Printing dwg files to Trotec Job Control 10 without Autocad?

[David Malicky]

I'm looking for low-cost software to print dwg files to a Trotec with Job Control 10.1. This is in a univ dept and our standard OS is Win7-64. Almost all of our work is vector. Here's what I've tried so far:
- Progecad 2011 and 2013: Our educational licenses insert an "Educational version..." header/footer on every print. They also don't work well with Trotec's driver.
- Autocad LT Educational: same header/footer problem as Progecad.
- Autocad 2004: Works but more cumbersome print process, only takes old dwg files, and needs a 32-bit OS.
- Draftsight V1R3.2: Works well after I found Jon Colley's archive post to select "ask for jobname" in JC. But it appears the license activation only works for 1 login ID. It does not print beziers, so vector-marked curves are start-stop and relatively slow (much slower than our old GCC Mercury, which apparently had a CNC-like motion planner that didn't need beziers). Draftsight is our best option, currently.

Here are some I'm looking at:
CorelCAD 2013 Education: from what I've read, it does not print an "Edu" header/footer. I don't know if it can print beziers.
CorelDraw Technical Suite X6 Education: accepts dwg files. I don't know if it can print beziers like regular CorelDraw, or about the "Edu" header/footer issue.
TurboCAD and NanoCAD: probably don't print beziers.

Has anyone tried any of those with Trotec JC, or have another rec for dwg files? Any insight is appreciated!

David

[Dave Sheldrake]

AutoDESK DWG TruView?

http://usa.autodesk.com/adsk/servlet/pc/index?id=6703438&siteID=123112

cheers

Dave

[Brian R Cain]

Coreldraw will be fine. The only issues you are likely to encounter will be compatibility with specific versions of Autocad. In other words, each release of Coreldraw will be a version behind on the file types it's compatible with, if that makes sense. Simplest thing to do is to download a trial from the Corel website and check it out. It gives you a couple of weeks to test it. If you're happy with it, buy the licence.

[Dan Hintz]

Export files as DXFs... should have no problem printing in any version of AutoCAD.

[Mike Null]

I would also go with CorelDraw.

[David Malicky]

Thanks all -- very helpful! I didn't know about TruView -- will give that a try. And good to know about the Corel trial and dwg versions. I'm just realizing regular CorelDraw can read dwg, so we don't necessarily need the "Technical Suite". I'll see how these go and may be back with other questions. Thanks again!

[David Malicky]

I did some testing with CorelDraw sending beziers to JC10, but the Trotec still has jerky motion marking the curves, and it's not much faster overall. It moves smoothly and quickly only when marking a circle or constant-radius arc (with a start-stop jerk only at 0, 90, 180, and 270 deg, like an old implementation of a CNC G2 or G3 command). Clicking on a bezier-sent curve in JC10, I see the Advanced Geometries feature breaks up the curve into arc-segments rather than straight line segments. But the Trotec motion controller still does a stop-start cycle at the end of each of those arc-segments, just like with line-segments. So except for circles, I'm not seeing much advantage to Enhanced Geometries.

I don't understand why Trotec's motion controller lacks a constant-velocity mode, which would prevent the stop-start behavior between line or curve segments. CNC motion controllers have had this for decades. Our GCC Mercury from 9 years ago evidently had it (and didn't need beziers) and is much faster and smoother than the Trotec for vector curve marking. Certainly, there is a lot that is impressive about Trotec. But so far I'm disappointed with vector curves on our "Speedy". Is there something I'm missing to get smooth and fast curves on a Trotec?

[Scott Shepherd]

I did some testing with CorelDraw sending beziers to JC10, but the Trotec still has jerky motion marking the curves, and it's not much faster overall. It moves smoothly and quickly only when marking a circle or constant-radius arc (with a start-stop jerk only at 0, 90, 180, and 270 deg, like an old implementation of a CNC G2 or G3 command). Clicking on a bezier-sent curve in JC10, I see the Advanced Geometries feature breaks up the curve into arc-segments rather than straight line segments. But the Trotec motion controller still does a stop-start cycle at the end of each of those arc-segments, just like with line-segments. So except for circles, I'm not seeing much advantage to Enhanced Geometries.

I don't understand why Trotec's motion controller lacks a constant-velocity mode, which would prevent the stop-start behavior between line or curve segments. CNC motion controllers have had this for decades. Our GCC Mercury from 9 years ago evidently had it (and didn't need beziers) and is much faster and smoother than the Trotec for vector curve marking. Certainly, there is a lot that is impressive about Trotec. But so far I'm disappointed with vector curves on our "Speedy". Is there something I'm missing to get smooth and fast curves on a Trotec?

