Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Things on February 19, 2011, 11:11:28 PM
-
So I recently completed my home made laser cutter.
Ive been generating gcode in various software, however sometimes the software will output the Y coordinates in the - values, and sometimes it won't. My table at the moment is setup to work in + coordinates, however why is there all this variation? Is fixing the -'s as simple as just removing all the - symbols from the gcode? Which brings me onto my next question.
Sometimes the output is upside down, mirrored, backfliped, frontflipped and whatever other orientation you can think of. Usually I can fix the mirroring by switching the axis around (Making X Y etc), though that is tedious, and can lead to my table smashing into the limit switches if I forget to switch it back when I do something else.
I think what I'm not understanding is how the coordinate system on a XY table actually works, and how software produces the g-code with coordinates. For example, if I draw a 10cm diameter circle in Adobe Illustrator, export it as a DXF, import it into Lazycam, then into gcode and into mach3, will the circle at the end actually be 10cm diameter? Any help is appreciated :)
Also, how do you calculate how many steps per mm the machine is? I am using 200step/rev steppers, connected to These pulleys: http://ausxmods.com.au/t5-pulleys/15t5-27-6.35mm-bore-timing-belt-pulley
With this belt: http://ausxmods.com.au/t5-timing-belts/t5-16mm-wide-open-timing-belt
Running at 1/8 microstepping.
Cheers,
Dan
-
For example, if I draw a 10cm diameter circle in Adobe Illustrator, export it as a DXF, import it into Lazycam, then into gcode and into mach3, will the circle at the end actually be 10cm diameter?
Provided that Illustrator outputs the .dxf properly. From what I've read, Illustrator's .dxf export is not very good at all.
One thing to note. You need to use the same unit of measure throughout the process. From design to machine.
.dxf files typically won't be drawn with any specific unit. What I mean, is that your 10cm circle will really be 10 units. So, If you draw a 10cm circle, and I draw a 10 inch circle, and someone else draws a 10mm circle, all 3 .dxf files could be exactly the same.
You machine will most likely be set up in mm's, so, typically, you would draw the circle with a 100mm diameter. Some programs that export .dxf files have a tendency to assume your drawing in a specific unit, and will output in another unit, by actually scaling the drawing when exporting. You'll need to do some testing to make sure the .dxf's you export are the size you expect them to be.
Since my machine is set up in inches, for me to get a 10cm circle, I'd draw it with a diameter of 3.937 inches.
Steps/mm can be easily calculated. Tell us how it's driven, and the specifics of the components.
Or, use the calibrate function in Mach3. I think it's on the settings page.
-
Hi, thanks for the reply.
Right after the modify button expired, I found the set steps per feature on the settings page. Seems my guess of 20 steps per mm was accurate.
I'll try a few softwares and see what results I get.
Do you happen to have an answer for my +/- coordinate problem?
Cheers,
Dan
-
OK well, I drew up a 10x10mm box in google sketchup, exported it as a DXF using a free plugin, loaded it into LazyCam, then into Mach 3, and I got a 10x10mm square :D
One more question I have, what determines where the engraving/cutting actually starts? All the g code I have seems to have an offset from home it starts at, and I have to test it a few times on scrap before I know its in the right position.
Cheers,
Dan
-
When you have a .dxf file, there is an origin, where X and Y = 0. Depending on where the part is drawn relative to the origin, you may have positive or negative values, or both.
Some CAM software will let you move the origin to wherever you want it. Some use it exactly as drawn.
Removing the -'s is not the answer.
As for the mirroring, backflipping and other stuff, you'll need to provide specific info for me to tell you why.
-
Just like the origin in the .dxf, the machine has a 0,0 position as well. This may or may not be the same as home.
You can set the machine's 0,0 position anywhere you want.
-
OK, that makes sense :)
Just one more question, which is probably more mechanical related.
I drew this up in Sketchup:
(http://dl.dropbox.com/u/203420/shg.jpg)
Then ran it on the laser, and ended up with:
(http://dl.dropbox.com/u/203420/shl.jpg)
I have circled 2 points, where I'm not quite sure what happened. It seems almost like the machine has lost track of where it was and kept going, before realizing it was off track.
