Cutting Software vs Printer Driver

[Andrea Weissenseel]

Just wondering if there is a difference in the quality of cuts, especially when cutting acrylic, if you use a laser cutting software instead of ie. Corel in combination with the printer driver for the laser. Any experience ? I've been told so, just beeing curious if that's true.

Andrea

[Doug Griffith]

I've heard some say that arcs such as those written in g-code produce smoother curves with less jaggies than what a print driver outputs (which is most likely HPGL polylines). I've never seen it myself and don't know for sure. If your laser produces these jaggies in a straight line, arcs would have nothing to do with it and the cause would most likely be from the frequency of the laser and/or stepping in your mechanical drive system. I would like to see a comparison of circles cut on the same laser using the same frequency/power/speed settings but with a driver vs g-code arcs.

[Zsolt Paul]

I have a vytek and it cuts via its own software. Its a pain sometimes to have to go to another software since I don't like designing in it, but they claim that its a necessary evil for the quality to be as good as possible. According to them its far superior to a print driver. I never cut with a print diver so I can't compare, but I've been told...

[paul mott]

Andrea,

I run GCode so can cut arcs easily.
The code for the disc on the left was produced from HPGL and is a series of small but straight lines, whereas the code for the disc on the right is one 360 degree arc.
Photography is not my strong point but if you had the two discs in your hand you could see that the difference is quite considerable.

Hope this helps.

Paul.

[Rodne Gold]

I think the interpolation of arcs and curves has more to do with the motion system than the design or interpolation software - with steppers , even micro steppers , you will get a more stepped type engraving. Thing is , the jaggies on pex are more to do with remelt etc than the amount of straight lines a circle is interpolated to.
Most drivers would do a really good job of arcs whatever method they use.

[paul mott]

Sorry to dis-agree Rodne but the two discs in my photo were made less than an hour ago (on the same machine with the same settings) and I think, prove that the software does indeed determine the final finish. :)

Paul.

[Andrea Weissenseel]

Paul, that really helped. I can see it on the picture, and that's what I've been told - that edges on acrylic almost look like flame polished, when cut with a cutting software. I have no problems with arcs on mine, they output perfectly - but the edges not really, mine look excatly like your left disc. Which software do you use ?

Thx, Andrea

[Frank Corker]

Andrea, a friend of mine also says that when they engrave a photograph using Photoshop, they get a better result than when using Corel Draw. I haven't tried it personally, maybe I shall put it on my list of things to do when it's raining (ok in my case, when the sun is shining).

[paul mott]

Andrea,

The circle for the disc on the left was drawn in ‘Front Designer’ which exports standard HPGL which I then converted to GCode.
The GCode for the other disc was created with the ‘Cut Circle’ wizard within ‘Mach3’.
I have attached the GCode (for anyone who is interested) which illustrates the difference in the two toolpaths. Obviously, physics determines that any change of direction is associated with a stop (which is where the ‘Remelt’ that Rodne mentioned comes into effect) but the stops associated with a continuous arc are considerably shorter in time than the stops associated with many straight lines thus the difference I am achieving in edge finish.

Paul.

[Andrea Weissenseel]

when they engrave a photograph using Photoshop, they get a better result than when using Corel Draw.

that sounds familiar, I couldn't validate it, but I remember doing some photographs engraving directly out of Photoshop und had a "WOW" effect - but also never tried to do the same pic via Corel

[Dan Hintz]

Some seem to be a bit mystified that one program (or driver) does better than another and are coming to the erroneous conclusion that (in this case) print drivers are better. It all boils down to resolution of the final output, regardless of who creates it. The best output will be created when its resolution matches or exceeds that of the motion control system, plain and simple.

Granted, a driver written for a specific machine will often be faster, but the qualilty can be equal. If, for example, Corel uses a print driver that outputs in a format approximating an arc with straight lines, a quality driver will allow you to set the final line length to the same size as the smallest stepper motor move. That is often not the case. Stand alone programs tied to a specific machine brand force you to import, but since they are specific to that machine, they are usually set to output that minimal line length (considering there's no practical way to actually output a true arc, it must be emulated with straight lines with any fixed step-size system... and yes, this includes servo motors as they use a fixed step-size feedback mechanism).

Paul, if you have the option of decreasing your line length more, you should have no problem with getting both circles to look identical.

[paul mott]

Hi Dan,


Paul, if you have the option of decreasing your line length more, you should have no problem with getting both circles to look identical. Quite right with this BUT if I reduce the line step length to the maximum (theoretical) resolution of my machine (0,00625mm) the resulting program could (and I can only guess on this) contain more lines of code than Mach3 can handle (10,000,000) :D. This is, I think, why circular interpolation was introduced as a machine control option within the GCode structure.

It is all a bit mystifying but (without, I hope, stirring up a hornets nest) the "Eurolaser foot keyring sample" shown in this earlier thread http://www.sawmillcreek.org/attachment.php?attachmentid=157014&d=1280346326 could well have been straight from the machine without any re-finishing or flame polishing.
As said earlier, I am not a photography expert but here is another photo of my edge finish using circular interpolation G02. My edge quality is not quite as good as the foot sample but it is not that far off.