Do you have it setup correctly in Corel? You know you have to set that to output beziers, right?

Also, it does have something like that, I believe. It sounds like the Intelligent Path Control feature. Turn IPC on for that color and try it, setting it to quality, not speed.

I'll also add that JCX is still in it's infancy. Trotec's powerful app, called TroCam is like $5,000 and it's always been the master for smooth curves. It's my understanding (which could be wrong), that they are slowly migrating a lot of that technology into their Job Control product for things like Speedy users. TroCam does lead in's, lead outs, and great curves. I'd expect it to be a little rough right now but as time goes on, it's going to be excellent, knowing what I know about it all. I can't wait. Only issue is they are slow to release things. Something said to be out in 3 months might come out in 12 months. I know changes that were supposed to be coming out this last spring and I've yet to hear anything about any of it. They tend to spend a lot of time refining it and worry about the product more than the deadline, which is fine.

[David Malicky]

Steve, thanks for the reply -- that's helpful. Yep, I had the "send beziers" box checked, and I also noticed the bezier had to be created within CorelDraw (an imported dwg spline printed as straight segments, just like from CAD).

That's interesting that TroCam makes smooth curves -- are they also fast? With the USB connection, I assume there is a machine controller chip inside the Speedy to direct the axis motors. I.e., the USB delivers the general plot commands or G code, then the Speedy interprets those into servomotor commands. If so, then the speed and smoothness issues would mostly depend on the Speedy firmware and electronics, not so much on the specific PC software. Maybe TroCam comes with special firmware, too? Still, the programming needed for fast and smooth vectors isn't rocket science -- grbl (freeware) and Mach3 both have it.

I've tried IPC a little before and found it was smoother, but slower. This time I did a systematic test and got results that are both bizarre and helpful. Here's the test so far: A spline and 2 circles were sent from Autocad 2004 to JCX as vectors ("Cut"). JCX saw it as short line segments as usual (no beziers). The plot measures 8.6" across:
270598

JCX was configured with 10% Power, 5000 Hz, and variable Speed setting. IPC was first set to Off, then Speed, then a few tests set to Accuracy. I timed how long it took to complete the job on a Speedy 300 (time is from laser on to off):
270599

Notes:


The bizarre cyclic behavior of IPC with Speed setting was not measurement error -- those times were repeatable (although the valley locations might be spline-dependent).
I also checked 40% and 100% Speed -- very similar to 20%.
I didn't measure the depth-of-cut, but the slower IPC runs were definitely deeper. I don't know if switching from 10% to 2% Speed would cut deeper (the intuitive result) or less deep since the job time is less. In any case, it seems clear that using Speed to control cut depth will be confusing, especially with IPC.
The fastest run (9 sec) seemed a little slower than our Mercury, but it was respectably fast, ~2x faster than my prior marking speed (20%), and it was surprisingly smooth, too.
For only doing vector curve marking, the take-home message seems to be: IPC = Off, Speed = 4%, do cut depth adjustments with Power.
Another important test is to check the time to complete some long straight lines... I'd expect that to improve ~linearly with % Speed. Hopefully soon.
Would anyone like to try replicating these results or other variations? The dwg and data files are attached, if helpful.

[Scott Shepherd]

David, I didn't see anything that looked wrong in that example. Remember, IPC is based on geometry, so the shapes of the part matter more than anything. I had one conversation with the guy that wrote JCX in Austria, 1 time, about it, so I'm no expert, but what he explained to me about it was far different than what I was hearing from others about it. My suspicion is that if you did a rectangle with square corners, IPC would have no real impact. However, if you did a small gear with lots of teeth, you'd see far different results from the test. So, to me, IPC is very geometry dependent, which would cause a variety of results.

Just for the record, a speed of 4 for vector cutting is very fast on the Trotec. I think 4 is at the point where you'd stop getting smooth cuts anyway, like most machines, there are some speed limits on things. If you were running a square, 4 would work fine, if you were doing a circle, I'd expect the curve to be not so smooth.

When it comes to vector cutting, I also wouldn't expect any machine to be significantly faster than another. Vectoring is very limited for things like that. I wouldn't expect the Trotec to be "faster" than your Mercury, an Epilog, Universal, or a Chinese machine. However, rastering is an entire different situation and that's where you'll see the speed increase.

But nothing you posted surprises me.

[Mike Null]

I don't have JCX but the Trotec vector speed is 31 IPS which is very fast. Like Steve, I rarely have mine set at anything higher than 2.

