Interpreting Speed setting on lasers - long post [Archive]

Richard Rumancik

06-18-2007, 11:47 PM

In other threads, a few of us have discussed the use of the speed settings when comparing different lasers as well as trying to predict settings for a new thickness based on know information. There are also often questions on the forum as to "what settings do you use" for a certain material.

I finally found some time to do some testing on my LaserPro Mercury. What I was hoping to confirm was that when you select a particular "percentage of speed" in the laser settings, it would be a linear relationship. That is, 100% speed would be maximum spec'd speed, 50% would be half-speed, 25% would be quarter-speed, etc.

I found that on my machine, this is approximately true for rastering. But for vector jobs, what I found seems very far from linear.

For raster, I determined that a speed of 50% will result in a job time of approximately twice the job time of a speed of 100%. Most of us would have expected this kind of result. It is not an exact relationship, probably because the laser has non-constant overtravel on the left and right of the image (overtravel distance depends on speed). But overall, the observed speed was quite linear for settings above 10% speed. (I don't know why the linearity was lost below 10% speed, but since that would seem very slow for raster I will ignore this for the moment.) I was also able to verify the manufacturer's specification of 42 ips maximum at 100% speed for this machine.

For vector cutting, however, I could not make any sense at all of the data I collected.

To test the vector speed, I plotted a rectangle 24" x 17.98" high (so it would fit into my height limit of 18".) This has a perimeter of approx. 84". I copied the shape on top of itself so it made 2 passes (168") to get a bit better timing accuracy at the higher speeds. The tests were all done with the door open.

What I found was that at S100, S50, S40, and S30 the elapsed time was the same: 16 seconds. (This is not very logical or expected.) At S20 T=23, and S10 T=47 seconds.

For S5 T=114, S4 T=156, S3 T=268, S2 T=404, S1 T=822. (at the slower speeds I used one rectangle or 1/2 the rectangle and extrapolated to save time, as it is a bit tiring to watch the laser cutting air.)

1. I did not see any indication of linearity in the vector cutting results.
2. I can't understand why there was no difference in elapsed time (observed speed) until I got below S30.
3. The maximum observed speed was no where near the maximum raster speed of 42 ips. It is closer to 11 ips. I don't know the logic behind this.

It gets more complicated. I wanted to see if there was a difference between cutting lines and cutting circles. (I had always has a suspicion just by watching the laser work that when cutting shapes it seemed to slow down and I wanted measured data.) I created a circle with a diameter of 13.369" (circumference 42 inches) and overlaid 4 circles on top of each other. This gave a total distance traveled of 168" for 4 circles (the same as two rectangles to enable quick comparisons). The results: S100 T=19, S90 T=21, S70 T=23, S60 T=26, S50 T=27, S20 T=46, S10 T=66, S5 T=120, S4 T=176, S3 T=248, S2 T=408, S1 T=832. These results showed that the observed speed in plotting a circle was quite different than the observed speed in plotting a rectangle, even though the commanded speed was the same %. The maximum speed achieved for the circle was 8.8 ips, and it changed for each % speed tested (which was not the case for rectangles). Rectangles plotted faster than circles even though the speed setting was the same. The observed speed at the 10% setting was about 3.6 ips when cutting a rectangle compared to 2.5 ips for the circle. This means that a cutting test done on a square (or a straight test line) will not be optimized for cutting a circle or a shape. Also, if you did a test using a circle, a square might not cut all the way through, due to the higher actual speed. The logic to this situation is not apparent, and the problems it creates are obvious.

It is possible that the circle cuts slower because the mainboard is getting bogged down in the math calculations associated with positioning, but that is only a guess.

Conclusions (applicable to the Mercury LaserPro):

1. For raster plotting, the observed speed is proportional to the % speed setting (approx.)
2. For vector cutting, the observed speed is not proportional to the % speed setting.
3. For vector cutting, observed speed is dependent on the shape being cut.

Conclusions 2 and 3 makes me think that comparing settings for vector cutting on different lasers has very limited validity. The settings may allow one to reproduce results on their own machine, but that is about all they can do.

I have not seen the laser manufacturers specify maximum vector cutting speed. It seems to me that this is a parameter that should be specified, especially if it is different from the maximum raster speed, as in appears to be in my case.

I know that not everybody is interested in this kind of analysis, but to those who are, I would be interested seeing results on other LaserPros (and later models) as well as on Epilog and ULS machines.

If anyone sees a gap in my logic or data, or can provide some insight to justify why the designers would have done things this way, please let me know . . .

I am attaching a file which I used to test my laser (saved as CorelDraw 10).