Paul.

[Rodne Gold]

I have the same machine as Andrea - I have used Corel, engravelab and Acad on this machine with exactly the same results.
The driver it'self has settings where you can increase curve resolution - but it makes no difference as to what program is used.
These machines do not use steppers in their motion system but use DC servo motors and work closed loop.

IMO if you are getting different results bypassing your machines drivers or using other software , then its driver is faulty.
I use Corel on my Tekcel , Isels , digital print and cutters and have NO issues whatsoever in terms of curve quality - I doubt it's got anything to do with Corel.

[Dan Hintz]

if I reduce the line step length to the maximum (theoretical) resolution of my machine (0,00625mm) the resulting program could (and I can only guess on this) contain more lines of code than Mach3 can handle (10,000,000) :D. This is, I think, why circular interpolation was introduced as a machine control option within the GCode structure.
Yep, and why the dedicated driver programs are usually faster/more accurate... you get to skip the intermediate stage of transforming the circle into a bunch of small, straight lines and storing them to a file first. The dedicated programs do the transform on the fly, but they do the same transform as the generic print drivers, just at the highest resolution the hardware will support.

In a nutshell, the best print driver will send any curves in their native vector format directly to the piece of software/firmware controlling the hardware. The controlling software/firmware should do the translation to straight lines/points... transforming the arcs anywhere farther up the chain leaves open room for positioning error, reduced speed, etc.

[paul mott]

Rodne,

Please excuse me for being a pain, I am a new boy to this forum and to lasers in general - I only finished building my laser about 4 Months ago (my 'home built' machine is actually a wood router with laser attached).

If you have the time, it would be great if you could post a picture of a 50mm or so diameter disc you have cut from 6mm acrylic. I would be most interested to see the edge quality others are getting - please.

(I suspect that all the standard machine drivers treat curves / arcs as a series of discrete straight lines rather than a circular interpolation which is why different software packages appear to produce the same results).

Paul.

[paul mott]

Yep, and why the dedicated driver programs are usually faster/more accurate... you get to skip the intermediate stage of transforming the circle into a bunch of small, straight lines and storing them to a file first. The dedicated programs do the transform on the fly, but they do the same transform as the generic print drivers, just at the highest resolution the hardware will support.

In a nutshell, the best print driver will send any curves in their native vector format directly to the piece of software/firmware controlling the hardware. The controlling software/firmware should do the translation to straight lines/points... transforming the arcs anywhere farther up the chain leaves open room for positioning error, reduced speed, etc. Excellent Dan,
Why, I wonder, don't all the standard machine drivers do this (if they did then everyone would get acrylic edges without vertical striations). :)

Paul.

[Dan Hintz]

Your guess is as good as mine. I haven't paid much attention to it on my machine, but since the more recent ULS machines are controlled directly by the computer, there would be no reason not to export converted curves at the highest resolution the machine can handle. There is a slider bar for selecting resolution (though like other ULS options, there's no quantitative value to choose from, just a "feel good" pictorial representation of the different levels), but I have not slid it all of the way over to see if it would make a difference in acrylic.

[paul mott]

There is a slider bar for selecting resolution (though like other ULS options, there's no quantitative value to choose from, just a "feel good" pictorial representation of the different levels), but I have not slid it all of the way over to see if it would make a difference in acrylic.

I would be most interested to see the results of your experiments with this Dan.

Paul.

[Rodne Gold]

Paul , I don't have a pic of that on this puter and am not at work to test cut.
We never get a perfect edge on thick stuff but it's perfectly serviceable and acceptable - and we make prestigious awards for some real critical quality conscious folk.

Striations in laser cut curves are not actually machine generated - especially at the slow speeds required to cut 6mm, they are more to do with air pressure , air assist direction , melt ejection, the substrate you cutting on and so forth. Also caused by material movement either due to local expansion at the cut point or a shaking machine - normally the items being cut are not secured down so can micro move.
I have isels XYZ tablers with steppers - wrote my own post processor from acad for these - and they have some quite gross step rate and even then , its real difficult to notice these steps - you see more marks from bit chatter and motion slop.

[paul mott]

Andrea,

Please accept my apologies as we seem to have hijacked your thread here.

As much as I have tried this afternoon I cannot cut anything circular (originated from Corel Draw X4) which does not have vertical striations when cut in acrylic. An exported file can, however, be 'cleaned up' using something like Vector Magic but it is an awful lot of extra work. I am not sure if the later versions of Corel Draw are an improvement but for me, Corel is not an option for creating artwork which is subsequently to be laser cut in acrylic.
I presume (although I don't know for sure) there are different types of DXF files with some allowing high definition curves (circular interpolation) and others not (a bit like HPGL and HPGL(2)) and I need to investigate this further. Draughting programs such as TurboCAD produce excellent results as does Splines in Solidworks. I think Dan is absolutely correct in saying that some software exports curves at a higher resolution than others and this is perhaps where the difference lies.