With my version of JC I have never experienced jagged lines and I cut rounded corner name tags and various curved labels every day.

[Scott Shepherd]

One thing I don't agree with in the factory material settings in JCX is the speeds. They prefer to take multiple light cuts for vector cutting engraving laminates. They do it at very high speeds. So their factory setting might be a speed of 10. Well, take a speed of 10 around a corner and you've got a mess. Take it on a straight edge and it's fine. I'm not sure why they sent factory settings for materials over with speeds that exceed the limitations of the machine, but they do.

[Dave Sheldrake]

Can somebody possibly try this modded file and see if it's any different?

270613

cheers

Dave

[David Malicky]

Thanks for the replies. Steve, I'm a little surprised you're not surprised, but that probably just reflects the difference in our experience with Trotec. I checked the effect of IPC on a large rectangle with diagonal lines (no curves) -- Yes, there is almost no relationship between job time and Speed setting for those long straight lines. But for those long lines, IPC=Speed was really slow: between 6x and 13x slower than IPC=Off (Speeds from 5% to 30%). What did the Trotec person say IPC was especially good at? Also, what did you mean by 'a mess' with 10% Speed around a corner?

To me, here are a few things that don't make intuitive sense and aren't user-friendly, from the last experiment:
- With IPC set to Speed, increasing the Speed setting from 2.0% to 2.2% increases the job time from 18 to 34 seconds (almost double).
- With IPC Off, increasing the Speed setting from 4% to 10% increases job time from 9 seconds to 16 seconds. I'd understand if the job time just stayed the same beyond a certain Speed % --that's very reasonable -- but why should it increase so much? Since it does increase, the user would then need to do experiments like this to find the optimum Speed setting for the fastest job. It also means that there is no Speed % that is optimally fast for a job with both straight and curved lines (see below).

Mike, thanks for the post -- to clarify I don't see any jagged or inaccurate lines. The jerky behavior is the stop and start of the laser head as it draws a curve -- it causes the whole machine to jerk, too. It's not a major problem to me, but I was expecting to see fast and smooth motion control like our old Mercury, and it may wear things out sooner.

Today I timed some runs for vector marking of long straight lines (a ~10" square with multiple diagonals and centerlines, no curves).
270687
- This went as expected: job times decrease steadily until about 20% Speed, above which there is no change.
- Job times for 20% Speed are about 3x faster than for 4% Speed.
- Optimum Speed for long lines is 20%, but for curves it's around 4% (probably depends on the curve). So for vector marking jobs with both lines and curves (like many of ours), any Speed setting will be a compromise and none will be particularly fast. It's disappointing to me that the Speedy has such great hardware, but the current vector firmware/software is apparently holding it back. Trotec makes a big deal of the machine's speed including its name, but for general vector marking jobs, it's currently looking like it will be slower than our old machine.

[Scott Shepherd]

Thanks for the replies. Steve, I'm a little surprised you're not surprised, but that probably just reflects the difference in our experience with Trotec. I checked the effect of IPC on a large rectangle with diagonal lines (no curves) -- Yes, there is almost no relationship between job time and Speed setting for those long straight lines. But for those long lines, IPC=Speed was really slow: between 6x and 13x slower than IPC=Off (Speeds from 5% to 30%). What did the Trotec person say IPC was especially good at? Also, what did you mean by 'a mess' with 10% Speed around a corner?

To me, here are a few things that don't make intuitive sense and aren't user-friendly, from the last experiment:
- With IPC set to Speed, increasing the Speed setting from 2.0% to 2.2% increases the job time from 18 to 34 seconds (almost double).
- With IPC Off, increasing the Speed setting from 4% to 10% increases job time from 9 seconds to 16 seconds. I'd understand if the job time just stayed the same beyond a certain Speed % --that's very reasonable -- but why should it increase so much? Since it does increase, the user would then need to do experiments like this to find the optimum Speed setting for the fastest job. It also means that there is no Speed % that is optimally fast for a job with both straight and curved lines (see below).

Mike, thanks for the post -- to clarify I don't see any jagged or inaccurate lines. The jerky behavior is the stop and start of the laser head as it draws a curve -- it causes the whole machine to jerk, too. It's not a major problem to me, but I was expecting to see fast and smooth motion control like our old Mercury, and it may wear things out sooner.