Lee DeRaud

06-19-2007, 1:12 AM

I know that not everybody is interested in this kind of analysis, but to those who are, I would be interested seeing results on other LaserPros (and later models) as well as on Epilog and ULS machines.I had similar results with a ULS VL200 for vector cutting. Rather than plotting rectangles, I was plotting lines the full width of the table. It was apparent at the higher speeds that the laser was only moving at the highest speed near the center of the cut. I assume the laser power is modulated during the acceleration at the beginning of the cut and the deceleration at the end of the cut to account for the actual speed at any given point. In any case, most materials vector-cut well below the speed (in this case, about 12%) where the motion goes linear.

Vertical (Y-axis) lines are slower than horizontal (X-axis) at the higher (non-linear) speeds, presumably due to the greater mass being moved (or more to the point, accelerated). Circles were somewhere in between...I suspect the effect is solely due to motion system mass limitations rather than processor overhead.

I didn't do much with raster. The only notable observation is that this ULS machine (or firmware/driver) does not do the kind of overtravel ramp up/down that you describe.

Rodne Gold

06-19-2007, 5:45 AM

Apart from all that , most lasers even tho rated at a specific wattage , vary. So in our case , we have 2 sets of 3 identical lasers , and cant even apply the same setting across the set of identical lasers.
There have been a lot of requests to set up a settings database , but its relatively useless.
There are also issues of ramping , hysteris etc that determine REAL speeds in terms of cycle times.
The only real way to determine whether one laser is faster than another is to actually time the same job on both.
There are a ton of ways to optimise cycle times on a job , we had a run of 10 000 labels that initially took over a minute to do per lable , we managed to reduce the cycle time to 14 secs with NO loss of quality.
for example , reducing dpi , clustering engraving where there are large areas of white space , substituting vector engraving for raster in some areas and so forth.
In terms of vector cutting , fastest speed is generally a non issue , as one is normally at full power and using fractional speeds unless you are vector cutting extremely thin material.
If you have to take a circle vs a square in terms of vector cutting , you will see that for the circle , the head has to change direction 1000's of times whereas with a square it changes direction 4x and has a chance of reaching full speed during the straight bit , whereas there are no stright bits in a circle , so the head cant achieve full speed in any segment.
You have to look at how a laser cuts too , it does so by treppanning , ie drilling a series of holes. The laser controller will modulate (or should do so) the spacing of holes compared to its speed in cutting , ie going round corners it will not be travelling as fast as in a straight line and thus it will reduce its pulsing to match , otherwise the holes would overlap too much and put heat into the material.
The ultimate conclusion of all this is that to get times you have to run the job , and settings are mostly trial and error.

Belinda Barfield

06-19-2007, 7:51 AM

I'm so glad to find out it's not just me! In my limited experience settings posted just don't transfer from machine to machine, even as Rodne pointed out, on machines made by the same manufacturer. I've tried to convert settings posted by others for use on my machine and things just didn't seem to work for me. Thanks to you guys for taking the time to do the analysis.

Peck Sidara

06-19-2007, 11:30 AM

Richard,

My reader's digest version of an answer. We're basically talking about motion control and how your laserpro is handling it.

Raster: For raster you have the acceleration and deceleration of each stroke. The engraving head doesn't start and stop on a dime, instead the engraving head has to over travel x-amount on both the left and right side before starting the next stroke. Variables such as the engraving file itself (how long of an x-axis stroke is involved), speed setting you're using will affect engraving time. That's why I wouldn't expect your raster speed to be exactly linear, close but not exact.

vector: There's more variables involved with motion control when it comes to vector cutting. There's the file itself, length of each line segment, speed and number of nodes in the vector cut. Vector cutting also has it's own acceleration and deceleration points. A 2" square cut at 10% speed will not take 5X less time than a 10" square cut at 10% speed.

Each manufacturer uses different components (motors, bearings, amplifiers (if any) & machine design in general) and will program their code differently to acheive the best possible raster and vector image.

Motion control is the good stuff behind the scenes. Motion control is what will cause you to lose hair if you try to figure out the logic behind your machine. It's also why you'll never get two different laser machines to cut exactly the same with the same times.

Hope it helps.

Richard Rumancik

06-20-2007, 1:15 PM

Peck, I appreciate you taking the time to respond. I realize you don't represent GCC and are giving general info, some of which may apply to GCC and comparable machines like Epilog, ULS, Trotec and others. I don't have any first-hand experience with these lasers so don't know which issues are unique to the LaserPro. I do have some comments on what you said . . . hope you don't feel that I'm giving you a hard time as I do like to see manufacturer's input on this forum.

Raster: For raster you have the acceleration and deceleration of each stroke. . . . the engraving head has to over travel x-amount on both the left and right side before starting the next stroke. Variables such as the engraving file itself (how long of an x-axis stroke is involved), speed setting you're using will affect engraving time. That's why I wouldn't expect your raster speed to be exactly linear, close but not exact.