I certainly get different results from using different CAD software as I have tried to show in my first two photos bearing in mind that my CAM software and laser settings etc. were exactly the same for both examples.

Paul.

[Andrea Weissenseel]

don't worry Paul :D most of the discussion only sound like railstation to me *lol* - the "Eurolaser foot" was the reason for my question.

[Joe De Medeiros]

As said earlier, I am not a photography expert but here is another photo of my edge finish using circular interpolation G02. My edge quality is not quite as good as the foot sample but it is not that far off.

Paul.

Does the machine run slower when using circular interpolation.

[Lee DeRaud]

It sounds like there are two completely different mechanisms in play here.

First, there are the vibration/ablation/whatever cutting artifacts that Rodne describes, which should be more-or-less uniform (or at least uniformly random, if you get my drift) around the perimeter. (Note that these can show up on straight lines as well.)

Second, there are what I'd call "path quantization effects", which is what Paul is talking about. Those should vary around the perimeter, being worse at the "flats", i.e. where the cut is running parallel to an axis.

Both could be described as "vertical striations", but they're not the same thing. However, at the level of detail we're talking about, which effect is dominant probably depends on things a lot more subtle than driver/firmware architecture.

[paul mott]

Does the machine run slower when using circular interpolation.

Joe,

The time taken to complete both discs is the same. So the answer is no - the feedrate is the same for both discs.

Paul.

[Bob Davis]

...really speaks for itself.
I really dislike the striations on cut acrylic, and I certainly find them to be very noticeable. The G-Code cut example appears to be far superior in edge quality.
As there were no other variables between the two items it would appear that the software is at least a large contributor to the result.

[paul mott]

Bob,

I think it is like Dan has already said it's the item which has the worst resolution which is the limiting factor. I can't get good results using Corel because it doesn't create very round circles (I suppose drawing at 100 x full size and then scaling back at cut time may resolve the issue) :)

Coming back to Andrea's original question - I don't think it has been established yet, just where the weak point in the chain is.

Is it the CAD software ?
Is it the post processor ?
Is it the CAM software ?
Is it the machine driver firmware ?

The Spirit 40W machine itself is certainly capable of producing excellent results so just what is holding it back with vector cutting ?.

The LaserPro website shows excellent edge quality on it's acrylic samples.

Paul.

[Dan Hintz]

In the case of Corel, the weak link is the print driver (supplied by the laser manufacturer). Corel will hand it a vector arc, if the driver supports it, and that's as accurate as you can get. What the driver does with it after that is entirely manufacturer dependent.

I still intend to sun some tests this weekend, time permitting...

[Rodne Gold]

Considering we use Corel cross platform with all manner of machinery - and get sterling results whether it be printing , lasering , vector cutting , cnc cutting etc - It seems Corel isnt the issue,
And having tried other software in Corels place an having got no better output - seems on our GCC's that the software makes no difference....
What makes a big difference is settings in the driver
There is an option for enhanced vector cutting that will give better interpolation but slower cutting. It's a driver function and not related to the software used.

[Doug Griffith]

Andrea,

Please accept my apologies as we seem to have hijacked your thread here.

As much as I have tried this afternoon I cannot cut anything circular (originated from Corel Draw X4) which does not have vertical striations when cut in acrylic. An exported file can, however, be 'cleaned up' using something like Vector Magic but it is an awful lot of extra work. I am not sure if the later versions of Corel Draw are an improvement but for me, Corel is not an option for creating artwork which is subsequently to be laser cut in acrylic.
I presume (although I don't know for sure) there are different types of DXF files with some allowing high definition curves (circular interpolation) and others not (a bit like HPGL and HPGL(2)) and I need to investigate this further. Draughting programs such as TurboCAD produce excellent results as does Splines in Solidworks. I think Dan is absolutely correct in saying that some software exports curves at a higher resolution than others and this is perhaps where the difference lies.

I certainly get different results from using different CAD software as I have tried to show in my first two photos bearing in mind that my CAM software and laser settings etc. were exactly the same for both examples.

Paul.

For some reason I think we are talking apples and oranges here. Your machine is a g-code driven router with a laser head. There is going to be a cutting difference between an arc comprised of many lines (GO1) to true arcs (G02/G03/G??). Even if the mechanics of the machine or pulse frequency or slag ejection makes a difference, arcs will give better results. Most of us use laser engravers with print drivers programmed specifically for the hardware. I'd like to see Epilog or ULS or other another manufacturer chime in and let us know if their driver converts arcs to lines, if the hardware does, or if their machine processes the x,y as passed in without conversion. If the hardware converts to lines, dedicated cutting software wouldn't make a difference unless it reduces line length. Otherwise, I think a better comparison would be dedicated cutting software vs print driver on the same machine.

