you are saying that the entities as output by the cad _don't_ match?
I mean, the cad is outputing 6 place numbers that don't match in the 3rd digit?
Yes basically so, this is the free version in synaptic , dunno if the "pro" retail version would be better. May be crippled intentionally?
I can't believe that
[00:01:40] <SkinnYPup> http://www.ribbonsoft.com/qcad.html
if you do "trim to intersection" (or the equivalent) and the resulting endpoints "as output by the cad" aren't the same. the cad is _wrong_
I've used qcad for a total of about 5 minutes, so unfortunately I can't help much beyond that
I use easycad - costs modest money, but it is very nice
Yeah it makes small errors that really add up when you combine the rotate copy function
to its trimmed entities
oh, you are trimming first, then rotating
how are you translating from DXF to g-code anyway?
Some is trim rotate
how are you translating from DXF to g-code?
gcnccam conversion to gcode, however it contains the same small errors that can be traced back to the original .dxf
is gnccam open source?
if you go back and analyze the entities center and endpoints its there in qcad
yeah opensource http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?GCncCam
I wonder if it could be hacked to fix the problems
gotta make more chips.... (gawd, cast iron is so nice - cuts like buttah!)
Now I can make rectangles with trims or radius to their corners that work fine, but throwing out a few construction lines to make a square tooth "gear" and radius the root and crest , then rotate , copy there will be problems
Enjoy the chips !
SkinnYPup: have you tried changing the precision values under edit->current drawing preferences
A few months ago when I was first having this prob with qcad , I cranked those precision values up to 6 decimal places. Then constructed a new"part" to have the same locating errors where it places arc centers, begining and end points , if you went back to the entity properties they would reflect this.
However it won't "zoom" deep enough to show on the screen that the parts were disconnected by .0006 for my particular problem. I used mastercam X demo to verify there was an error in what it was trimming
some small "teardrop" type features on a round disc would work out with numbers that emc wouldn't choke on , but after the third rotation the features entities began to shift in accumulating error.
got my vertical into production: http://www.youtube.com/watch?v=KT_cAiyks5U
more later next week.
This Qcad user sounds like he was having the same problem http://tech.groups.yahoo.com/group/qcad-user/message/1795
Canada is famous again..... dont take the greyhound bus.....
[01:30:30] <LawrenceG> http://www.cbc.ca/canada/saskatchewan/story/2008/07/31/greyhound-transcanada.html?ref=rss
That vertical lathe is awsome
hahahaha that IS awesome
That this cool !
that's a very long endmill
that thing yes ... can't even type after that
OK Time for my stupid question of the day (part 1?)
2 identical 12-0-12 CT transformers , 1st gets wired with a bridge rectifier, 2nd is wired with 2 diodes using the center tap. Can the outputs be connected in series before the the cap (liner power)?
SkullWorks_PGAB, Why would one want to do this. It does create different voltage between the two?
summing the voltage
Oh. You want to connect the negitive of one to the positive of the other?
Why not connect primaries and secondaries so you get the voltage you want before the rectifiers?
2x 24V in series gives about 68VDC after the cap - thats too much
true if there were suitable bucking trans that could let me hit the target Voltage without loosing Current capacity
other option is get a torrid and start unwinding... :(
So you want one of the transformers to produce 34 volts and the other 16?
What is the peak voltage of the device you're feeding?
target is 48VDC @ 10A - I was going to slip a diode inline for a slight drop.
Well one of the transformers will be working harder than the other. But...
greek to me - was told if I am hooking up in series the trannies had to be identical for load balance.
load balance may not be right term
Load balance only counts in parallel, I think.
anyway I thought that maybe setting one up old style using the center tap and 2 diodes would work
vrs a 4 diode bridge rectifier
skipping the center tap & using the bridge I get 34VDC at 10A - the other way I get 17VDC @ 10A
not sure how that works - kinda half wave rect?
You are rapidly reaching past the extent of my knowledge.
But the half wave on each half of the transformer seems right.
I'm out of my own depth at the moment
Just don't ground anything in the middle of it.
if anything including the driver is grounded at one end or the other.