Peck, I don't have a huge problem with the raster results. It is "linear enough" to make some predictions (for the same machine), which is my objective. I did acknowledge how overtravel adds some "noise" to the data. On my machine it, doesn't seem to matter how long (wide) the x-axis stroke is as you suggest it would. If I raster plot a vertical line .006 wide it will accelerate from zero, reach maximum velocity, plot a few dots, then decelerate back to zero. The overtravel seems to be about the same (about .90 inches) for a .006" wide raster plot or a 24" wide plot, based on my limited testing. Yes, of course, the speed setting will affect engraving time. . . but I'm not sure what you are saying here.

vector: There's more variables involved with motion control when it comes to vector cutting. There's the file itself, length of each line segment, speed and number of nodes in the vector cut. Vector cutting also has it's own acceleration and deceleration points.

You have mentioned 2 variables that could affect observed speed: length of the line segment and number of nodes (although they are closely related.) At higher speeds the acceleration and deceleration just before a node (change in direction) probably affects observed speed. I wish it didn't. At low speeds it shouldn't be such a factor.

A 2" square cut at 10% speed will not take 5X less time than a 10" square cut at 10% speed.

At speeds of 10% and higher you are right. But if you use lower settings, the time required for acceleration and deceleration on the corners becomes almost insignificant. (I don't have a means to directly measure it.) I did your test on the Mercury, and it is almost exactly a 5X relationship below 10% speed, and even at 10% speed, it was 4.5X which is still not bad.

Each manufacturer uses different components (motors, bearings, amplifiers (if any) & machine design in general) and will program their code differently to acheive the best possible raster and vector image.

I understand this, and I trust that they will try to optimize, but I also feel that it is the designer's job to make all the hardware behind the scene "transparent" to the end user. By that I mean that they should provide a machine that moves at 10 inches per second when it is commanded to do so. Not move at 10 ips if it a square, and move at 8 ips if it is a circle, and move at 6 ips if there are a lot of nodes, all without telling me this is going to happen. I'm not suggesting that you can violate the laws of physics but if a setting is accepted, the user tends to assume it is being used.

When I say that I am trying to figure out the logic to my machine, it is because I want to be able to operate the laser effectively from the user point of view, which requires a logical and predictable interface. (I don't need to know the math behind the motion system.) But the laser does not always operate in the manner I expect. This results in too much trial and error in setting up a job and it makes it impossible to compare results with other users.

Richard Rumancik

06-20-2007, 1:22 PM

Thanks for your input Lee. One comment . . .

I didn't do much with raster. The only notable observation is that this ULS machine (or firmware/driver) does not do the kind of overtravel ramp up/down that you describe.

Are you sure? You can see for yourself by rastering a vertical line at 100% speed. Just draw a line .006" wide so it will raster, and turn the red beam pointer on. You will probably see the carriage swinging wildly about 1-2" even though the "plot" is only .006" wide. Or try a 1" wide filled rectangle. Then the carriage will probably travel 2-3" to plot 1". The overtravel is where the laser is changing direction. It doesn't plot in that zone because it would be difficult if not impossible to program the laser to lay down consistent dots when the speed is changing.

Mike Null

06-20-2007, 1:43 PM

Richard

First, this is a really great thread. It would be helpful if we had some manufacturers chime in with some factual info here.

Second, as already pointed out, ascertaining speed settings in a vector mode is virtually impossible. If all other things were equal the drawing itself would cause varying speeds. Thus comparing them one machine vs. another is an effort in futility.

It would appear that empirical data gathered from ones own machine is the only way to be reasonably secure with the settings for vectoring.

Third, if I can horn in on the thread with what I think is a logical way for us to make comparisons for the raster mode that would be this: List speed as inches per second (if your machine operates at 80ips then at 50% speed it would be 40ips.) Same with wattage; if you're using 50% of your 60 watts then your setting is 30 watts for descriptive purposes. PPI and DPI settings are relatively constant from one machine to another.

Lee DeRaud

06-20-2007, 1:45 PM

I understand this, and I trust that they will try to optimize, but I also feel that it is the designer's job to make all the hardware behind the scene "transparent" to the end user. By that I mean that they should provide a machine that moves at 10 inches per second when it is commanded to do so. Not move at 10 ips if it a square, and move at 8 ips if it is a circle, and move at 6 ips if there are a lot of nodes, all without telling me this is going to happen. I'm not suggesting that you can violate the laws of physics but if a setting is accepted, the user tends to assume it is being used.Eh, maybe not. What you really want is for the laser to be consistent in results: if it cuts through material X at power setting P and speed setting S when cutting a square, it should still do so with the same settings if you are cutting a circle or any other shape. Further, you should get roughly the same results if you change to P*2 and S/2 (within the linearity limits of the material being cut).

Now if you have two machines that are consistent in this manner, there will be a fixed relationship in how they react to settings: you can determine the appropriate factors to apply to speed and power settings on one machine to get the same results on the other.

Anything beyond that is useful only in the sense of predictability of run-time, which is subject to a whole slew of factors (eg. file layout) that have nothing to do with the motion system or firmware/driver of the laser.