[paul mott]

For some reason I think we are talking apples and oranges here. Your machine is a g-code driven router with a laser head. There is going to be a cutting difference between an arc comprised of many lines (GO1) to true arcs (G02/G03/G??). Even if the mechanics of the machine or pulse frequency or slag ejection makes a difference, arcs will give better results. Most of us use laser engravers with print drivers programmed specifically for the hardware. I'd like to see Epilog or ULS or other another manufacturer chime in and let us know if their driver converts arcs to lines, if the hardware does, or if their machine processes the x,y as passed in without conversion. If the hardware converts to lines, dedicated cutting software wouldn't make a difference unless it reduces line length. Otherwise, I think a better comparison would be dedicated cutting software vs print driver on the same machine. I think you are absolutely right here Doug. The most ironic thing is that the GCC's hardware is capable of producing far superior results, when it comes to cutting acrylic, than it actually does in practice.
Consensus of opinion is that it is the print driver which is the limiting factor then. :(

Paul.

[paul mott]

don't worry Paul :D most of the discussion only sound like railstation to me *lol* - the "Eurolaser foot" was the reason for my question. Andrea,
Looking at the video of them cutting that foot they may be using a very short focal distance lens (would be unusual for 10mm thickness). If this is the case it would explain the 'barrelling' which is seen on the edges of the work.
I think I have shown that it is possible to cut a curved surface without any visible vertical striations (without looking at it through a magnifying glass that is). http://www.sawmillcreek.org/attachment.php?attachmentid=158085&d=1281615614 however, the striations are still there - they have to be there because the machine only moves in small but straight lines, it doesn’t really do arcs it just approximates them.
Just for the hell of it I traced the web picture of the foot, gave it a bit of editing and traced it again into a DXF then used LazyCam to create the GCode. My edge finish may not be quite as good as theirs http://www.sawmillcreek.org/attachment.php?attachmentid=157013&d=1280346315 but I now know for certain that it is possible for their machine to have produced that quality of edge without any re-finishing or flame polishing etc. (After all if I can get these results using my 'homebuilt machine’, I am sure they can get excellent results with their professional stuff). I still have to decide what I shall engrave on my acrylic foot (probably my mobile phone number) but I think this will make a very nice keyring. (Incidentally mine has been cut from 6mm, I don’t have any 10mm otherwise I would have used it). Perhaps a lesson to be learned by all of us is that if we can’t do something that does not mean that it cannot be done.
I don’t know how close you are to Eurolaser but it would be nice if you could ask them to send you a sample of the foot then you will be able to judge for yourself, first hand, if it has been re-finished after cutting or not.
In the absence of any other pictures which have been posted here it is impossible to make a true comparison but, from comments posted, I think it is fair to say that, when it comes to cutting curved acrylic items, a dedicated cutting software / firmware solution (similar to that which I am using), will certainly out perform the standard printer driver. It would still be nice to see some pictures of rounded stuff cut with the enhanced vector cutting option that Rodne mentioned. Do you have this option with your GCC ?.

Paul.

[Doug Griffith]

In the absence of any other pictures which have been posted here it is impossible to make a true comparison but, from comments posted, I think it is fair to say that, when it comes to cutting curved acrylic items, a dedicated cutting software / firmware solution (similar to that which I am using), will certainly out perform the standard printer driver.

I still think we're talking apples and oranges here. You are assuming that all dedicated cutting software uses arcs instead of lines and that all print drivers use lines instead of arcs. What you have proven is that true g-gode arcs produce better results than polyline g-code arcs and that your hardware is not the limiting variable. Personally, my next laser will be g-code driven because almost all I do is cut and the striations drive me nuts.

[paul mott]

Hi Doug,

Quite possibly.

Incidentally these little feet were cut with lines not arcs. :)

Paul.

addition;

I have attached part of my GCode just in case proof of lines as opposed to arcs is required.

[Rodne Gold]

The point you are not getting is that striations when cutting acrylic are mostly NOT a function of arc/line interpolation - heres an article by synrad that illustrates that

http://www.synrad.com/e-newsletters/08_24_06.htm

Note that a perfect flame polished edge is not possible. Even the best edges exhibit very slight vertical striations due to the periodic ejection of vapor. Although decreasing cut speed can eliminate these striations, micro cracking then occurs on the cut face.

In fact if you search the archives for "striation" or "striations" you will see it even applys to metal cutting.

[Scott Shepherd]

I'd like to see Epilog or ULS or other another manufacturer chime in and let us know if their driver converts arcs to lines, if the hardware does, or if their machine processes the x,y as passed in without conversion. If the hardware converts to lines, dedicated cutting software wouldn't make a difference unless it reduces line length. Otherwise, I think a better comparison would be dedicated cutting software vs print driver on the same machine.

It is my understanding that the ULS driver converts everything into lines. I'm not positive on that, but I worked with some of the software people on some issues, and it's my unprofessional opinion that it does exactly that.

[Lee DeRaud]

It is my understanding that the ULS driver converts everything into lines. I'm not positive on that, but I worked with some of the software people on some issues, and it's my unprofessional opinion that it does exactly that.That's true for my VL200: see my post in the 'cutting PMMA' thread. YMMV on higher-end machines like yours.