I don't think it's my machine skipping steps, because other stuff drawn after those points lined up OK, so seems more like a glitch.
EDIT: I just ran the exact same program again, and it glitched in exactly the same spots. To me, that doesn't seem like a machine problem (skipped steps etc)
Cheers
Dan
-
Dan,
The glitch top left is usually caused when the object vectors are not joined (LazyCam takes the shortest route to the next vector).
What did the whole thing look like on the LazyCam screen ?
Did it show these errors ?
Tweakie.
-
Hi, this is how it looks in LazyCam straight from the DXF (No optimizations)
(http://dl.dropbox.com/u/203420/shlc.png)
Cheers,
Dan
-
I tried rotating the image 90 degrees, and that causes the triangle at the bottom to be on the left side of the vertical line and not partly thru it, though I still have the problem with the little box and the diagonal lines not meeting up.
-
Dan,
how the coordinate system on a XY table actually works
Depends on how you have your machine setup. You can set your 0,0,0 anywhere you want.
Have a read of the Mach Mill manual.
software produces the g-code with coordinates
In general the software will look at the dxf, it contains the location of all the entities which you have
drawn and extracts that data on import. Best example is to save a file as a ver 12 DXF and look at the
file with Notepad since it's a text file. You can't do that with later dxf versions and they will contain additional
info on what was contained in dxf ie; the third axis locations etc. In general you now must define what you
want to do with each entity that was drawn, no software is a "mindreader".
will the circle at the end actually be 10cm diameter
Well it depends if the program used outputs the file as vectors based on precise data or as vectors
based on screen pixels/units, whatever. You want to use a vector based CAD drawing program and
will remark that not all programs generate dxf's exactly the same and some are just outright terrible at it.
[qoute]how do you calculate how many steps per mm the machine is[/quote]
Try the calculator in Members Docs or use the axis calibrator in Mach to check.
http://www.machsupport.com/forum/index.php/topic,16315.msg110286.html#msg110286
Can you post your DXF file of what you have drawn?
RICH
-
Hi Rich, thanks for the info.
I tried the text feature in Lazycam, and that comes out flawlessly, so I've come to the decision something is going wrong between the DXF and gcode conversion.
Here is the DXF: http://dl.dropbox.com/u/203420/moreshap.dxf
Cheers,
Dan
-
Just thought of something. The DXF export plugin for Sketchup has multiple export options. They are polyface mesh, polylines, triangular mesh, lines, and stl.
Are any of these better than others? I've been using polylines.
-
Post a screenshot from Mach3. That will tell you if the machine is missing steps. I'm pretty sure that's what you're seeing.
You can also post the g-code and I'll verify if it's correct.
Technically, there's nothing wrong with the .dxf file. However, it's not perfect, at least for CNC use.
One all the closed shapes are 3D polylines. Not all CAM programs will support those entities. Note that they are not draw in 3D. They're just drawn with a linetype that's capable of being drawn in 3D.
Two, all the shapes are closed, which means the edges between closed shapes (just about all) get cut twice.
I'm not a big fan of using Sketchup as a 2D CAD program, but I know some people do and it works fine for them.
Here's a shot of the g-code I created from your .dxf in Mach3.
-
Hi, would it be possible for you to post the gcode you created? That way I can try it on my machine and see if it's just the code, or my actual machine. I'm having a hard time thinking it would be my machine, as I can create some text in lazycam and it'll come out perfectly on the machine, jut seems that anything I bring in from a DXF doesn't work properly.
I did this using all the exact same settings:
(http://dl.dropbox.com/u/203420/DSCF2210.jpg)
What makes me thing it's not skipping steps or slipping the belt, is it can draw a misaligned line, but then it can go right to the corner of the square and draw the diagonal line. If it was skipping steps, the diagonal line wouldn't line up with the corner perfectly.