Today I timed some runs for vector marking of long straight lines (a ~10" square with multiple diagonals and centerlines, no curves).
270687
- This went as expected: job times decrease steadily until about 20% Speed, above which there is no change.
- Job times for 20% Speed are about 3x faster than for 4% Speed.
- Optimum Speed for long lines is 20%, but for curves it's around 4% (probably depends on the curve). So for vector marking jobs with both lines and curves (like many of ours), any Speed setting will be a compromise and none will be particularly fast. It's disappointing to me that the Speedy has such great hardware, but the current vector firmware/software is apparently holding it back. Trotec makes a big deal of the machine's speed including its name, but for general vector marking jobs, it's currently looking like it will be slower than our old machine.

David, IPC is not something I'd recommend using on a daily basis. It's a tool that's available to solve a particular issue and that issue is dealing with the problem all machines have and that's computing small curves, etc. Let's take a 1" x 4" rectangle with .125 radii on the corners. If you had it at a speed of 10, the machine cannot move at a speed of 10 and calculate such a small move around the corner at that speed. The machine can't accelerate and decelerate fast enough. If you did the calculate on how fast the machine is moving and how little distance it is around the corner, you'd see it's not enough room/time. IPC is made to resolve that particular issue. All lasers do it. On our Universal we have a particular problem with it. Our machine slows down to go around a corner and then never goes back up to full speed on some pieces. It causes a job that should be 30 seconds to take 2 minutes or more. Doesn't work so well when it's 100 of them on a sheet. I've sent videos to them, I've shown people at trade shows, and no one seems to care. So it's not a new issue (needing to slow down around corners). It's just how they deal with it.

If a small area requires a speed of 1 to get around smoothly, and you put a speed of 10 in, then I would expect it to take longer because it now how to do calculations to try and make it work. I only mentioned trying IPC because you were having an issue. For the shapes you posted, I would never use IPC.

We do see serious increases in speed on the Trotec, even in vectoring certain things. Some things it doesn't matter, some things it does matter. I don't recall the exact numbers (I posted them on here somewhere), but we finished a job not long ago that only required light power, so our two machines should have been close in times. They weren't. The Trotec was at 6:30 or something and the Universal was double that.

I understand your frustration, but I don't share your findings. On cast acrylic, the edges aren't anything special. On extruded acrylic, the edges are the best of any laser we've seen. If I were doing the job you posted, I'd turn IPC off and I wouldn't exceed 3 on the speed. I also wouldn't expect vectoring to be much faster than any other machine.

You also can't push the machine faster than it should go. I know the Epilog's and Universal's allow you to put in speeds the machine can't handle well either, so I'm not sure how Mercury caps their speed to keep users from doing that, but just know that anything above 3 on speed is going to have an issue if you are expecting high quality. I also know that 3 on the Trotec is faster than 10 on the Universal, so it's all relative.

I remember a friend of mine getting his Trotec several years ago. I got an email from him and he was not happy. Quality on vectors was awful and that's mostly what he did. This went on for days. His speed was 13 or something like that. He finally got someone logged into his computer from Trotec and they looked at it and said "Oh, speed of 13, you can't do that.", and they changed it to something far less and the machine cut like butter. He's been happy since. To him, 13 wasn't fast. To the machine, it was way faster than it should run.

If if were me, I'd cap the speeds, but I don't program them, so I have no say so.

[Mike Lysov]

Hi David,

it seems you are having the same issues as me. After using Spirit GX for 2 years my new 240W laser with cnc control and stepper motors was a nightmare for me until I adapted to it. As Scott mentioned cnc controlled lasers cannot move at the speed you want and either change speeds based on the shape geometry and size inside their motion controllers or change the shape outline as they cut it out.

On my machine if I want to cut out a huge rectangle(say 500x500mm) at the high speed(10000mm/m) it will move fast but it won't do sharp corners. The size of that shape allows to move as fast as the speed set or close to it but at this speed it cannot suddenly stop at the corner(unless it told to do it with settings), make 90 degrees turn and go to the other direction. So to do the rectangle with 90 degrees corners I have to use dwell settings for some specific range of angles(I use 20 to 120 degree range) and dwell stop time(I use 1-5 ms). It tells the controller to stop for dwell time set if the required angle with the specified range. It must be a small range may be even smaller than 20-120 because if it is set from 1-359 degrees it will stop for dwell time at each spline start point and that's when you will get jerky movements.

If I try to do 20x20mm rectangle at the same speed it will never do it at 10000mm/m because as Scott mentioned above it just cannot accelerate to such a fast speed on such a small length of curves. So on small shapes it moves much slower(may be even 600mm/m) but sometimes it does sharp angles without using dwell settings because of much slower actual speed.