[paul mott]

The point you are not getting is that striations when cutting acrylic are mostly NOT a function of arc/line interpolation - heres an article by synrad that illustrates that
http://www.synrad.com/e-newsletters/08_24_06.htm
Note that a perfect flame polished edge is not possible. Even the best edges exhibit very slight vertical striations due to the periodic ejection of vapor. Although decreasing cut speed can eliminate these striations, micro cracking then occurs on the cut face.
In fact if you search the archives for "striation" or "striations" you will see it even applys to metal cutting. This is, of course, quite true Rodne but what I am looking at here is the striations that are visible on the work once all the above issues have been addressed (or at least as well as they can be).

Why is it that we can all get good edge finish (no visible striations) on the straight edges but not get good edge finish on the circles or curves ?.

Paul.

[Rodne Gold]

Often the reason you cant get a good finish on the curves is due to the ramping function of the driver.
You have 2 main parameters to get the ideal settings - speed and power
On curves speed decreases mostly but if you have not enabled a ramping function your settings change and you are now putting more energy into the material.
The laser power should be able to vary as it speeds up or down thru curves etc- not all drivers have this and some allow it as an option.
If you can enable this - you will probably have the same performance on flats vs curves.
If you have coin effects around curves , its most likely a remelt/ejection issue - if you have visible facets - its most likely an interpolation issue.
Easy way to check is to use black anodised ally and engrave something at the speed you would use for acrylic - inspect the white line under magnification - this will soon show facets and then you know what your issue is.

[paul mott]

Easy way to check is to use black anodised ally and engrave something at the speed you would use for acrylic - inspect the white line under magnification - this will soon show facets and then you know what your issue is.

Thanks Rodne,

Excellent information, why didn't I think of that :o
(now to find something anodized black).

Paul.

[Lee DeRaud]

Why is it that we can all get good edge finish (no visible striations) on the straight edges...Can we?

The answer to your question may be, "Because we're lazy buggers and tend to run our straight-line tests on vertical and horizontal lines." A line slanted at an arbitrary angle to the motion system's axes is a distinctly different animal.

(And no, I don't have all, or even any decent percentage, of the answers. I'm still trying to figure out why the edges of a hole are usually smoother than the edges of the disk that was cut from it. :eek:)

[paul mott]

:D:D Nice one Lee :D:D

Paul.

[Lee DeRaud]

:D:D Nice one Lee :D:DI said "arbitrary". :cool::p I wouldn't be surprised if some/most motion systems also handle lines at 45 degrees as a special case...

I'm also confused by the fact that when this thread started, polylines, as opposed to "real" curves, were the problem, and now they seem to be the solution.)

[paul mott]

I said "arbitrary". :cool::p I wouldn't be surprised if some/most motion systems also handle lines at 45 degrees as a special case...My motion system has its axis at 90 degrees but the step size can be very small (0,00625mm) I think any angle can be handled as long as the trajectory planner gets good information.


I'm also confused by the fact that when this thread started, polylines, as opposed to "real" curves, were the problem, and now they seem to be the solution.) I'm not sure that is really the case as the problem with curves is that different software treats them differently so there is no obvious common solution. The easy way out is to consider straight lines, if they can be cut at any angle then place 100 of them around in a circle (without joining them into a poly-line) and cut that. If the flats are visible then increase the number of lines until the flats cannot be seen.
My motion system only moves in straight lines so I could not actually cut a true curve even if I wanted to.

Hope this helps with the confusion :)

Paul.

[Lee DeRaud]

I'm not sure that is really the case as the problem with curves is that different software treats them differently so there is no obvious common solution. The easy way out is to consider straight lines, if they can be cut at any angle then place 100 of them around in a circle (without joining them into a poly-line) and cut that. If the flats are visible then increase the number of lines until the flats cannot be seen.The problem isn't that you can see the flats, it's that you can see the corners: they show up as vertical, um, striations on the edge.

And for the life of me, I can see absolutely no difference between drawing a curve as N separate line segments rather than a N-node polyline, except maybe that the separate-line implementation is likely to be (much) slower.

My motion system only moves in straight lines so I could not actually cut a true curve even if I wanted to.The confusion comes from the fact that just a couple of days ago you said:

The code for the disc on the left was produced from HPGL and is a series of small but straight lines, whereas the code for the disc on the right is one 360 degree arc.
Photography is not my strong point but if you had the two discs in your hand you could see that the difference is quite considerable.

[paul mott]

Hi Lee,


The problem isn't that you can see the flats, it's that you can see the corners: they show up as vertical, um, striations on the edge.This is not quite the case and you would probably have to do it to see the proof. It's probably (just guessing here) the melting which occurs on either side of the kerf which blends the corners, in any event they are not visible and this is shown here http://www.sawmillcreek.org/attachment.php?attachmentid=158085&d=1281615614 This disc is comprised of approx. 400 straight lines and therefore has approx. 400 flats around its circumference but neither the flats or the joins/corners/intersections are visible.


And for the life of me, I can see absolutely no difference between drawing a curve as N separate line segments rather than a N-node polyline, except maybe that the separate-line implementation is likely to be (much) slower. I fully understand your reasoning here but again you would have to actually do it to see the difference. Because acrylic is cut slow the machine cutting time is exactly the same between a file of 400 separate lines of code (where each line of code represents one pair of X and Y axis moves) and a file of 7 lines of code (which represents one circular interpolation move).