Cheers,
Dan
-
Dan,
Have a look at the attached pic as it shows what is being exported based on your drawing. It really leaves something to be deisred and is no wonder that LC or some other porgrams will be confused. I strongly suggest you have a read of the LC Manual.
RICH
-
Dan,
I used your dxf and LC to generate gcode. I converted the gcode and backplotted it back to CAD. Red lines are rapids and blue is the pathing.
Nothing wrong with the gcode as you can see from the pic. ;)
Should machine just fine.
RICH
-
Here you go.
You might have a mechanical issue, like a loose coupling somewhere.
-
Hmm, OK.
This is why I am doubting it's a mechanical problem.
This is the image drawn normally:
(http://dl.dropbox.com/u/203420/LaserCNC/shapes1.jpg)
And this is the image flipped 90 degrees in lazycam:
(http://dl.dropbox.com/u/203420/LaserCNC/shapes2.jpg)
How, notice how the outer square is perfect? Parts of that outer square are drawn after the glitches.
You'll also notice that the diagonal lines meet up in the corners of the outer square perfectly. 3 of these 4 diagonal lines are drawn AFTER the glitch in the corner of the box, yet they are still where they should be.
My theory is, if the steppers were skipping steps, or the belt was slipping, wouldn't the outer square also be effected, and the diagonal lines wouldn't match up either?
Does anyone have some different software they know works with their machine, that they could do up a simple DXF like this for me? Just so I can eliminate some things?
Cheers,
Dan
-
The outer square is NOT perfect. My eyes are going bad, but not that bad. :)
Reduce your X and Y acceleration by half, and see if it's better.
-
Yeah, I do see the tiny little offsets on it in the images.
I just reduced the acceleration from 500 all the way down to 50, and it didn't make any difference.
Cheers,
Dan
-
Any chance the pen is moving?
-
I think both I and Gerry have shown you that it is not a code problem.
I don't need to try a different program or run the program as the backplot overlays exactly onto the original file.
I believe you have a problem with you machine. It may be your plotting device, axis, setup etc.....so the quicker you accept that fact
the sooner you will define and solve the problem.
RICH
-
Well, it's actually a laser, as its only temporarily setup, it has the potential to move. I did try setting home on the machine to a different position, just to make sure it wasnt a certain section of the machine, but got the same thing. I will secure the laser down better tomorrow and give it another try.
I never said it could have been bad axis setup or an option somewhere in mach3, but there are no pointers that my steppers are skipping or belt is slipping.
-
but there are no pointers that my steppers are skipping or belt is slipping
Probably 9 out of 10 (or more) people that have these types of issues always find out in the end that it's their machine. And most always want to point to the software first.
-
Well, how would I go about checking if they are skipping steps or the belt is slipping? I've tightened the belts enough that they have minimal sag, maybe it needs to be even tighter? My steppers have a fair bit of torque, it takes a fair bit of effort to stop them, so i'd be surprised if they were the problem.
-
Make note of your current steps per unit. Go to the Settings Tab and use the axis calibration in the lower left corner of the page. Do a command via MDI line to move each axis one way and then the reverse. That will check your steps per unit.
You can also try different velocity and accelration settings, start low and just double the value until the calibrated steps per unit do not agree anymore with DRO displayed.That will give you an indication of when your starting to skip. That's why Gerry suggested you cut your velocity in half.
Let us know how you make out.
BTW, You can loose steps due to belt tension and you would not know it, but the amount would be small, don't want to dwell on the belt.
RICH
-
I wasn't implying that the motors are losing steps. My guess is that the laser is moving. I was just telling you why we tend to look there first.
-
Ive been generating gcode in various software, however sometimes the software will output the Y coordinates in the - values, and sometimes it won't.
I didn't see that anyone addressed this so I'll take a shot. If you're laser homes to the left, for instance, and the table is set up so the a G0 Y1 moves the laser 1 unit to the right, then G0 Y0 would move it to the left one unit. This assumes you're O,O origin is at the lower left of the drawing. If origin is in the middle of the drawing, for example, and you start at the lower left of the drawing, left hand moves that go past the 0,0 origin will be negative numbers.