SkullWorks_PGAB: you can do what you are describing
or you can save a couple of diodes, and just connect one half of the one transformer's winding in series with all of the other one, then rectify the result
either way, one of the transformers will be at half its VA rating when the other one is fully loaded
ah - was not sure if I could do that
paralleling is where you have to be carefull
if the transformers aren't perfectly matched and you parallel them, you can get circulating currents
if the mismatch is bad, you can overload the transformers before you even draw any load current
but in series, the currents are forced to be the same
so that would cause lots of inductive heating without any applied load?
in parallel, you can have one transformer (the one with higher voltage) delivering 10A, of which 9A is backfeeding the other transformer (since it has lower voltage) and only 1A is going to the load
in series, if the load is drawing 1A, then both transformers must each also be carrying 1A, no more
but I should be safe getting 2 units same mfg and p/n.
"getting" as in buying?
or do you already have two?
if you are buying, you don't want to do that
you will be wasting half of one transformer
either get one transformer of the proper voltage, or get two with the same _current_ rating, and voltage ratings that add up to whatever you need
bought - 2 was to be for 2 diff projects but I decide one would require too much space
heh, I can't tell what you are saying - do you have the transformers already or not?
have the allready
lol - I thought that is what I asked you a few lines ago
if you got 'em, use em
Do you really need the 48 volts or would a single transformer be sufficient?
even buying new - some units are less than half price because of the volume of sales over the less needed odd voltage units
but suppose you need 24+12
you can get 2 500VA 24s, use one center tapped to make 12, and put it in series with the other one for a total of 36
you can draw a maximum of 750VA from that combo, but you bought and paid for 1000VA
if you get a 500VA 24 and a 250VA 12 (or 250VA center tapped 24), you can still get 750VA total, and thats all you paid for
thats the plan - 2x $13 24v units vrs a single torrid @ $49
more cabinet space, less cash
yeah - toroids usually aren't the cheapest, but they sure are small for the same power
and this project is for a guy who is unemployed - medical issues - will need more surgeries on his foot and will have to stay off it for about 9 months
I'm donating what spares I have that will work for his unit
I mean, "is it a stepper machine", not "do you have the steppers already"
yeah - He has a CNC ready unit that needs Nema 34's
you are aiming for 48V DC at what current?
maybe a bit less
hmm, dog is whining, better take him for a walk
Thanks John and Ray - I'm off for an extended coffee break.
hello mr Joni
yet another hal question...
man motion is nice for documentation of pins the motion module presents to the hal-"user"
But some (ok at least one :-)) IN field is not described fully... The "axis.N.motor-pos-fb IN float" is described as "The actual position for this joint." but not how it is used in the code.
The same goes for axis.N.neg-lim-sw-in IN bit and axis.N.pos-lim-sw-in IN bit
Is this described somewhere, or should I simply look in the code?
pmbdk: motor-pos-fb is used for multiple purposes. here are some that come to mind: kinematics, determining following error, for the DRO when in "actual position" mode, for the result of G38.x probing moves, and to set the commanded position when going into machine on mode
pmbdk: lim-sw-in is used to trigger a fault condition when a limit switch is closed
(actually in machine off mode the commanded position tracks the feedback position at all times, and then that 'tracking' is turned off when you go into machine-on mode)
pmbdk: you might try to catch jmkasunich, he prepared a nice drawing showing the uses of commanded and feedback positions inside the motion controlller earlier this summer.
jepler: Thanks... I'm trying to find the use of the var in the code right now...
busy hacking up the documents :)
ah fix the arctan crap
years ago .. when sim first came out I installed it on a FC5 system. Now I can't seem to remember how to start it ..
where at archivist_ub ?
gcode docs theres a table with three arctans
* BigJohnT looking
tool file and compensation?
language overview user doc
what is the title of the table? The manuals have changed a lot since the last version
Table 13.3 Functions
it's 6.3 now but I'm there
it needs an explanation that you use the top one for all three
can you elaborate a bit please?
I tried using the other syntax but failed , it just gets error messages
yay -- now I have working X and Y home switches
archivist_ub: it's an error in the documentation -- there is not a one-arg ATAN function in gcode, only a two-arg one.
ok thanks jepler
hey jepler ... how do I start the sim version??