The confusion comes from the fact that just a couple of days ago you said:…….. You have to forgive me if I make mistakes but I am not the expert here I am learning from you Guys and a couple of days can be a long time when I am learning new stuff.

Although I cut a lot of acrylic and I learn fast, it is all straight line stuff and easy to obtain a good edge finish. It was only when I started to cut profiles that the troubles started.
This is where I was at some 5 weeks ago http://www.cooperman.talktalk.net/pic1.jpg (http://www.machsupport.com/forum/index.php?action=dlattach;topic=12444.0;attach=217 30;image) and I thought this was quite good work. What really bugged me was that the edge finish on the crossbars was good compared to the rest – why – because these were straight lines parallel to a machine axis (just what Lee said in a previous post).
From reading others posts on this forum, following links that had been posted (mainly to manufacturer’s websites) and digesting all the information I could swallow I moved forward to this http://www.cooperman.talktalk.net/pic2.jpg (http://www.machsupport.com/forum/index.php?action=dlattach;topic=12444.0;attach=217 38;image) quite a considerable improvement I thought (which, incidentally, I could not have done so quickly without you guys and this forum). I have to learn from others because life is just not long enough to make all the mistakes myself.
Then I started contributing to this thread where my problem was that I knew how to cut a circle with striations and I knew how to cut a circle without striations what I didn’t know was exactly what was causing the damn striations.
I had already overcome the re-melt, ejection of vapour, air assist, vibration, laser power, cutting speed, focal point etc.etc. problems and knew it had to be something else. Rodne’s suggestion of drawing onto a black anodized surface was the key – this way I can, under a microscope, see my actual toolpath on the surface of the work. A long time buddy of mine (who doesn’t actually have a laser) who lives some 6000 miles away kindly agreed to do some work on analyzing my toolpaths as they are sent to the machine and here is part of just one result http://www.cooperman.talktalk.net/pic3.png (http://www.machsupport.com/forum/index.php?action=dlattach;topic=12444.0;attach=221 02;image) Now I was expecting to see a series of flats not a wobbly line but it was similar to what I was seeing under the microscope and indicates that my machine was cutting exactly what it had been asked to cut. Wobbly line in = wobbly line out. As you are aware clear acrylic is not very forgiving optically and even the slightest deviations show up as a mountain range.
As I said before different software treats circles differently and some are better than others at approximating the curve / arc. I was totally unable to obtain a good circle exported from Corel (doesn’t mean that it can’t be done just that I couldn’t do it). But straight lines in = straight lines out, this is the direction I have now taken and I think that my results speak for themselves. This, however, is not the end of the story because there is always more to learn and progress to be made and I am certain that a new idea or something better will surface.

I think I have resolved my problems, for the time being, but for you Guys, if you want to profile cut acrylic without striations I believe that it can be done and that the straight line approach (as a way of getting the Corel export data through the print driver) is well worth investigating but initially you have to get everything else right, in other words you have to be able to cut straight lines, at any angle without striations – Is this possible ? I think so. There is nothing special about my machine (it’s a wood router) and if I can cut acrylic without striations then so can you.

Sorry if I have waffled on a bit but it’s a rainy day here and work is a bit slack – roll on Friday.

Paul.

[Lee DeRaud]

This is not quite the case and you would probably have to do it to see the proof. It's probably (just guessing here) the melting which occurs on either side of the kerf which blends the corners, in any event they are not visible and this is shown here http://www.sawmillcreek.org/attachment.php?attachmentid=158085&d=1281615614 This disc is comprised of approx. 400 straight lines and therefore has approx. 400 flats around its circumference but neither the flats or the joins/corners/intersections are visible.So basically the method depends on the melt from the slow cut to smooth ("interpolate") between the segments. That almost sounds reasonable, but I have to ask why that same effect doesn't also smooth out whatever jaggies/striations are induced by doing the cut as a curve.

Note also that we've already established that some driver/firmware/motion-system combinations in common use implement a curve as a short-segment polyline. I admit that the one I have doesn't go to as small a segment as I'd prefer: it uses 128 segments for a 2" circle...I haven't looked at the output file for larger-radius curves yet. From the discussion above, it sounds like my only optimizations to that are (1) lower speed to get more melt or (2) constructing the curve as a polyline directly to increase the segment count, which probably is also going to slow the cut down, resulting in more melt. (It's definitely going to slow down my drawing process...:eek:)

I fully understand your reasoning here but again you would have to actually do it to see the difference. Because acrylic is cut slow the machine cutting time is exactly the same between a file of 400 separate lines of code (where each line of code represents one pair of X and Y axis moves) and a file of 7 lines of code (which represents one circular interpolation move).Yes, but what I was commenting on was the differential cutting time between 400 separate short lines and a 400-node polyline of the same path length. I suppose one difference is that you could cut the 400 separate lines in a non-contiguous manner (e.g. do all the 'odd' lines, then all the 'even' lines), but I guarantee that would slow things up enormously.