Now regarding actual speeds and time to do a job, on your first graph where there seems to be no direct correlation between the speed and the time to complete your job I have this problem too. Since speed is something based on geometry and size your laser motion controller decides what speed to use. I have seen some strange decisions my controller makes where there is some magical point in speed set up by me when the job can be completed in minimum time. What that speed should be I do not know, all I know if I try to increase the speed after that point it actually takes more time to complete the same job. So it goes like the faster I am trying to do the same job the slower it is actually done unless I have found and set up this optimal speed. That's the most annoying thing I have ever seen. It looks like a stupid logic inside my controller that does not have actual threshold in speed for some particular shape and changes actual speed of cutting with every new speed set by me.

For the smooth motion I just use prefetch function and Pline output settings that tells the machine to do smooth cutting from splines. You must have something like this in your JC software.

And one more thing, my software that has come with the laser does not like very small segments that are created in Corel Draw. If there is something like 0.1-0.2mm long segment between two nodes in my artwork software generates G-Code to do it as an arc which is somehow done as a full circle. It ruins my designs because it actually trying to cut out a circle where there is no such thing in my design. But that's not all. The radius of that circle can be ridiculous. I had one job that stopped my controller working and I had to reset firmware of the controller. I made a cnc file with g code and emailed it to my friend who is an expert in CNC controlled machines. He checked G-Code and he found out that for one tiny segment my software generated a code to cut a circle with 139 000 000 kms in radius. No wonder the controller stopped working because when it reached this command it was trying to go from the center of that circle to a point 139 millions kilometers away at the highest speed possible. It did not even bother my controller that the size of my laser bed is billion times smaller, it just could not move that fast to reach that point far-far away.

Now I check and modify all my designs before I bring them to my laser control software. First I reduce the number of nodes until there is no tiny segments left in them. Then I export them as a dxf file from Corel Draw and import them into my laser control software. Then I use a prefetch function. It reduces the number of G-Code commands and makes cutting smooth.

[Scott Shepherd]

I might also add that in my experience, there is no perfect laser that does everything right, all the time. While it's unquestionable to me that you could make every machine on the market beat the Trotec in some test, I also believe the biggest picture is which one gives you more pluses than minuses when compared to other machines. I have test files that will make an Epilog fail miserably. I know the weak points of the Universal. I also know some of the weak points of the Trotec. I think I can make all 3 come out on top of some tests I can run. But day in and day out, our Trotec STRONGLY out performs every other laser we own or have owned.

I'm sorry you're having issues and believe the Trotec isn't as great as you thought, but I suspect as time goes by, you'll change that opinion. I love our Trotec a lot more today than I did in the early months. I also remember about a year into it, realizing I had been doing some basic things wrong because I didn't know what I didn't know (I still have that happen frequently :) ).

Hang in there, give it some time and I think you'll be happy.

[Mike Lysov]

there is no perfect laser that does everything right, all the time.

may be you right Scott, but my Spirit GX never cuts out circles out of nowhere and it increases the speed of cutting if it is changed by me. With Spirit GX I can at least cut thin material at the maximum speed possible.

I am kind of glad to know that Trotec works the same way as my new laser. At some point I started thinking that it had a bad motion controller or software it came with was bad. Now based on David's experience with his Trotec I think it is the way these CNC controlled lasers work.

[Scott Shepherd]

may be you right Scott, but my Spirit GX never cuts out circles out of nowhere and it increases the speed of cutting if it is changed by me. With Spirit GX I can at least cut thin material at the maximum speed possible.

And that's exactly my point. Your GCC might do that better than the Trotec, but when it comes to other things, the Trotec might leave the GCC in the dust. It's a balance on what's right for your business. If the things you do all day long work better on the GCC, then the GCC is a better option for you. If the things that an Epilog does better are what you do all day, then the Epilog is a better choice, and so on. I think it takes a lot more research than most people realize to pick the machine that's best for their needs. I don't equate buying a laser to buying a new computer, I think it can be much more complex.

Having owned an Epilog, Universal, and a Trotec, I have no problems standing by my Trotec and it's ability to make us money. Is it perfect? Nope. Does it have some bugs? Sure. Need me to name the bugs in the Epilog or Universal? I can easily do so. They all have issues. It's how those companies respond to those issues that's important to me. I've had a LOT of discussions with anyone at Trotec that will listen to me, which included the guy that wrote Job Control. I'm confident that when issues are brought to their attention, that they will be resolved. I share that same confidence with Universal. Granted, I don't think any of them solve issues as quickly as I'd like, but it's a much tougher job for some tasks and most of the time, the things we ask for aren't simple fixes.