This results in too much trial and error in setting up a job and it makes it impossible to compare results with other users. If your machine is self-consistent in the manner I describe above, the first simply should not be a problem. If it isn't, then I agree that someone at the factory hasn't their engineering homework, but I suspect if a given brand/model was subject to grossly inconsistent results due to shape/orientation of cut objects, you wouldn't need a computer to hear the customer base complaining about it.

The second is really just a question of developing a data base of conversion factors between machines of differing brands and models...non-trivial, admittedly, but certainly doable. (Note that I am completely ignoring the further calibration factor required between any two laser tubes of the same "nominal" power.)

Dean Carpenter

06-20-2007, 1:46 PM

all lasers travel beyond the actual point of engraving (rastering), especially plotters but even galvo lasers. This has nothing to do with the laser tube firing.

It's so that the motion system can compensate for the vibrations created by the tollerances that allow the motion system to move.

we call it overtook.

The better your machines motion system the smaller the distance required for overtook.

The faster you drive your machine the greater the overtook will be.

Some plotter machines need up to 1". One of our plotter models needs as little as 1/4" (S300) but even this machine could not produce a good engraving result without any overtook.

Lee DeRaud

06-20-2007, 1:53 PM

Or try a 1" wide filled rectangle. Then the carriage will probably travel 2-3" to plot 1".That's pretty much what I did. If there's any overtravel, it's small fractions of an inch...not visually discernable.

Consider this: what happens when you raster and solid rectangle at the left or right edge of the work area? If the overtravel is indeed required, that shape would always have to raster from the inside to the outside, and end up taking twice as long to raster as the same shape in the center of the field. Doesn't happen...well, not on my machine anyway.

EDIT: Ok, ignore the preceding. Went up and tried it again, this time marking the area being engraved so I had a reference point to look at. Yup, about an inch of overtravel. The really odd thing is that the overtravel was one-sided when the square was over at the edge...and the job ran faster there. I haven't noticed any difference in raster quality when I have objects at the edge...then again, raster is about 2% of what I do on this thing. (Looks like I'm having fried crow for lunch today.)

Do people who do a lot of rastering have to give up the right/left-most X inches of their work area?

Dean Carpenter

06-20-2007, 2:12 PM

and there's more......!!

regarding comparing settings/speed manufacturer against manufacturer for raster engraving: impossible. Even impossible comparing machines of a different model from the same manufacturer.

There are so many variables too many to mention here.

A good test is to engrave 3 solid boxes (landscape) as seperate jobs all at full speed:

1) 1"x1"
2) 1"x4"
3) 1"x20"

Use a dpi as close to 500 dpi that the driver will allow.

Take the total time in seconds and divide it by the dpi you have used. That's gives you the magic number - the time taken to perform a raster pass using a range of overtook compensations - and makes all things reasonably equal for comparison purposes as far as top speed, turn delay, overtook, etc - ultimately top speed - is concerned.

the deeper the test (more than 1") the more accurate the comparison would be.

But that only tells you how fast a motion system will travel. It does not give you any indication about how good the engraving quality is at such speeds.

A good test is to engrave a line of + set with thickness 1mm at full speed. Look for varience in thickness between the vertical portion and the horizontal portion. Also look for fuzziness. This will not only show how good/bad the motion system is but also how good the laser tube is.

IMPORTANT: use the same (ideally 2") focal length lens for the tests

Dean

Dean Carpenter

06-20-2007, 2:19 PM

Lee,

the turn/delay is at the end of the travel between first and last laser pulse ON and OFF.

In terms of speed, with the same job, it should make no difference where it is performed on the table (work area) but depending how good/old your motion system is it will probably make a quality difference to your engraving because of things like varying belt tension, etc.

Dean

Mike Mackenzie

06-20-2007, 4:37 PM

Lee,

You will be surprised to see if you run slower raster speeds you will actually get faster times. The reason being the Raster margins are smaller when the speeds are slower.

Sometimes running 90% speed will get you a faster time than running 100% speed this is of coarse material dependent.

This is the reason that some machines are two inches or larger in X axis than the actual engraving area "Raster Margins"

Vector speeds are also a little deceiving if your graphic has a lot of curves then slow down the speed and it will cut faster than if you were to use a higher speed to begin with.

Lee DeRaud

06-20-2007, 4:52 PM

You will be surprised to see if you run slower raster speeds you will actually get faster times. The reason being the Raster margins are smaller when the speeds are slower.Understood (well, now anyway :p ).

The anomaly I noted today was that, since the VL200 has no margin or "run-off area" at the left and right edges, there is no overtravel at those points. So the same area (at a given speed) engraves faster when it is positioned at the edges since the "wasted" motion is halved. I don't know if the engraving quality is degraded by that, but I don't recall it being mentioned anywhere in the documentation as a limitation of this particular machine. (As I said, I use the thing in vector mode almost exclusively.)

