Print/cut alignment

[Peter Boyford]

I have a GCC Mercury laser and just stumbled into a small problem, that can turn out to be vast if not solved:

I have printed some small signs on polystyrene and need to contour cut it on my laser. I have mounted the vector table very sturdily on the bed of the machine so the material will be fixed in the same position every time i cut something.

Hence I know exactly where my X0 and Y0 is on my table. BUT:

1. I cant seem to find anywhere in the driver that lets me specify an absolute offset
2. when I cut along both axis the machine inaccuracy is actually up to 3mm over 600mm. This means that when I send a rectangle of 600x400mm it turns out to be 597x401mm.

There seems to be some sort of correction option in the driver that lets me set scaling factors for each axis in steps of 0,1%. But simple math tells me that I need to scale X by 0,05% and Y by -0,025%

Any takers to these problems? How do you align laserparts to precise cutting?

[Dave Johnson29]

There seems to be some sort of correction option in the driver that lets me set scaling factors for each axis in steps of 0,1%. But simple math tells me that I need to scale X by 0,05% and Y by -0,025%


Hi Peter,

There was a posting about this I read just recently. I think it was by Richard Rumancik.

OK, found it, read here...

http://www.sawmillcreek.org/showpost.php?p=995460&postcount=30

[Bob Davis]

I also have a GCC Mercury and have battled to calibrate it accurately. There is a setting in the driver that can be used to adjust the scale in .001% increments. This can get a fairly close accuracy with a lot of trial and error, but I think there are other factors involved that effect the result.
On my (well out of warranty) machine if I send a raster job that extends over a wide area and then send a vector job to cut around the images I will usually (but not always) find that the two jobs are not accurately registered to each other on the x axis despite the same scale being set in the driver. However, if both jobs are sent at the same time it will usually (but again, not always!) keep correct registration - at least with each other if not in absolute terms.
This lack of consistency is a real pain. I've swapped out the RAM without improvement, but generally the machine is doing smaller jobs where the inconsistency is not appreciable to the eye, so I've just lived with it and used our Epilog laser where accuracy is required.
An aspect that appears to be a factor is the number of objects horizontally in the job. An example would be cutting 6of 100x30mm plastic plates, which the machine does faultlessly, and then stacking them across the top rail. In Coreldraw design a block of text to fill the first 100x30 and then duplicate it at 100mm distance another 5 times and send the job. Despite the 6 x 100mm plates extending to (very close) to 600mm, the raster job results in the text creeping progressively left on each plate until the last is out by around 3mm. This amount is very noticeable on a centre justified text block.
If anybody has any clues on what could be the problem I would be pleased to hear from them.

Bob Davis

[Albert Nix]

Peter what software are you using with the laser and what software did you use to print the signs. What size are the signs and how are they laid out. Can you post a picture of what you what cut????

[Rodne Gold]

There is a problem with the sealing of the X-motors shaft encoder housing on the Mercurys , dust gets in and the accuracy is compromised, take the motor out , slide off the housing and dust off the encoder and the reader led thingy , then slide the housing on and wrap and seal any crevices or gaps with masking tape.
The mercury knows where the limits of its home is cos of little tabs that interrupt a beam , if these are faulty or loose , the home will differ slightly each time it homes.
The x/Y scaling will allow you to correct any of these incorrect X and Y problems
You can easily put in an offset if you want by setting the driver NOT to start from home but from "relative" and then use the buttons on the machines display panel to manually jog it to that position (displays coordinates of x andf y in the display pane) or put the offset in in the drawing.

One problem you might find is this - that the table edges are NOT square with the laser travel , so jamming a square of cut materal into the corner to register skews it with respect to the cutting. You can make a "set square" with the skew factors cut and then put that in the corner and register material to that set square.
Prior to my getting a large format digital print and cut printer , we used to print the most intricate stuff on a deskjet printer and die cut with the laser , i could do big qtys of small and fiddly die cut stuff VERY accurately, leaving even cut borders around graphics or cutiing a printed die line.
So registration with the mercury to extremely high tolerance is attainable.
Another area to look at is if you want to print and then cut , make sure its not the printers X/Y thats not out...
We had made a jig to hold the printed item and used to use registration marks to test the accuracy of cuting , print one top left and bottom right and send the file to the laser ands run the crop marks.