My mouse stopped working properly with the latest version of Adobe Illustrator. It's been reported as a bug THREE months ago. Still no fix. Works with Photoshop, just some features with Illustrator don't work. If Adobe can't fix my mouse zoom in Illustrator in 3 months, maybe the timeline laser manufacturers have for fixes isn't so slow :)

I will also say that I've never had a smooth circle come from a DXF file into a graphics file. You can draw perfect circles in a CAD package all day long, but I've never seen one get imported to my liking yet.

[David Malicky]

Steve and Mike, thanks for the replies!

Steve, That's really helpful info about IPC -- now I know when and when not to use it.

To clarify, I do realize the Trotec has many strengths and I appreciate those. I'm disappointed with the current vector marking speed and smoothness, although being in a teaching environment means these weaknesses of the Trotec affect more than just me. We want to show our students what sophisticated motion control looks like, and other faculty here will recognize these motion control problems and will form their own judgments. The Trotec will likely be used as an example of how *not* to do it (unless Trotec fixes the problem).

These Trotec weaknesses aren't an inherent limitation of software, electronics, or hardware. Any digitally-based motion-controlled machine is a CNC machine, and almost all of those were forced to solve the smoothness issues a long time ago. GCC evidently solved it. An example of smooth and fast motion control: http://youtu.be/Zop1z9RNsyQ?t=6s
Compared to a laser, CNC routers/mills have extremely heavy axes, which means they need high torque motors that are inherently slower. Yet, even they achieve fast and smooth motion control. They do slow down for tight corners, but do not stutter and are very fast in straighter curves.

Here is our Trotec at 6% speed, jerky but better job time than 4% for these gentle curves (~19 seconds): http://youtu.be/9-lfbDfxz2E

Here is our Mercury running the same plot (1:48 to 2:02, ~14 seconds): http://youtu.be/_vTMwOHV-wI?t=1m48s

The Trotec clearly has superior mechanicals based on the rapids. Job time is slower because the motion controller can't achieve constant velocity through the curves, and needs to be crippled to a low % speed so the curves aren't even slower. There is no good technical reason for these problems today. It's probably a firmware, communication, or look-ahead problem, and it should be fixable.

Mike, that's interesting yours has a similar problem--thanks for your post. I believe prefetch is the same thing as (more) look-ahead and I agree that is a likely solution. I'd love to try (more) look-ahead, but I don't see any setting for it in JCX.

I imagine these Trotec problems still exist for 2 reasons:
- Laser heads are very light weight, so stop-start motion isn't catastrophic like it is for a mill, so Trotec hasn't been forced to fix it yet.
- Most users have accepted the problem, limiting the machine to 2-3% to keep motion smooth. Steve, it's great to hear Trotec has been responsive to customers -- contacting them sounds like the next step on this, or perhaps they read SMC. I would be more than happy to report back to SMC that they fixed the issues.

[Michael Kowalczyk]

Export files as DXFs... should have no problem printing in any version of AutoCAD.
Hey Dan,
I try not to use any DXF files since they are segmented. I can hit a button on the JC but that adds another step and another file so a DWG works best for me.

To the original poster...Try Rhino5. Great academic price and great program. Have not tried to output in JC 10 yet but it should, so I have heard.;)

[Michael Kowalczyk]

David,
What are your settings?
Do you have the expert version?

There is an optimizer with a right click of the mouse. I use it frequently but can't recall the name as of now and my laser is tubeless right now and disconnected. Regassed one should be on it's way tomorrow. lasted OVER ten years. wooot hooooooo.
Never go above 3% vector speed on a speedy II. other newer ones may be higher % for velocity.

[Mike Lysov]

Hi David,

I know what you mean comparing your Mercury and Trotec and that's the thing Scott seems to miss. It is really strange that your Trotec and my new laser somehow have been designed with different from GCC lasers motion controllers and these controllers decide what speed to use and how jerky they move. As I have mentioned before my Spirit as well as your Mercury seems to have completely different motion controllers and I do not think it has something to do with gantry weight of on bigger machines. So as long as your Trotec as well as my new laser are supplied with motion controllers like they are now we have no choice and just have to adapt to them. They won't fix it because all these problems come from the motion controllers they use.

I do not think your Trotec has something called Prefetch but may be that "IPC" thing in Trotec does exactly the same.

