stepper/L010378%20-%2023Y9%20Series%20Spec%20Sheet.pdf
Those are peak currents in the motor data sheet I believe.
no, but that's not the least expensive drive that can do it :)
(the Gecko G251 or G250 should handle it as well, and is $50 or so)
I want to microstep at 1/256...
I don't see why
microstep positions aren't reliable
I don't need good absolute position...
I know that I won't get the benefit of most of that due to friction and other effects...
what benefit do you hope to get by doing that?
then the question is - why do want to microstep at 1/256
So I have this EDM process that is running with a stepper with 1.8 deg/step and micro stepping at 1/8th steps.
Jon_geo01005: Are you working with micron tolerances?
Due to a goof up in a calculation I ordered the wrong gearbox (1:10 I should have ordered a 1:100) and I don't have quite enough resolution to stabilize that arc.
Jon_geo01005: and are you connected via paraport?
So I'm trying to increase the resolution of the motor before I buy a new gearbox $$$
no, I'm running with a 5i20.
microstepping isn't going to help. even half stepping isn't very accurite - position wize.. you really can only count on full steps. microstepping really only helps smoothing out motion and resonence.
has the gearbox got anti backlash
wonder if a small servo with a high resolution encoder would work better
gearbox has about 5 arc min of backlash.
a 1024 line encoder (super standard) gives you .08 degree resolution
I only ever use half step because the rest is snake oil
you'll have trouble getting under 1? .5? degree with a stepper
belt drive to that gearbox maybe
yeah but it would take two belt stages to get that 1:10
well I will get better resolution with .9 degree motor...
and like I said, I don't need positional accuracy.
I just need it to move less than it currently does with the same number of steps. (I don't know if that makes sense).
in hind sight I should have used a servo motor.
what you really need is smaller position steps right?
er you just said that
I take it - you're not using emc?"
just the hal.
lost again - couldn't you change whatever hal setup you have to output less steps?
Dave911_ is now known as Dave911
I really need it to physically move less.
Well, any body know of a good BLDC (nema 23) brushless motor setup for under $400 that I can get/install/tune by Tuesday morning? :)
you can do 10-1 on belt at a push
one ally plate some holes two pulleys and a belt
I'm sort of stuck with the current drive train because of some dimensional constraints, so adding another belt drive isn't really an option.
>>Well, any body know of a good BLDC (nema 23) brushless motor setup for under $400 that I can get/install/tune by Tuesday morning? :)
There is a guy in Canada selling setups like that for less than $400. I can find the website link if you are interested. Actually I am not sure it is Nema23 might be a metric motor. They do have nema 34s about the same price. Pretty cheap. I haven't used them yet. They are also selling them on Ebay.
Sure, I would like to know who that it.
I'm thinking a keling motor: KL23BLS-95: $26/pcs
tomp is now known as tom3p
A mesa 7I39-LV $149
and a AMT102-V CUI 2400 ppr encoder $30 on digikey.
so that is about $250 with shipping.
These guys have really inexpensive servo setups. I haven't tried them out yet but I did have a fairly lengthy email exchange with the company guy. He answered all of the technical questions very quickly and they were not the "normal" questions..
[16:39:02] <Dave911> http://www.dmm-tech.com/
If anyone buys one of their setups and tries it out please report back.. :-)
Wow, those are fantastic prices!!!
so is the honest opinion from everbody that a .9 degree motor with the 1/256 step driver won't have much if any better resolution that my current 1.8 degree 1/8 step driver?
that's my vote
for very low EDM speeds, you may have issues from stiction and variations in the "back pressure" from the fluid, which will create a varying load on the motor, which will make the microstep positions inaccurate
you also won't get any benefit from "smoother stepping", since you'd be running the motor too slowly
hmm, but do you think that commanding 1 step at say 1/64th step will move the motor?
A .9 degrees stepmotor would have a steeper torque/deflection angle slope
so if you were running it with large loads at slow speeds it might be a little better
(there is very little load on this motor ~ 1-5 oz inches)
microstepping at slow speeds wont move anything till stiction is overcome
.9 degree motors are probably less efficient (flux leakage) and more expensive as well
so how about comparing that to a bldc motor with a 8000 count encoder...