More to the point, if the engraving quality isn't degraded by not having overtravel at the edges, it implies that the overtravel used elsewhere is either excessive or unnecessary.

Peck Sidara

06-20-2007, 5:50 PM

Richard,

No worries, I don't think you're giving me a hard time. Inquiring minds need to know! I'll start by saying I'm not an expert when it comes to motion control and do agree with some of the comments in this thread and will disagree with others. What I'm saying is that there's just too many variables to compare model x to model y.

Peck, I don't have a huge problem with the raster results. It is "linear enough" to make some predictions (for the same machine), which is my objective. I did acknowledge how overtravel adds some "noise" to the data. On my machine it, doesn't seem to matter how long (wide) the x-axis stroke is as you suggest it would. If I raster plot a vertical line .006 wide it will accelerate from zero, reach maximum velocity, plot a few dots, then decelerate back to zero. The overtravel seems to be about the same (about .90 inches) for a .006" wide raster plot or a 24" wide plot, based on my limited testing. Yes, of course, the speed setting will affect engraving time. . . but I'm not sure what you are saying here.

There's over travel in the x-axis when raster engraving. The higher the speed and longer distance you're traveling the more time you'll need to slow down and turn around for the next stroke. Using a car analogy, you can't expect a car to slow down and make a u-turn going 60mph in the same time it takes going 20mph.

If your GCC has the same over travel doing a .006" raster line as a does doing a 24" raster line then the overall machine engraving speed is consistent (consistently slow, imho). I say this because doing a .006" raster line is a short stroke, I would think the engraving head isn't even reaching it's full maxium velocity speed. Doing a 24" raster line gives the engraving head enough time to reach it's full speed so it should take longer to slow down and turn around. Try this same test using 100% speed. At higher speed settings, you'll likely see more of a difference in time.It's also VERY possible that GCC just designed their machines this way and most likely their code is different than ours(what my experience is based from). Proving that you'll never obtain consistent settings from one machine to the next.

You have mentioned 2 variables that could affect observed speed: length of the line segment and number of nodes (although they are closely related.) At higher speeds the acceleration and deceleration just before a node (change in direction) probably affects observed speed. I wish it didn't. At low speeds it shouldn't be such a factor.

Agreed, slower vector speed settings aren't affected so much by these variables.

At speeds of 10% and higher you are right. But if you use lower settings, the time required for acceleration and deceleration on the corners becomes almost insignificant. (I don't have a means to directly measure it.) I did your test on the Mercury, and it is almost exactly a 5X relationship below 10% speed, and even at 10% speed, it was 4.5X which is still not bad.

Wrong testing noted on my behalf. You should test this at a higher vector speed setting like 50% or higher to see the affects of vector acceleration and deceleration. You're results are likely to be more skewed.

understand this, and I trust that they will try to optimize, but I also feel that it is the designer's job to make all the hardware behind the scene "transparent" to the end user. By that I mean that they should provide a machine that moves at 10 inches per second when it is commanded to do so. Not move at 10 ips if it a square, and move at 8 ips if it is a circle, and move at 6 ips if there are a lot of nodes, all without telling me this is going to happen. I'm not suggesting that you can violate the laws of physics but if a setting is accepted, the user tends to assume it is being used.

When I say that I am trying to figure out the logic to my machine, it is because I want to be able to operate the laser effectively from the user point of view, which requires a logical and predictable interface. (I don't need to know the math behind the motion system.) But the laser does not always operate in the manner I expect. This results in too much trial and error in setting up a job and it makes it impossible to compare results with other users.

Asking for a machine to be consistent in speed regardless of the many variables is like to asking your Honda Accord to maintain 90mph on the highway regardless of weather conditions, elevation changes and whether the highway is thru the mountains of Colorado or the Arizona desert. At slower speeds it's easy and predictable, at high speeds alot of different factors kick in and that's when things get interesting.

I think all in all, laser manufacturers are doing the very best they can when it comes to making their equipment fast, efficient while maintaining quality. At least that's where I feel we stand. Your points as an end-user is well taken and as a consumer, understandable. I just think there's too much going on behind the scenes that most would understand (myself included).

sorry for all the car analogies, It makes sense to me so I assume it makes sense to most.

Interesting thread indeed. I'm just not sure if we'll ever find the answers you're looking for.

Richard Rumancik

06-20-2007, 7:49 PM

Do people who do a lot of rastering have to give up the right/left-most X inches of their work area?

Well, yes and no. I can only speak for my laser. I don't give up any of the 25" width which is the spec on the LaserPro. The area I am "giving up" is a zone that GCC never promised me in the first place. But a few years after I got my laser, it occurred to me that the carriage could actually address about .80 or .90" beyond the 25" "plotting zone" on both the left and right. So it can go farther left (over the rule) and farther right (past the 25" mark) enclosing an area about 26.60 (or a bit more) in width.