I am learning more with my new laser everyday and now it seems that the Prefetch function can create much more problems than it can fix. It actually reduces g-code when you run it but at the same time it generates really strange numbers in g-code like a few hundred kilometers radiuses for arcs or points on a bed with 8-13 digits on the left side of the number. When it does it my laser stops and I have to reset firmware in its motion controller. I have found the better way to get corel draw designed graphic into my laser control software. Here are the steps I do now:

1) I reduce number of nodes to a minimum in CorelDraw until there is dramatic changes done to a shape. In most cases I can reduce from 1600 nodes to just 300 and the shape still looks the same
2) I export that shape to DXF but not with CorelDraw built in DXF export but with DXF export that comes with e-cut macros(cost only $50 but has a lot usefull stuff). I use 0.2mm tolerance in DXF export settings for 15cm tall shapes.

3) I import the DXF file into laser control software and cut it as it is(no prefetching). E-Cut Dxf export actually creates a DXF file with arcs and polylines instead of splines so there is no need in using the Prefetch function at all. It cuts smooth continuous lines and g-code is compact straight after importing it.

[Scott Shepherd]

I know what you mean comparing your Mercury and Trotec and that's the thing Scott seems to miss.

I'm not missing anything Mike, I understand his issue. What I was stating was that it's easy to find something you think doesn't work correctly and compare it to something you believe works correctly by your standard. My point was that we can do that with every single machine on the market.

But I do understand his point and I've emailed him for the file to see if it's something I can look at on our machine and see if it's something it handles different on the Trotec, or if it's a setting that needs to be changed and the issue goes away. Before we pass judgement on it, let's see if it's a new user with an incorrect setting or if it's a machine issue that is a legitimate issue that needs to be looked at by the factory. At this point, I have no idea because I don't have enough data to know.

[Dave Sheldrake]

AutoCAD Method:

Select spline
Input "Splinedit"
Hit return
Input P (polyline)
Hit return
Input 1 (quality)
Hit return
Spline converts to arcs and lines all joined with the same shape and node positions but no splines.

cheers

Dave

The free Cad package Draftsite does the same using the same commands

[Mike Lysov]

Sorry Scott, you are indeed missing my point.

I do not mean that my new laser or David's Trotec is bad. I just do not understand why these two lasers(and there are obviously others done in the same way) that are built using G-Code commands and motion controllers to support them which do not do any good to accurate and speedy laser cutting. As David mentioned there are no heavy spindles mounted on their gantries like in CNC routers/mills/plasmas, laser head assemblies are very lightweight.
GCC lasers are build based on HPGL commands and they seem to use different motion controllers. As far as I know HPGL is used for plotters. What good does it do to laser cutting on GCC lasers? The answer is it cuts as it is, without any jerky movements, it does square corners and it moves as fast as it's been told to cut.

So for me after using ULS and GCC lasers with HPGL controllers for a while my new laser was some kind of shock. I just cannot see any benefit of using G-Code over HPGL as all factors including speed and accuracy of cutting now depend on the motion controller and I have no control over it at all. If my new laser had been built based on HGPL I would have not had any troubles to do all my products on one laser choosing the best speed based on the thickness of material. But I cannot do it because my new laser does not allow me to set the speed based on how thick material is and what it is. Instead there are a lot of other factors like my artwork size and whether it has sharp corners or not. Not to mention conversion problems between artwork done in CorelDraw with splines into DXF/DWG formats with arcs and polylines for smother cutting.

If you have any explanation why Trotec or my laser manufacturer have decided to use G-Code or so called DSP based controllers over HGPL ones please provide it. Unfortunately it won't change the way my laser or Trotec is built but at least we will know why they are done the way they are.

[Mike Lysov]

AutoCAD Method:

Select spline
Input "Splinedit"
Hit return
Input P (polyline)
Hit return
Input 1 (quality)
Hit return
Spline converts to arcs and lines all joined with the same shape and node positions but no splines.

cheers

Dave

The free Cad package Draftsite does the same using the same commands

Draftsight does it but it does not like most AutoCad versions. I believe it works only in one version but I do not remember what it is. When I tried to do it with R14 version I usually use it did not work for Spline to Polyline conversion in DraftSight. I had to google for it and found some advise about the correct version number on one Draftsight forum. It may be R11, but I am not sure.

[Scott Shepherd]

Sorry Scott, you are indeed missing my point.

Sorry Mike, I'm not missing your point at all. I fully understand what you are asking. I don't know the answer, but I do understand the question. I've had a few discussions with people over the last year with people at the Trotec factory discussing something very similar. I know they had a lot of things in the pipeline at the time, but I haven't heard or seen any details of them, other than some general discussions.

I can't speak for why any manufacturer decides to use one method or the other. Like I said, I think I can point to weaknesses in all machine. I can make the ULS that uses hpgl fail quite easily. So I'm not sure that hpgl is the utopia of laser engraving that you may believe.