I think the main advantage of higher microstep ratios is less noise at low speeds
in step or servo tuning for edm, the smallest command that actually moves is an important thing to test. ( 1 step or 1 bit on the DAC, does it really move?)
of course the accuracy of the bldc motor will be better, but could I actually get it to move in any smaller steps that the stepper?
currently 1/8 step always moves the motor
At slow speeds with static loads a servo can creep along line a stepmotor, 1 step/encoder count
move at least 1/2 of the command ( if you ask for .0001" position, make sure you can move .00005 [better if you can move .000025"])
but with a hair of backlash that goes away right? (changing directions alot)
then the edm can adjust for an error below the requested position
do you have backlash on the position detector ( the encoder )?
There would be no backlash between the motor and encoder.
if not, the servo will move to position and 'handle' the backlash, (tho backlash is a bad thing, ok if tiny )
I wish I just had a servo I could throw on there and see how it performs.
the servo may get you smaller 'increments', its all about being able to respond to tiny changes. if the stepper can move ok, then stepper is ok ( sort of calculus... using a bit or a step doesnt matter, changing to the correct voltage difference anode/cathode is important)
theres some point where the gap error isnt important also ( like do you respond to .1volt error, .01volt, .0001? ) at some point it doesnt matter, at some point you cant respond.
I'm thinking I should go remeasure and make sure that it is moving every microstep I ask it to.
i bet 1/4 micron is great for most work ( what i use now , so i really do bet on it )
it not, then I think that I can safely conclude that more microsteps won't do any good at all.
4 count to a micron was notciably smoother than 1 um, but i didnt try 1/8.... it got hard to scale the amplifiers output ( yaskawas and panasonics)
the top speed suffered
yeah, I need about 10X more resolution to get to 1/4 micron.
if the workin area is large( > 1cm cross section) you may not need submicron
(hence the error in original calculations)
if its tiny (bonding tools or tiny holes) then 1 or 1/4..
you'd have satisfactory but >maybe< not optimum with 1um
for 1/64 dia
oops 1/16" dia
yeah 1.5mm dia is coarse enuf, most taiwanese hole drill have 1um feedback linear scales, and they can cut well
Yeah, I just measured it... it does in fact move every microstep commanded a distance of ~.0003 inches or 7.6 microns.
however, I can visually see on the dial of the runout gauge the vibration of the motor due to microsteping.
the backlash at the output is a little less than a single step so less than .0003"
so say that I were to get a .9 degree motor, from what I understand that will automatically double the resolution (correct me if I'm wrong) that get me down to about 4 microns.
Then perhaps the additional micro stepping may give an additional doubling effect...
so that would put me down to about 2 microns, if I'm lucky all the way down to 1 micron.
That ought to be good enough to get my data for my thesis.
(it works alright as it is...)
I'll let you all know if I fall on my face or not :) but I'm going to try this stepper setup... If that dosen't work on Monday, I'll have to get a new gearbox.
Thanks for the help.
down at those levels of movement, elasticity in the machine shows
to the extent that a person walking by will flex the floor and it shows
yes, my machines are massive, but i can move the tooltip by hand ( with a grunt, i can deflect the Z .0005" at full extent ). so not the stiffest, but way better than a 'test mockup'
its gotta be stiff to respond to tiny errors
can you believe that my current setup is mounted on a piece of 1" thick UHMW ?
Jon_geo01005: do you use a guide? a guide just above the work reduces tool wandering ( just like a drill guide, just a few um over the tool dia)
tomp looks up uhmw
This isn't really and EDM machine, it just uses the principals of edm. Instead of machining a part, I'm producing particles.
Ultra high molecular weight polyethylene (UHMW)
yep, i use 'greenglass' when i prototype. very stiff, very insulated
but yes there is a guide.
tom3p do you do EDM for a living?