[Peter Boyford]

Thanks for your replies so far.

I have, of course, measured the print before agonizing about misalignment of my laser. The print is spot on.

Bob-> where is that option to adjust i 0,001% increments? I can only find one to adjust in 1/1000 increments (0,1%)...

The method using Corel Draw to compensate for inaccuracy is actually one I have tried with success, but it seems insane to me that the laser doesn't have this feature built in.

I have compensated for skew - with success (and trial and error).

[Bill Jermyn]

Bob, with regard to your problem, what I have done on my Epilog is to reduce the page size to the individual item size, then use layout imposition for the number of items across and down, and adjusting the page size to get the spacing just right. In Corel you can adjust the page size in .001" increments (or maybe less, I'm not sure).

[Rob Bosworth]

You are also compounding the accuracy problem by using CorelDRAW. CorelDRAW is dimensionally unstable. It is a pixelated graphics program that just is all over the board on dimensional stability.

So let's see. We have a machine that runs by a rubber bands (drive belt). We have a movement system that is run on rubber tires. We have encoders that could or could not be dusty. All of these movement components are running on extruded aluminum rails. We use a laser beam that hits two, three, four optics that are mounted on fairly light weight fixtures. Most of these machines sit on casters with very little machine leveling ever done. We use a program that is dimensionally unstable. Now you are trying to hold really close tolerances on a printed piece of paper?

As Scotty used to say, "We can't defy the laws of physics?"

[Richard Rumancik]

You are also compounding the accuracy problem by using CorelDRAW. CorelDRAW is dimensionally unstable. It is a pixelated graphics program that just is all over the board on dimensional stability.

Rob - you said this once before in another thread - someone challenged it and you did not reply (if I recall correctly). What do you mean when you say CorelDraw is pixelated? It is a vector program, not a bitmap program (like PhotoPaint). Yes, of course, everything ends up as dots if you print to a laser or inkjet or display on a screen. But the entities in the file are mathematical features, not bitmaps. You can scale up and down without losing any accuracy. I have never found Corel to be "inaccurate" from a mathematical sense. Although it is not a CAD program, and does not have all the normal CAD tools, I have not found anything to suggest it is inaccurate on a mathematical basis. Can you elaborate?

Your comments about other sources of error are well taken. However, with proper compensation, tooling, and procedures, you can still do quite accurate work. Even though belts may be made of rubber, they are surprisingly repeatable. Like Rodne, I have had to develop my own methods because my supplier did not appreciate the accuracy I was trying to achieve. If you are marking a plaque extreme accuracy is not that important. But if you are making production parts for manufacturers you need to use different methods.

[Rob Bosworth]

Richard, I am not sure why dimensional cut parts from CorelDRAW tend to be less accurate than parts run directly from a CAD file. I am assuming it has to do with a vector contoured line in CorelDRAW is actually a series of dots. A line in a CAD package comes out as true arc. Could it have something to do with the CAD package using three points to plot an arc and CorelDRAW is basically connecting the dots?

I have generated same vector layouts in AutoCAD and CorelDRAW, only to get more accurate results from the AutoCAD generated file than from the CorelDRAW generated file. Repeatability is not effected, but accuracy seems to be. I assume that repeatability is a function of the machine and accuracy has to do with both the machine and the artwork generator.

[Doug Griffith]

Richard, I am not sure why dimensional cut parts from CorelDRAW tend to be less accurate than parts run directly from a CAD file. I am assuming it has to do with a vector contoured line in CorelDRAW is actually a series of dots. A line in a CAD package comes out as true arc. Could it have something to do with the CAD package using three points to plot an arc and CorelDRAW is basically connecting the dots?