If I turn the speed to 15 on my Universal and tell it to do a rectangle with rounded corners, having a 1/8" corner, then that corner is going to be very erratic and jagged. I haven't seen a single laser out there that doesn't do that. So in fact, you do need to base your settings on the work, the geometry in the work, etc. Those things do have a factor in running any job.

What Trotec has done with their Intelligent Path Control is to try and remedy that exact situation mentioned above. If you turn on IPC and tell it "Quality" is the priority, and you do a rounded corner rectangle, set the speed to a very fast speed, it's going to recalculate that speed settings to the slowest possible speed it can do with a smooth cut.

Having run lasers for 6 years now, 3 different brands, I can't say that I haven't seen this exact same problem in every single machine I've ever run.

If I get the opportunity, I'll have this discussion with some Trotec factory people and see if I can report back.

But I do understand your statement quite well. As someone with 30 years experience programming NC and CNC machines as well as various CADD and CAM packages, believe me, I understand.

[Scott Shepherd]

David sent me the file and this video shows me running the file at the speed David used (6.0 on speed). I opened his DXF file in Adobe Illustrator, saved it as an eps file and imported it into CorelDraw as an eps file. I welded the 2 pieces together and that's all I did to it (it was 2 pieces). There is a slight glitch in the motion at one point, but I believe that to be a node issue in the file, not a flaw in the machine or motion system.

Take a look. Not sure how to get much smoother than this :


http://youtu.be/96XByRQlUuI

[David Malicky]

Steve, yes, that looks very smooth and much faster. Those rapids are also quick. Thanks for doing that! The question now is how to get the Trotec to do that directly from the dwg file, preferably from a CAD print. Illustrator -> EPS -> Corel is a lot of programs (time, $) and steps -- and more time than saved on the laser. Our Mercury received the same output from Autocad as JCX, yet was smooth and fast. Another interesting test would be tighter curves. The guitar shape has very gentle curves which probably match well with the Enhanced Geometries feature (long arc segments with fewer transitions). When I tried CorelDraw + NativeBeziers + SendBeziers + EnhancedGeometries with tight curves, the Speedy still stuttered a lot. I can make a Corel file and post if helpful to test.

Michael, Yes, we're running JCX in expert mode. Settings: standard process, 500 dpi, 4-6% speed, 30% power, 1000-5000 Hz, IPC off. All our work is vectors. Thanks, I'll check out Rhino5. We're using Autocad 2004 for now on a 32-bit machine as it seems to have the least bugs of the CAD programs we have access to. I also tried NanoCad and DoubleCad but those didn't mesh with JCX very well. And I tried CorelDraw but I couldn't edit the dwg import very well; maybe there's an easy way I don't know about. As you all know, it's helpful to make changes during print iterations, and a pain if we have to go back to CAD to edit, then import in Corel just to send to JCX.

Mike, Right, I didn't mean the gantry weight helped, just that with a light laser head Trotec hasn't been *forced* to fix it. GCC evidently implemented a sophisticated motion planner voluntarily. From the JCX pdf manual, IPC does use lookahead -- BUT, there's apparently something off in the implementation as it's extremely slow for the medium and large plots I've tried with it. Sounds like your Prefetch implementaion has some bugs, too. I don't think this is a HPGL vs. G-code issue; the math and computations in the motion planner should be very similar. Properly done as with most CNC machines, with any language, a motion planner should be able to use lookahead to generate smooth and fast motor commands. With the right firmware, the Trotec's hardware should be able to leave the Mercury in the dust.

Thinking more about the symptoms from earlier in this thread:
- Stutter steps / stop-start motion / jerks at higher % speed
- Overall slower job time at higher speed compared to around 5%
--> A lack of (enough) look-ahead explains both of those very well:
- At a low enough % speed, the laser head arrives at the destination *after* the motion controller has processed the next move command. So the controller can keep the motors running at full speed through the destination and onto the next segment.
- At a higher % speed, the laser head arrives at the destination *before* the motion controller has processed the next move command. So the controller must decelerate and stop the motors, finish processing the next move, then reaccelerate the motors. That would explain why the Trotec's jobtimes for curves are slower above ~5% speed, rather than just leveling off.

Dave and Mike, thanks, I will try splines vs polylines to see if any difference.

[Scott Shepherd]

David, I only posted my software steps for reference. I'm on a Mac which is why it opened in Illustrator. I saved it as eps just to open in Corel.

You can skip that and just import the drawing file directly into Corel.