* Jon_geo01005 thinks he has asked this before.
a coarse judge of ability to control the gap... a long ( like i use 24" extended on .026" dia tubes)... a good gap control wont bend the tool. ( think 'push up a rope' ;)
yes ( or did until yesterday ;)
that is too bad.
i built edms, i'm very intersted in your 'particle production'. my recent work was all medical & aerospace.
btw: a huge effect on the process is keeping the gap 'normal'... so removing any particles is necc to get the environment back to normal ( flushing ). the faster you remove particles, the more flow is needed, and thru a tiny tube that means pressure
1100 psi thru a 24"long .009" dia hole is barely adequate. it wont even dimple your finger.
we are doing a pressure relief hole of .2mm for someone
8thou is small, hope its not too deep
not hard with a typical edm drill.
i've done 7thou at 1" ( tricky!) but 60thou deep is ok
we dont have edm so will be using normal drills
peck like crazy or use a 'sensitive drill' ( a hand fed sliding spindle )
we do small holes on a regular basis, clockmakers
the real problem is that we are drilling cheap pipe fittings that have a pip in the middle
sensitve drill http://www1.mscdirect.com/CGI/NNSRIT?PMPXNO=1785669&PMT4NO=71687675
oh a lump casting... like mismatched mold halves
likely right where you need to drill
exaacly where we are supposed to drill, the parting off pip
file a flat ;)
one of the pips is in a hollow
heh, where do you get a .006" center drill
need enuf done to pay for sending them to me? ;)
we can hand center under a microscope
nah only 60 off
a spindle mounted microscope is handy for that stuff
and the prototype qty is about 6
with cross hairs, likely a cheapo web cam can do similar
I have stereo zoom on the lathe
ok, off to the store bbl
What on earth has happened to my Z-motor I wonder?
It used to run at 1500mm/min and now will only do 600mm.min
andypugh: if you get that figured out, I have a patch for you to test that may improve the initial spindle sync
archivist tested earlier but I think his results were not conclusive
um, I lost the url
The same amp drives the same type of motor as well as it ever did on the mill z-axis. That motor is equally poor on another amp, so it is the motor.
broken wire? overheated and shorted a turn?
if using parallel windings, is one open?
I have looked and there is no junk in there, and my second guess that one of the pole stacks had come loose is definitely not the case.
Series windings, resistance checks out right, no sign of damage
does the shaft feel right when you turn it?
Yes, it all feels fine
andypugh, see if you can put a couple of volts on it and measure the stall torque... compare with your other motors... maybe the drive faulted and demagnetized the motor
(Thanks for the patch, I will try it when there is some performance in the axis again)
That is the sort of thing I am wondering about.
short each winding in turn and make sure it makes the motor turn harder
I don't suppose running a 2.1A motor at 2.5A would do that?
measure the current, dont believe the switches
if you have the specs, one of them will be the torque / amp
The same amp runs another motor exactly as before, though.
I don't have torque/amp. I have a torque/pps graph, but I assume that is at rated max current?
There probably isn't much point worrying about it, I might as well just buy another motor at £25
and may assume half step driver
andypugh: or two
Do you think I am made of money?
compared to me yes
I hesitate to guess what anyone is made of
You do realise that this is all hobby work for me?
I believe I am made of slugs and snails and puppy-dogs' tails.
hobby is a very fuzzy concept 'round here
Very expensve hobby
Not compared to motorcycle racing, for example.
I suspect it depends on degree. I would expect to get a nice set of CNC kit for the price of a MotoGP team. Whereas my small-scale Enduro habit costs less than my mill/lathe conversion.
cradek, with the rotary table thread, will it be able to set an exact start angle, just wondering about loss of gearing/ creep after n cycles and a lot of revs
Newbie question, how do I apply a patch to emc?
I am running 2.3.0, so I assume first update to 2.3.3?
hmm I tested with the 2.4.pre version
else its git am 0001-Improve-initial-threading-synchronization.patch
assuming you are in the repo and so is the patch
I am scared of cutting-edge stuff. More so after a swarf-based incident this evening has left my finger and thumb rather leaky.
I actually did git am ../0001-Improve-initial-threading-synchronization.patch
leaky! you dangerous person
Hmm, can you explain in a way that a Mac-user might understand :0)
mac no never heard of them
ok you need a git checkout http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Getting_the_source_with_git
OK, I reckon I can handle that
then apt-get build-dep emc2
I might do it tomorrow, as I have just remembered that my problems are in software so don't require working hardware.
er.... dunno about that
OK, I don't need the Z-axis working to see if the veocity is osciallating
What does "run in place" mean in this context?
it means a non installed version running in a directory
which you need to not interfere with the running setup
Ah, OK. I am a little unclear what non-installed means, but the last bit was the important part.