BigJohnT, I found it writing my helix milling code
I've fixed it
archivist_ub: anytime you find stuff like that feel free to let me know
the home position repeats to much better than .001 inch
I found mine to be more accurate than I gave the switch credit for
dave_1: the sim version of emc2 starts just the same as the rt version, but you have to choose a configuration that doesn't load any hardware drivers. the ones in the "sim" folder of sample configurations fall into this category, as do several others.
jepler, you using cheapo micro switches as well
I have my supplies, I must fit
I need to figure out how to sanely route these wires and hook them to my I/O board
dunno how to home the rotating axis's nicely yet
what about a inductive sensor
and a index pulse
or absolute encoders ;)
nowhere to fit on a Vertex HV4
hasnt even got sensible bolt holes to fit to table
ah.. yeah :/
I have one too
I need another for the 5th axis
but it has sane bolt holes :)
you drilled it?
I think a hv6
but it's got a bit of slop
this is not perfect, but good enough
jepler ... I've spaced ... typing emc only get me "emc: command not found"
for the price I think reasonable
no... older install ... right after sim was announced ...
I don't get the usual popup to select config
IIRC it took as multi word command line
maybe I should just reinstall with a more modern version
anonimasu, I add a jog at the start of the gcode to remove backlash, so I always push against if I can
dave_1, do you have any executables that begin with `emc` on your path?
like the old emc.run (or generic.run, or whatever.run ...
ah .. finally stumbled across it. $scripts/emc does it :-)
that gets me the popup
oh, is this a compiled "run-in-place" install?
if so, then you should source scripts/emc-environment
that will set up the path and other stuff so things like man work (in that shell only)
SWPadnos: if you have any plans to visit the Northwest, look me up. I'd like to show you our shop sometime.
sounds good to me!
your brother lives in Beaverton right?
I'll let you know - I have a brother in law in Portland and friends in Seattle
yeah - Beaverton (close enough :) )
hm, motor offset isn't available on a pin
joint.n.motor-cmd-pos vs. axis.n.motor-cmd-pos
at least according to the manpage for motion
SWPadnos: don't think that's really motor offset
it says it includes things like backlasn and screw comp
ok, might be
KimK is now known as KimK_
hi - are you the kim guy from cnc workshop?
Hi Chris. Yeah, that's me.
welcome. how's it going?
Still fooling around getting alternate nicks registered. Not bad, got a potential retrofit on a pair of chuckers. We'll see. Hi Alex, sorry I didn't get to meet you at the 2008 Workshop. Maybe next year.
KimK_: I wasn't there :)
maybe next year
Anyone have any opinions on early 1980's Cincinnati Milacrons?
either iron or electronics?
I have zero experience with them, but I'd say scrap the electronics unless they work perfectly ;)
Hi SWP. Yes, I agree, especially since even the one that *works* takes five rolls of #$%^&* paper tape to bring it back from the "dead".
tape == scrap IMO
Yes, it's certainly an EMC2 candidate. At least that's what I have in mind. But the operators panel has some peculiarities. For example, it has 4 "select jog direction" buttons (membrane keyboard w/ adjacent state LED) and only one actual jog button (industrial pushbutton). Stuff I'll be glad to be rid of.
you could probably even to that in EMC (not that you'd want to - but you can)
actually, the 4 buttons and state indicator are more in keeping with the "NIST principle" than a multi-position knob
sounds like an awful user interface...
if the GUI has an "active axis" output, then the LED show changes from the GUI as well as the buttons
mine has two buttons, X/Y/Z and JOG/WHEEL/INCREMENTAL so you have to stab at the two of them until the display shows what you want
SWPadnos: only if you accept that a "direction + go" button is a good way to do jogs
it sucks. forget doing anything without looking at it.
true, direction is pretty dumb, I was thinking it was the axis select
axis select + jogwheel would be nice
jogwheel axis select should be a hard switch IMO. you should be able to switch it to the axis you want without looking.
yeah - since each set of jogging controls is independent, you can have an axis selector+wheel at each one - keeps someone elsse from changing what the control in your hand will do
cradek: exactly - the "push one button N times until the indicator says the right thing" is lame - it is used only when buttons (or I/O points) are way too expensive to afford more
like alarm clocks
Yeah, the electronics are junk, and that can be fixed with EMC2. But my "research project" is this: The owner is reluctant to spend any money on them at all. He seems to think that nobody would want a Cincinnati-Milacron, preferring instead (of same approx vintage, price, etc.) Okuma, Hass, (pretty much name your competing brand). So: does C-M have a bad rep in the used market? And is it because of the electronics (which I ca
n agree with, and there's a remedy for) or is it because of the iron? (or something similar, maybe something not easily fixed?)