-
Yeah, I understand why it could go into the -s now. Originally I didn't have a solid software to test with, so my gcode was all over the place. Now that I know sketchup works OK (Unless the machine isn't the problem), it makes sense that the machine draws it in the exact same spot as you did in sketchup.
I'll mount the laser on it properly when I get home and try it again.
-
Don't rely too heavily on Sketchup it was never designed for CNC and may not be doing the best job.
Tweakie.
-
OK well. I mounted the laser on the machine nice and steadily, it doesn't move at all now. I also ran the steps per calibration a few times on both axes, and they matched up exactly the same every time.
So not really sure what else could be wrong with it now, apart from software.
Cheers,
Dan
-
So not really sure what else could be wrong with it now, apart from software.
Dan,
Which software do you think is causing the problem ?
Tweakie.
-
Well that I'm not exactly sure on. The page before, the g code I was using was proven to be OK, and I can create some text in lazycam and it'll draw it all fine. Maybe mach3 is interpreting the code wrong? Or maybe there is a setting (offset or something) that mach3 doesn't like?
Does anyone have any proper CAD software they know works with their machine, who could create a dxf or some gcode for me to try?
Cheers,
Dan
-
Can you post the dxf you used from Sketchup?
RICH
-
This logo has just been downloaded from "Brands of the World".
Loaded in Lazycam and toolpath produced.
GCode and dxf attached - see what you can do with it.
(this has been set up for my vinyl cutter so you will need to edit the Z in the GCode and scale etc. to suit your machine)
Tweakie.
-
Here is how it turned out:
(http://dl.dropbox.com/u/203420/DSCF2213.jpg)
So, it would appear the problem is either in my electronics, or my mechanics.
Does anyone know any guides that could help me narrow down the problem? Or any gcode/tests I can run to help?
Is it possible microstepping could be playing a part in this? I have the controller (Pre-made, generic 4 axis board) set on 1/8 microstepping. I really do like it as it makes everything a lot quieter and smoother, not to mention more accurate, but if its causing me problems I may need to find another solution. But I'm was under the impression almost everyone used mcirostepping without problems?
I built this whole machine myself, and this teeny little glitch is all thats stopping me from finally completing it.
If hi res, detail pictures of the machine will help, I'm ready with the camera.
Cheers,
Dan
-
Apart from the obvious it came out very well.
On the 'R' you have a vertical shift and on the 'A' a horizontal shift so the problem is occuring on both axis.
Surely that gap is too large for missing steps (how many steps would have to be missed, consecutively, for a gap that large?) so something loose on your machine ? or your PC multitasking ? or drive belt jumping a tooth ?. I think you need to be extremely methodical in testing this to discover the cause, just try one thing at a time until you find it.
Best of luck.
Tweakie.
-
Do you still have the steppers being used as idlers? Could that be the problem as they may partially lock for a split second every now and then.
Hood
-
Dont know if it's any help to you but attached is the GCode file (thanks to Sam) I used for initially testing my laser. (it is Inch units so you may need to scale it)
http://www.cooperman.talktalk.net/laser1a.wmv
Tweakie.
-
Yes, I still have the steppers as idlers. I'll see if I can remove some of the magnetic structure so they spin freely and see if it makes any difference.
If those gaps were skipped steps, it'd take about 20 skipped steps to make a gap that size.
Multitasking PC: I think that'd be a more random occurrence, and I dont think it'd happen at exactly the same time, every time.
Belt jumping a tooth, possible, though not exactly sure how I can test that. Are my belts better off tight enough that they don't sag much, but not stupid tight, or stupid tight?
Cheers,
Dan
-
OK, I think I've found my problem. I set my original shape pattern on F60, and watched the belts move over the pulleys. I notices that sometimes the teeth of the belt actually ride ON the ridges in the pulley, and not the grooves like they are supposed to.
Now, why it's doing this has lost me. The pulleys and belt are the same pitch, so I'm thinking maybe my belts arent tight enough?