[paul mott]

Hi Lee,

I am only guessing that the melt smooths the surface I don't know that for sure but what else could it be ?. I am open to offers on this one.

I will go and cut a 128 line 50mm circle and post the result for us to see.

When I said I cut slow I mean real slow.I get good results at a feedrate of 120mm per minute (approx. 1 minute to cut a 50mm disc) but 60mm per minute works better for the really small stuff (this increases the width of the kerf and enables the part to come out of the stock easier). At these low speeds there is absolutely no issue with program execution time no matter if it contains 10 times the number of code lines. I set the feedrate I require and the trajectory planner takes care of organizing the motion control system (at least I think that is the way it works).


At the moment I don't know just how to create a circle with exactly128 lines but how hard can it be ?.


Paul.

[Doug Griffith]

At the moment I don't know just how to create a circle with exactly128 lines but how hard can it be ?.

That takes about 2 seconds. Just create a 128 sided polygon.

[paul mott]

That takes about 2 seconds. Just create a 128 sided polygon.

Doug,

Thanks for the tip. Does that create a circle comprising of 128 separate lines or one poly line with 128 faces ?.

Perhaps you could post one for me please.

Paul.

[Lee DeRaud]

When I said I cut slow I mean real slow.I get good results at a feedrate of 120mm per minute (approx. 1 minute to cut a 50mm disc) but 60mm per minute works better for the really small stuff (this increases the width of the kerf and enables the part to come out of the stock easier).Ouch! My normal settings only take 30-32 seconds for a 2" circle...you're talking up to 4X slower. :eek:

Ok, I get it now: cut and flame-polish, all in one step.

[Joe De Medeiros]

Doug,

Thanks for the tip. Does that create a circle comprising of 128 separate lines or one poly line with 128 faces ?.

Perhaps you could post one for me please.

Paul.

if you're using autocad it's a pline.

[Lee DeRaud]

Does that create a circle comprising of 128 separate lines or one poly line with 128 faces?Yes. :cool:

Depending on when you look at it.

I'm sort of kidding, but the only thing we care about is what it looks like when it gets to the laser. To Corel it's a 128-side polygon...probably only takes a few dozen bytes to describe it internally: bounding box coordinates and number of sides.

[Doug Griffith]

Here it is as a polyline. EPS because I don't use Corel.

[paul mott]

Ok, I get it now: cut and flame-polish, all in one step.:D:D That's about right :D:D

Paul.

http://www.machsupport.com/forum/index.php?action=dlattach;topic=12444.0;attach=218 24;image

[Doug Griffith]

Here are 2 more at 256 sides and 512 sides if you feel inclined.

[paul mott]

Thanks Doug,

I have converted the 128. (My program now tells me that it is 132 sides but that's near enough for a country job ;) ).

Paul.

[paul mott]

Hi Lee,

These little discs are 25mm and 50mm diameter and each were cut using 132 straight lines (the code uses 134 pairs of X-Y moves). I wasn’t able to arrive at exactly128 lines but I thought this would be near enough. Looking very closely I can just start to get a hint that the 50mm disc is comprised of straight lines but there again I know it is, so I am looking for them, which is perhaps not a very fair test. What do you think ?.

Paul.

[paul mott]

Thanks Joe,

I sort of missed your posting while I was uploading a picture. :(

thanks again, Paul.

[Dan Hintz]

Ouch! My normal settings only take 30-32 seconds for a 2" circle...you're talking up to 4X slower. :eek:

Ok, I get it now: cut and flame-polish, all in one step.
Yeah, and I also wonder what kind of kerf comes from such a slow setting...

[Lee DeRaud]

These little discs are 25mm and 50mm diameter and each were cut using 132 straight lines (the code uses 134 pairs of X-Y moves). I wasn’t able to arrive at exactly128 lines but I thought this would be near enough. Looking very closely I can just start to get a hint that the 50mm disc is comprised of straight lines but there again I know it is, so I am looking for them, which is perhaps not a very fair test. What do you think ?Better than what I normally get, but I haven't tried slowing down my machine yet. Wonder what yours looks like if you speed it up 2X-4X? :cool:

The other thing on my to-do list is to generate a 128-side polygon in Corel, size it to 2"x2", and see how its output coordinates compare with the polyline nodes from a 2" "circle". (Warning: I'll probably mangle some terminology on what follows.) I'm wondering how the RMS error from the true circular path compares between the two: I suspect the two 128-point sets are not derived the same way (e.g. the polyline may be the output of a DDA computed around the ideal path). It could well be that the polygon is less accurate in that sense, but appears smoother because it's more consistent (for lack of a better word).

Or maybe it's all "melt effects", I have no idea.:p

[paul mott]

Hi Lee,


Better than what I normally get, but I haven't tried slowing down my machine yet. Wonder what yours looks like if you speed it up 2X-4X? :cool:If I speed it up the results are bad :( (if you look at the videos of professionals cutting acrylic you will see that they cut slow as well).