It seemed to me that there was no reason why this zone could not be put to use for vectoring, as you don't use left/right overtravel for vector cutting. I think they could have moved the rule over and called this a 26.60" machine (with 25" rastering zone). The customer would have been able to cut a wider piece. Of course it means that you need separate limits for rastering and vectoring built into the driver but this should be straightforward. It also would need to warn the user if the raster is clipped by having inadequate margin.

Is the extra inch and a half worth it? If I was a manufacturer of lasers, and could give and extra 1.6" vectoring width "for free", I think would do it. I know, not everybody needs/wants it, but if the cost is zero to negligible, why not? There are lots of users on the forum doing different things with their laser. Maybe there is some technical limitation I don't see. I proposed this to GCC but I don't think they really understood what I was suggesting.

I don't know how the overtravel is positioned on other machines relative to the legal plotting area.

Mike Null

06-20-2007, 8:25 PM

On my machine that area is designated for ramp up/down when rastering and the laser is not operating at full power in that zone. I've had a couple of instances when I was engraving against the ruler and the engraving was clearly at lower power out a quarter to 3/8" from the ruler.

When I contacted tech support they quickly advised that engraving jobs (that is the image itself) should not be positioned so close to the ruler. It does not ordinarily pose a problem but I am at least aware to be on the lookout for it.

I doubt that I could vector over the ruler but haven't tried it.

Richard Rumancik

06-20-2007, 9:09 PM

I doubt that I could vector over the ruler but haven't tried it.

I didn't mean to imply I could actually vector cut on the ruler. This is a prohibited zone because of firmware. But I can move the red-beam into that zone. So it would seem that it would be possible (with design changes) to vector cut in that zone. Seems your machine handles it differently - perhaps more like what I was suggesting to GCC.

Richard Rumancik

06-20-2007, 9:18 PM

. . .What you really want is for the laser to be consistent in results: if it cuts through material X at power setting P and speed setting S when cutting a square, it should still do so with the same settings if you are cutting a circle or any other shape. Further, you should get roughly the same results if you change to P*2 and S/2 (within the linearity limits of the material being cut).

I agree with your comment 100%. And I agree changing to P*2 and S/2 will ideally give the same results because the "energy delivered per unit length" should be identical. My machine seems to behave this way fairly well.

But I thought my statement supported yours. If I want to cut a circle and I optimize power setting to "just get through", and then cut a square with the same settings, there's a good chance the square won't cut all the way through because it plotted faster. When I say I would like to see all shapes cut at the same speed it is with the idea of achieving the same cut quality. Either I am wasting laser time, or using excessive power in some areas (with the possibility of degraded cut quality.)

If your machine is self-consistent in the manner I describe above, the first simply should not be a problem. If it isn't, then I agree that someone at the factory hasn't their engineering homework, but I suspect if a given brand/model was subject to grossly inconsistent results due to shape/orientation of cut objects, you wouldn't need a computer to hear the customer base complaining about it.

Well, I haven't done specific testing on this - maybe some day - in the heat of the battle you still have to get the job out. Then you go on to the next job and forget about the issue for a while. But I do cut some wood kits from 1/16" basswood and this job comes to mind. There are a lot of unusual shapes in the same panel. When I cut the more intricate shapes, I have to change the color, and set special parameters for that color. Why should I have to do this? I would have preferred that the laser held the same speed on all shapes, so I could do a 1" cutting test, set one speed/color, and be done with the setup.

As Rodne suggested in this thread, there are a lot of things you can do to optimize a job. Many times we are probably cutting slower than we should because there are a few "problem areas". And for large runs it pays to optimize. For small runs it might not. You slow it down till it cuts through everywhere and hope it doesn't ruin the fine detail where the laser gets bogged down. This isn't ideal.

Richard Rumancik

06-20-2007, 10:16 PM

Richard, If your GCC has the same over travel doing a .006" raster line as a does doing a 24" raster line then the overall machine engraving speed is consistent (consistently slow, imho). I say this because doing a .006" raster line is a short stroke, I would think the engraving head isn't even reaching it's full maxium velocity speed. Doing a 24" raster line gives the engraving head enough time to reach it's full speed so it should take longer to slow down and turn around.

Well, I will have to look at this again. It was my impression that rastering would not start until constant velocity is achieved. I visualized a ramp up, constant velocity zone, and ramp down at the end, with plotting occuring only during constant velocity. I need to think about this and check my laser again . . .

[Re: A 2" square cut at 10% speed will not take 5X less time than a 10" square cut at 10% speed.]

You should test this [relationship] at a higher vector speed setting like 50% or higher to see the affects of vector acceleration and deceleration. You're results are likely to be more skewed..

Well, I can't really. . . My laser will vector cut straight lines at a maximum speed of 29% of spec'd (raster) speed which is about 11 ips. If I tell it to cut faster, like 50%, it will accept the value but still will only cut at 29% (11 ips.) I couldn't cut much at this speed anyway due to power limitations. I agree that results will be more skewed the higher the speed. And this is not a good thing . . .