Oops, talked too long, I'm a flooder, LOL
unfortunately only a fairly small percentage of the people in here are familiar with production-class machinery
for example, I have a 66 year old Van Norman knee mill, and a 10 year old chinese 3-in-1
the 3-in-1 is CNC'ed with EMC, the mill is manual
I have a 25 year old bridgeport and 30 year old hardinge lathe
Hi John. Thanks for joining in.
oh, and I have a 18 year old "Targa" (taiwan) 6x12 manual surface grinder, that I am reconditioning
There is nothing wrong with Cinci. Good machines. I don't think I'd try to sell one that I'd fitted EMC2 to but that would be equally true of any retrofit not named Fanuc.
you might want to post your question on the emc-users mailing list - there are about 600 people there (vs. 47 online here)
Hi Ray, good to hear from you.
But for production it would work great. Hi Kim.
you might also want to post on CNCzone - they seem to have a higher proportion of shop owners, etc (at least in some channels)
rayh: Your point about the popularity of Fanuc is certainly valid, but the expected cost vs. resale price makes a Fanuc retrofit unlikely, and since these two have been sitting around for some time, I think it may be EMC2 or the scrapyard.
Hate to see them made into patio furniture.
You bet it does.
Fanuc costs less now than a while back but still not worth it for my thinking.
While I consider everyone's valuable advice... John, what reconditioning work does your grinder need?
rayh: curious, why would you not want to sell a retrofit machine?
fenn: the need to support it
(would be one of my reasons)
many/most big shops will look at the brand of control, then move on if it isn't on their short list of favorites
The machine would be fine. Selling it with a "noname" control would be the difficult part.
Yes, some (big) shops are that way, Fanuc only, Haas only, etc. But maybe small or medium shops would be OK with it. And occasional ongoing support is not necessarily a bad thing either.
Are there INI files (or whatever?) to make EMC2's behavior more Fanuc-like, more Haas-like, more whatever-like?
it's already pretty close to fanuc
no, if you're talking about the specific G-doce dialects
there has been some discussion about making custom G-codes using a macro language of some sort (Ken Lerman is thinking about it IIRC)
I had in mind just so that someone who was very used to a brand X control would feel comfortable, or at least not be booby-trapped by some difference.
yeah - it's not so configurable that you could make a Fanuc-like UI. at least not without writing some code
KimK_: we have a great manual :P
It would seem (I am no expert) that macro's or pre-processing the different dialect NC files could generate an independant version of NC files. ... that wouldn't get over user interface issues
I don't think the UI would be a big problem. I was more worried about some obscure, seldom-used G command working "backwards" or something (yikes!). But servant74's idea of pre-processing, might be just the ticket.
if a list of "trouble codes" can be made, it's pretty easy to look for them
Yeah, that's good. And the fact that EMC2 is "already pretty close to fanuc" helps also. (Not sure if we are both following the standard or if the standard was set because of Fanuc, but, whatever, it works for me.
KimK_ is now known as KimK
KimK: regarding reconditioning (I was working when you asked) - it has a lot more wear on the cross-slide ways than I knew when I bought it
it was only $210 tho, so I don't feel too bad
I'm planning to rescrape the ways - I did one flat surface already: http://jmkasunich.com/cgi-bin/blosxom/index.html
at the moment I'm working on an angle straight-edge to use for spotting the dovetails
dushantch_ is now known as dushantch
jmkasunich: Nice report, thanks. You've worked hard on that grinder, it should be nice when done. Keep us posted.
i need to come up with a banner for my webpage
to replace the default banner that comes with the framework =(
I was discussing Qcad yesterday, here's a little something I threw together real quick as an example of it goofin on the arc center placements .
[21:47:38] <SkinnYPup> http://imagebin.org/23498
[21:49:27] <SkinnYPup> http://imagebin.org/23499
The toolpath shown in axis is a 30,60,90 triangle it had no problem radiusing.
it looks like qcad is outputting 3 significant digits
can you change it to 4?
also, for the next screenshot, could you make the line with the error visible? :)
SkinnYPup: what did you use to convert from dxf to g code?