I would be most interested in your results of polygon/polyline/circle investigation.
Have you got any pictures of your cutting progress yet ?

Regarding cutting speed etc.......
Although the power density may be identical between 'High Power / High Speed' and 'Low Power / Low Speed' the thing that seems to be important is not to saturate the material with energy. In other words don't put in more than the thermal conductivity of the material can handle.
When cutting at low speed the power settings have to be reduced accordingly so that the beam just cuts through the material and no more. If too much power is used here the material over gasses (boils) and the surface finish then gets really bad.
Another, quite important factor is to reduce the air assist flow to the minimum necessary to protect your lens or better still use an 'air knife'. Cold air directed into the kerf causes premature cooling which results in a milky white color edge (this is probably the 'micro cracking' which Synrad refers to in their documentation).

Paul.

[Rodne Gold]

Acrylic is a sort of anomaly to normal stuff - you actually want to put in more energy than the material can handle and have a very pronounced heat affected zone - so it gets that flame polished look. Basically the laser works by treppaning when cutting (drilling a series of overlapping holes to cut) and relies on vaporising the material to make the hole. In Acrylics case you want not only vapour , but melt.
In terms of the micro cracking , what happens is that perspex or acrylic is highly stressed and heat or any machining operation involving heat stresses the material even more and thus the pex will stress crack where it has been laser engraved or cut due to the heat. Extruded perspex is far worse in this regard than cast cos of its short polymer chain...some solvents - even their vapours will result in uncontrolled cracking where the pex is machined this can happen many months or even years after the product is supplied.

[paul mott]

Acrylic is a sort of anomaly to normal stuff - you actually want to put in more energy than the material can handle and have a very pronounced heat affected zone - so it gets that flame polished look.Guess we will just have to 'agree to differ' on this one Rodne.

Incidentally, I have posted lots of pictures of the circular stuff I have cut in the way I have just described and the results speak for themselves. I would really like to see some pictures of the circular parts you have cut using "more energy than the material can handle", just for comparison purposes.
Please :)


Paul.

[Rodne Gold]

What are you disagreeing about Paul?
The reason you get the polished look is because of the melting of the acrylic and it only melts cos you have gone beyond sheer vaporization - you are actually relying on the extra heat to melt the stuff - flame polishing does the same thing.
Same with wood , ideally lasers would vaporise the wood and you would not get the charred or branded look but rather an super clean engraving with little contrast from the background.
The only way to get dark engraving on wood is overpowering it or having a setting that will NOT vaporise but char or allow resins to discolour.
The Holy Grail in laser engraving is to have as little a heat affected zone (the zone that is not vaporised) as possible - but as I said , for pex or wood , that wont give the best results.

If for eg , you are cutting paper then you will have HUGE issues with heat affected zones and would thus use a different strategy to get best results - you would use minimal power and set your PPI (Pulses per inch - also sometimes called frequency = the no of times that a laser pulses or drills holes per inch) to a minimum that will give a smooth look and should then have clean cuts with NO edge discolouration.

[paul mott]

Hi Rodne,

What I do is - I experiment and build on the information I have gained from others in order to acquire a sound practical knowledge. (the theoretical knowledge is easily acquired by searching the internet)
I value the information I have gained from yourself and others on this forum and I would not be able to have reached the stage I have without it, however, from my practical work the results of using no more heat / power than is necessary to cut through the acrylic has, for me, so far, produced the best results. If "using more power than the material can handle" produces better results then that is exactly what I want to do but my tests have shown that increasing the power (which I can 'do on the fly') has a negative effect on the edge finish not a positive effect.
So what I am really saying is - if that works for you, then fine, but it just doesn't work for me so we have to agree to differ.
But if there is anything I can do to improve what I already have then I dearly need to know about it so please keep the information coming.

Come on all you Guys reading this thread, if you have cut a circular path or profile in acrylic then please post some pictures - I know I am not the only one who has done it ?


Paul.

[Michael Hunter]

Paul - How powerful is your laser?

I think that the power could be significant, affecting the way that the "best" cut is achieved as follows -

A low power (<45W) laser probably depends on melting its way thorugh some of the acrylic - and so also relies on the air-assist to eject the molten stuff.

A high power (>100W) would do most, if not all, of the cut by instant vapourisation of the material. If the pulse frequency is also high, then I would expect the cut to be smoother and far less affected by the quality of the air-assist, since the ammount of melt would be reduced.


It is very useful to know that high quality cuts are possible and I shall certainly be re-evaluating my settings to see if I can improve on my current output.
I think though, that there are so many factors in play that there is unlikely to be a one-size-fits-all solution to getting rid of the striations.

[Doug Griffith]

All techno-babble aside, I get decent quality cuts if I choose to do so. Not perfect, but good enough. Normally, I'm trying to make a buck and time matters, so I use the fastest possible setting that is acceptable for the application at hand. I always start at 100% speed / 100% power and work down from there.

If I wanted a perfect edge and time didn't matter, I'd build a fancy hold down fixture for my CNC, machine the part (which would give a dead-nut straight kerf), then flame polish.