Asking for a machine to be consistent in speed regardless of the many variables is like to asking your Honda Accord to maintain 90mph on the highway regardless of weather conditions, elevation changes and whether the highway is thru the mountains of Colorado or the Arizona desert. At slower speeds it's easy and predictable, at high speeds alot of different factors kick in and that's when things get interesting..

Well, it is at the slow vector speeds that I am most concerned - let's say 20% (8 ips) and below for a 30 watt machine. I'm fairly satisfied with rastering behavior.

If you set your cruise control on your Honda to 40 miles an hour (not 90) then is it unreasonable to hold 40 in the desert and in the mountains? If Honda told you the cruise doesn't work on hills - is this acceptable?

Interesting thread indeed. I'm just not sure if we'll ever find the answers you're looking for.

But I'm going to keep trying . . .thanks for participating.

Rodne Gold

06-21-2007, 12:51 AM

You will find the GCC acts differently at speed of less than 3% than above that in terms of vector cutting
The later models than the mercury also have smartact and skipwht , and both of these also affect raster speeds and quality
Added to that ,there is the ramping options as well as the optimisation options that affect speeds.
Actually trying to work out some table based on graphic size and trying to predict a speed before running a job using that table is pretty much mission impossible.

Im a little confused as to what is being asked for?
Are you wanting total linearity so you can extrapolte time from one job to another ?
Do you want to reduce cycle times?
Ultimately the idea is to run as short a cycle time that you can with the most acceptable quality.
We probably dont do this as we have a higher cut off point re quality than our customers do , IE we exceed their expectations.

Peck Sidara

06-21-2007, 10:36 AM

Im a little confused as to what is being asked for?
Are you wanting total linearity so you can extrapolte time from one job to another ?
Do you want to reduce cycle times?

A little of both Rodne. Some predictabilty and consistency in both low and high speeds for both raser and vector is what I believe Richard's looking for.

With this information, Richard can use what is known and apply it to new materials, thicknesses and share the information with other users of the same and different manufacturers. He's looking for some data on other equipment as well to see if we can somewhat standardize power/speed settings when members ask about lasering certain materials.

Rodne Gold

06-21-2007, 11:46 AM

I dont think the laser guys have witheld anything in these areas. The settings in my manual were way out anyway , so even my machine mnfgr got it wrong.
However there are guidelines one can follow when trying to predict times (I dont think a time based model of pricing works properly or well) and work out settings.
The nature of the material and thickness are paramount , so the first thing to do is bone up on the materials and its properties.
I believe , albeit a time based model doesnt work for pricing , that in terms of strategy , it does , thus I always use 100% power or speed and get the operation to be as fast as possible.
So I always start from a shortest cycle time point of view.
If its rastering , generally I will start at 100% speed and vectoring at 100% power.
The melting point relative to acrylic will give me an indication of total power needed to engrave and cut as will the thickness. Composition too will influence my setting decision and the edge properties will influence my pulse freq per inch.
Of course how material reacts to a laser is vital , for example glass frosts by realtively uncontrolled fracturing , acrylic frosts by vaporization. On of the most important things is to try mininmize the HAZ - heat affected zone , IE the area around the spot that changes due to being heated so a lot of strategies are formulated with that in mind.

So one can make educated guesses as to where to start. We use 3mm Acrylic as a base line product. We know exactly at what speed and power it cuts and engraves.
What I did when we first got our lasers was get samples of just about anything and tried lasering or cutting it , even now I look at stuff and think "Surely i can engrave and cut it and there must be some application" and I bung it in the laser and try. I keep a record of the settings and how it behaves.
Problem is , the material i get in my part of the world might be different to yours , like for example , my mdf might contain a more laser resistant binder than yours , so what strategy I adopt on mine , might not work on your mdf.

The big problem is that a lot of laser engraver users are not that technically inclined and werent informed to to use the tool effectively , you have to do a lot of your own experimentation. Also new users without the material knowledge or baselines to work from feel as tho they are floundering in the deep end , not knowing where to start.
A better idea is not to have a materal settings data base but a materials/laser behavior data base where the properties and suggested strategies are listed , adjuct to that , one could add another listing of settings that have worked for owners of particular machines.

Mike Mackenzie

06-21-2007, 1:25 PM

Rodne,

The only other thing that will cause issues with this is the actual power of the laser tube on the table. Each tube will vary with this output power / peak power and most operators do not have a method of testing this.

Just another wrench to throw into the equation.

Richard Rumancik

06-21-2007, 11:26 PM

I appreciate all the comments in this thread; I will try to respond to a few when I get a chance. In the meantime I came across an Epilog 36EXT manual and found this:

"Because there are many factors that influence the time it takes to engrave or cut a given image, the Speed settings were designed to be reference numbers only. The Speed setting scale of 1% to 100% is not linear – i.e. 100% speed will not be twice as fast as 50% speed. This non-linear scale is very useful in compensating for the different factors that affect engraving time, but using speed to predict a jobs engraving time is not practical.
The Power settings are linear – i.e. 50% power is half as much as 100% power."