the files of a installed version are spread in too many places for it to be nicely manageable when changing the code and compiling and testing often
although it can see your hal and ini
and there can practically be only one installed version of one program on one system
Right, makes sense.
because there's only one right place for it
heh except mysql servers celeron55_ , one can have fun there
archivist: there is no creep except for the loss of floating point precision as the joint gets far away from zero. but that is a very small effect now that we have double precision all through EMC and HAL.
cradek, Im wondering if one can store the 0 for the resets
<digression> I suspect all the Mac OSX apps are done that way, as they don't care where you put them, and are monolithic lumps of code directly corresponding to their icon. (But you can look inside, if you want). Installing a Mac application involves dragging the icon off the disc to where you want it.
archivist: not without very invasive changes. I do not think it will be a problem now that we have doubles through hal.
double is 16 significant digits. Enough for most purposes.
cradek, I know Im going to get down and dirty with hal for a hobbing machine due to the need for the gearing to remain in gear through an emergency stop and rerun
You will run out of precision with a double after about 10^12 revolutions of the gear. I suspect none of us will live that long.
(hmm, one rev of the gear per second, that is only 30,000 years. But you will probably have to reboot in that time. :-)
the gear arrangement of a hobbing machine http://www.collection.archivist.info/hobbing.jpeg
its the funny small numbers in the differential etc
I look forward to seeing a HAL implementation of that!
it is going to be fun!
archivist: I really hate when ppl see/use/consider "E-Stop" when it really should be a "PAUSE" button.
electronic gearing is sure more convenient than the 1948 style
jymm, e stop must NOT break any thing
jymm: on the other hand, it's bad to be scared to push it - when it's time, you don't want to think twice
the real machine has real gears
archivist: It's E M E R G E N C Y Stop, not the "oh shit I fscked up" or "I need to take a break" button.
type type type type type type C-u
^ me not getting into this discussion
cradek, by the way that picture was from an article using sound/vibration to detect errors in the cuts
estop should shut down everything to a SAFE condition. HALT should place things into a "holding pattern" and just maintain current position.
My dad worked for 48 years at David Brown's. They have some serious gearmaking kit there, they can do gear up to 14m diameter (and no, I don't mean 14 feet)
safe means held in gear
does the work advance an entire tooth each time? ( doesnt need to advance in an analog way ) can you add an electric brake on the turntable and open/close the control loop in between advances?
If hitting E-Stop means things the workpiece is fubared, but safey is maintained, so be it.
* archivist has a damaged David Bownd book I want a clean copy
IMO, estop means the different kinds of energy (electrical, kinetic) should be removed from the machine as fast as possible, to minimize or prevent damage to the humans nearby
dangit I typed
jymm, wrong that means a smashed part and bits flying
it's a terrible tradeoff
archivist: that's what brakes are for.
"try keeping position" means you have to leave power applied. leaving power applied means if the control circuitry freaked out and that's the reason for your estop, estop doesn't help
I guess I am not the only one to have run out of I/O pins before getting an e-stop rigged up?
how did making sure the hob was in the correct relation to the work go into estop discussion?
what cradek said
tom3p: everything devolves into an estop discussion or a comparison to the nazis
it's an internet rule
I sure sympathize with both sides of the estop debate
I thought that was vi v emacs the operating system
another problem is servos stop under power/control faster than they coast to a stop, but same problem applies
cradek: Well, that's why I said brakes, and just just remove power.
remove power and add brake (or just short the motors) is best I bet, but none of my machines have that provision
can emc2 'open the loop' on an axis?
dc inject for brake
maybe we all ought to decide what we want and then do that :-)
tom3p: sure when you turn 'machine off' you get that
uh, i mean, in middle of a program, then close the loop then open then close....
you'd have to fool it, but you could do that using hal
'unhooking' a servoed spindle and running it in velocity mode would be a use
tell the position loop to 'forget about it' ... oh thx, cool
Hal says "I am sorry Chris, but I can't let you do that"
yeh it must be possible ( i sit here fiddling with the THC code ... doh!)
are there any of the MESA Anything PCI cards that should be avoided?
Ask on the mailing list the Mesa guy hangs out there. I don't think I have seen him here.
L84Supper: i'm pretty sure they all work great, and if they dont i want to know about it