I have generated same vector layouts in AutoCAD and CorelDRAW, only to get more accurate results from the AutoCAD generated file than from the CorelDRAW generated file. Repeatability is not effected, but accuracy seems to be. I assume that repeatability is a function of the machine and accuracy has to do with both the machine and the artwork generator.

Files run on my Epilog using Adobe Illustrator are fairly accurate and repeatable. Close to .005". About as good as its going to get using a belt drive system with a tapering beam that incinerates plastic. Could it be that the driver you are using is converting Corel's bezier curves to segments while leaving your CAD arcs as true arcs? Not the machine or the art program but the driver.

Cheers,
Doug

[Dan Hintz]

Rob,

As Doug just alluded to, there's still quite a bit of "going's on" between the vector file in CorelDraw and the output of the laser. The internal representation of an arc/bezier curve is, from a mathematical model standpoint, identical between CorelDraw and any other CAD package. At some point, both packages have to turn that mathematics into a drive command for the laser, and it's in that path that something is going awry for you. I suspect it's the link between the laser driver and how CorelDraw turns that math into a bitmap, though without experimentation I cannot say with any certainty.

[Richard Rumancik]

When I cut small circles and arcs they tend to be a bit more choppy than I would like, but I have never attributed that to CorelDraw. The mathematical curves have to be interpreted by the driver/firmware/hardware and converted into machine motion. This is where problems can occur. When you approximate a circle with a series of straight lines you will get some roughness to the contour; how much depends on how smart the motion control system is. I don't think there is much "look ahead" (if any) in laser systems like you see in a CNC system.

However, if you are using the same driver and hardware and see different results between two vector packages, then I cannot explain why. Corel won't store the shape of a curve as a series of nodes (like dot-to-dot). It stores a mathematical equation which defines the shape.

[Richard Rumancik]

where is that option to adjust i 0,001% increments? I can only find one to adjust in 1/1000 increments (0,1%)...

The method using Corel Draw to compensate for inaccuracy is actually one I have tried with success, but it seems insane to me that the laser doesn't have this feature built in. . .

Peter, I'm not in front of my laser, but what you see sounds like what I have. The driver allows you to select enlarging or shrinking in 0.1% increments. However, this is not fine enough to be useful. On a 25" length, an adjustment of 0.1% changes the length by .025". What if I want a stretch of .010" over 25"? Then it doesn't work.

In addition, I found that with my system, it rounded off the % increment that I set. It would only make a measurable change for every second setting. e.g. enlarging .1% or .2% caused the same effect (of a .2% change). You will get the same effect using .3% or .4% (GCC verified my observation.) So at that point I decided the scaling feature in the driver was rather useless and I have not attempted to use it for some time. With the driver, you can scale up or down .2%, .4%, .6% etc. The machine error on the worst machine probably would not be greater than .2%, so anything more than about .2% isn't even relevant.

The method of compensation by stretching the drawing in CorelDraw is a time consuming work-around, but it does work. The biggest problem I have is that you have to have a design file (which stays to scale) and a plot file (used for lasering only.) If you decide to make any changes while testing the plot file, you need to make them in the drawing file and re-generate the plot file. If you don't, you run the risk of fixing a problem and then having it "reappear" at a later date.

I agree that the laser should have a way of entering a compensation factor that is built-in. (In my opinion, it should really be input into the laser firmware at the control panel.) Once it is determined, it is likely stable for a long time. I use the same amount of compensation all the time (and I do check it ocassionally.)

It would seem to me that the correction factors should be initially factory set, with provision for tweaking in the field if needed. A brand new laser should plot a full-bed rectangle spot-on for size without any tweaking (at least to +/-.005" accuracy per 12"). The manufacturers should have the equipment to accurately measure the actual travel in each axes. But as these lasers were not really designed for accurate work, I don't think the manufacturers have paid much attention to that issue. Some of us are trying to use these systems for purposes beyond what may have been intended. With a few work arounds, it is doable, but it is painful knowing it could have been so much simpler.