EDIT: I just took a pulley off, and wrapped the belt around it. I could get 1 full rev, but after that, the teeth started riding on the ridges. Maybe the machine was very slightly off when these pulleys were machined?
Cheers,
Dan
-
It's more likely that the belts are innacurate. (almost the same pitch is different to exactly the same pitch)
Tweakie.
-
Here is a pic of the belt , the left side, the teeth are riding in the grooves as they should, but as you can see, they start riding up onto the teeth in the pulley.
If I pull on the belt VERY tightly, I think it manages to fit, but that kind of tension would rip my machine to shreds.
(http://dl.dropbox.com/u/203420/bpulley.jpg)
I'll email the place I got the belts and pulleys from and see what they say.
Cheers,
Dan
-
It seems as if I may just need more belt tension. However, it's going to require a fair bit of cross bracing on the machine. Gotta do what you gotta do I guess!
-
DraftSight is a nice, free CAD program. It is offered by the company that puts out SolidWorks.
http://www.3ds.com/products/draftsight/download-draftsight/?xtor=SEC-6-GOO-[]-[]-S-[draftsight]
-
OK, I tightened my belts this afternoon about as tight as I'd want to take them on my current setup, and the results are MUCH better:
(http://dl.dropbox.com/u/203420/shl4.jpg)
Does still seem to be a little variation there though, which I'm not sure on.
Cheers,
Dan
-
The results are not really any better, OK the mis-allignment may be less but overall it is still a poor result.
IMO the reduction in the amount of error does tend to indicate that it is the belts/pulleys which are the main cause of the problem though.
Tweakie.
-
Its not going to be perfect, as the laser can not turn on and off (thus the dark spots in places its been over more than once, or stopped at.
What is odd is that square at the top is pretty much perfect, however now the quadrant on the lower right is off.
Not really sure what else I can test now? the belt does not appear to ride on the teeth anymore, however sometimes it clunks, and the gap seems just under the width of a single tooth.
Cheers,
Dan
-
I've tightened the belts even more, and still have the problem. I am pretty sure my belts are not riding on the teeh at all now.
I think the problem may be the teeeny offset on the pulleys, meaning they very slightly loosen and tension the belt. Seem like it could be the issue?
Maybe I'll see about taking a stepper and a pulley to my local alu place, and see if they can make me 6 sleeves that'll keep them perfectly centered.
Cheers,
Dan
-
Sorry for the triple post, but I have good news :D
I was just mucking around with the belt and pulleys on the X axis, and for some reason, it's decided to work properly now:
(http://dl.dropbox.com/u/203420/shlyay.jpg)
(http://dl.dropbox.com/u/203420/shlyay2.jpg)
No idea what I did that fixed it, but as long as it stays fixed, I'm happy :)
Cheers,
Dan
-
That's great news Dan, looks like you are well on your way now.
Tweakie.
-
Yep, very happy with it now :)
Just ran this up to make sure my steps per setting was right:
(http://dl.dropbox.com/u/203420/DSCF2225.jpg)
Cheers,
Dan
-
I have used Draftsight to create DXF files but on occasion I like to use plotter files. I cannot create the plot files directly in this program so I was trying to find a plotter driver that I could print to file. The ones that were recommended to me were the drivers for HP 7475a or HP 7580a. The problem is the HP download site does not have these drivers anymore and I cannot seem to locate them. In addition, I read somewhere that they are not supported on Win XP. So, question is what plotter driver is available and where can I download it ?
Thanks
-
I have used Draftsight to create DXF files but on occasion I like to use plotter files. I cannot create the plot files directly in this program so I was trying to find a plotter driver that I could print to file. The ones that were recommended to me were the drivers for HP 7475a or HP 7580a. The problem is the HP download site does not have these drivers anymore and I cannot seem to locate them. In addition, I read somewhere that they are not supported on Win XP. So, question is what plotter driver is available and where can I download it ?
Thanks
Perhap WinLine may be suitable for your needs http://www.winline.com/
Tweakie.
-
thanks, good suggestion