I think it goes as high as .0000000x but it just doesn't calculate the radius trims to that accuracy.
qcad lets you design and output GCode AFAIK
* BigJohnT looks
I've used both gcnccam and dxf2gcode to generate that. But double verified that Qcad is to fault using mastercam to compare trim radiused start end and centerpoints
SWPadnos: I think it's 'camexpert' from Ribbonsoft
oh - does qcad not output GCode?
bummer - nevermine me then :)
I see it says (Generated with dxf2code)
I use sheetcam and qcad with no problems
it could be qcam or dxf2code that needs to be fixed
well, I didn't observe that well :)
if they are open source this should be simple
SkinnYPup: can you post the dxf file that qcad generates on pastebin?
Sure ! hangon
also I get errors all the time from mastecam with arcs
no matter the source of the dxf file
DanielFalck: my wife says the bottom of the first one is too fat and don't look right for a butterfly
hobbyist & linux noobie here, need a little help compiling emc 2.2.5 on ARCHLinux kernel 2.6.25 patch with Xenomai 2.4.4
Here's the pastebin
[21:58:46] <SkinnYPup> http://pastebin.com/m15b7d39c
BigJohnT: I've never looked at them that close, so I wouldn't know
SkinnYPup: one moment
the dxf looks fine. it has full precision
Oh no hurry been bangin at this for a while
so I assume 'dxf2code' is nuking it
opens fine in sheetcam ...
compare these to the endpoints in the error message
you can see that dxf2code truncated them at 3 places after the decimal
you should be able to just change it to 4 or 5
SkinnYPup: works ok with sheetcam
[22:04:19] <BigJohnT> http://pastebin.ca/1091262
you will have to comment out the o<touchoff> call line
SkinnYPup: which program did you use again for converting to gcode? Where did you get it from?
Jamie is now known as Guest26797
* BigJohnT heads upstairs for butterfly inspection :)
Ok had to try out a few things there. BigJohn, yes that is a good toolpath. You generated that with sheetcam ?
I'll have to try it thankyou
it's free at the moment and has a linux distro
DanielFalck: I used dxf2gcode
SkinnYPup: was it from here:http://dxftogcode.googlepages.com/dxf2gcodedxftogcode
If the post gets you confused I can send you one
of was it a simpler c program from http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Cam
Its not c , .py
[22:15:46] <SkinnYPup> http://translate.google.com/translate?hl=en&langpair=de|en&u=http://www.christian-kohloeffel.homepage.t-online.de/dxf2gocde.html
* BigJohnT passed butterfly inspection
DanielFalck: Didn't know there was more than one dxf2gcode !
there's a bunch with that name
SkinnYPup: in the dxf2gcode postproc file, set post_decimals to 4 instead of 3
cradek: I'll check that. Thanks
it's the first setting in the file :-)
mhel: your question got lost in the traffic I think
if it has to do with compiling, etc, you might be better off asking on #emc-devel
thnks, I'm just waiting for possible reply :)
actually, you didn't ask a question ;-)
I was noticing that too
you just saId you were having a problem
I did hope that someone would ask me what tho.. he.he
i'm gonna ask at devel then.
cradek: The postproc file with post_decimals you refered to . Is it dxf2gcode A or B ? http://dxftogcode.googlepages.com/dxf2gcodedxftogcode
[22:31:44] <SkinnYPup> http://translate.google.com/translate?hl=en&langpair=de|en&u=http://www.christian-kohloeffel.homepage.t-online.de/dxf2gocde.html
I don't have any M.S. windows running to try http://dxftogcode.googlepages.com/dxf2gcodedxftogcode
at the moment.
don't worry about it, I bet a lot of folks here don't either : )
use the source Luke....
shame theres no real decent cad/cam under linux
pro/e is ok for cad but cam is a bit lacking
there's is it's called Unigraphics NX
it costs a lot
there's also sheetcam and webersys now
dont think nx is all that good for cam eitehr
my place used nx5 for cam but binned it off for hypermill
colinb ... I think synergy (webersys) will do just about anythink you want including 4th and 5 axis but the learning curve is really steep.
at least is is steep for me but maybe that is an age thing. I don't do python either.
not heard of webersys tbh
opps ... anythink == anything
learning delcam powermill at the moment for my new job
thats quite a complex program
Weber Systems wrote and sells synergy
2.5 D thru 3D 5th axis
$250 gets you 2.5D cam and then the price goes up from there. .... if you are interested email webersys.com ... check the website.
ah yeah just had a look
iv seen it before
I use it for a lot of 2.5 D but have not been successful in the 3D mode