So, it seems they have re-defined the meaning of "%" - I have never heard of non-linear percentages before. And they explain why they did it: "This non-linear scale is very useful in compensating for the different factors that affect engraving time."

They obviously had what they thought was a good reason, but I can't understand the explanation. Can anyone please explain this statement to me? Peck?

Peck Sidara

06-25-2007, 1:15 PM

Because there are many factors that influence the time it takes to engrave or cut a given image, the Speed settings were designed to be reference numbers only. The Speed setting scale of 1% to 100% is not linear – i.e. 100% speed will not be twice as fast as 50% speed. This non-linear scale is very useful in compensating for the different factors that affect engraving time, but using speed to predict a jobs engraving time is not practical.
The Power settings are linear – i.e. 50% power is half as much as 100% power.

Richard,

That's a pretty good find you got there.

What does it mean? It basically means what I've been saying. There's alot of variables in our driver/firmware/motion control design that effects engraving time. Because of these "variables", don't go using the speed parameter to determine engraving time. The speed scale of 1-100% is not linear, the power scale of 1-100% is linear. A non-linear % scale just means there's some math algorithm and other factors going on in the background. My cars stock radio volume goes to 50, when I have it at 25, it sure doesn't sound like it's 1/2 as loud. Non linearity in its simpliest form.

What are these variables? From my non-engineering perspective it's acceleration, deceleration, laser control, load on components such as motors, belts, driving weight and lastly; it probably has something to do with a flux-capacitor.

EVERY laser manufacturer uses some type of motion control to control how their axis's move in raster and vector mode. This motion control has alot to do with the overall raster & vector quality. Since EVERY laser manufacturer does it differently, it becomes difficult to have "settings" for specific materials that will apply across the board in both raster and vector modes.

If my car radio analogy doesn't work as a non-porportional % scale example, please refer to your original posting;

I found that on my machine, this is approximately true for rastering. But for vector jobs, what I found seems very far from linear.

Conclusions (applicable to the Mercury LaserPro):

1. For raster plotting, the observed speed is proportional to the % speed setting (approx.)
2. For vector cutting, the observed speed is not proportional to the % speed setting.
3. For vector cutting, observed speed is dependent on the shape being cut.

Conclusions 2 and 3 makes me think that comparing settings for vector cutting on different lasers has very limited validity. The settings may allow one to reproduce results on their own machine, but that is about all they can do.

Tom Cole

06-25-2007, 6:40 PM

All other settings being equal, on my epilog mini18, using vector cutting, if I cut a 1 inch circle defined by 100 nodes, it takes a LOT longer than if I cut a circle defined by 4 nodes. I always select all nodes in an object and use "reduce nodes" set to 7 in coreldraw. this setting reduces the number of nodes dramatically while not effecting the object shape that much.

Richard Rumancik

06-25-2007, 11:09 PM

. . .if I cut a 1 inch circle defined by 100 nodes, it takes a LOT longer than if I cut a circle defined by 4 nodes. .

Thanks for this info - it helps me understand how other machines perform. I find similar results on my LaserPro. I have tried to use the offset command in Corel to enlarge a part for kerf allowance on inlay experiments - but then the laser slows down because of the extra nodes, and tends to overburn. If the laser could achieve constant speed independent of shape, then the extra nodes should not cause a problem.

I always select all nodes in an object and use "reduce nodes" set to 7 in coreldraw. this setting reduces the number of nodes dramatically while not effecting the object shape that much.

Yes, for some things the "reduce nodes" command works fine. But for inlay, the "reduce nodes" will just get you into trouble, because it distorts the shape enough to make for a poor fit.

Rodne Gold

06-26-2007, 2:11 AM

Upgrade to X3 , it does not have the zillion modes when offsetting problem.
The laser has to interpolate what the software feeds it , it is not the lasers fault that a complex item with a million points takes longer , it deals with the info it has.
A tip when reducing nodes is to copy the item and paste it on itself , reduce the nodes on one of these and you can plainly see how the reduced node shape differs from the original.

Richard Rumancik

06-26-2007, 11:45 PM

Rodne - you are right, I guess I have to bite the bullet and get X3. I found out that deleting nodes does not work for inlay by doing exactly what you suggested. The gap between the original shape and the offset shape should remain constant as you reduce nodes but it does not. Which means that the inlay piece would not fit uniformly.

I suppose that if there are excessive nodes then I could forgive the laser for not keeping up. But even with small numbers of nodes the speed goes way down.

Richard Rumancik

06-27-2007, 12:23 AM

RE: ". .100% speed will not be twice as fast as 50% speed. This non-linear scale is very useful in compensating for the different factors that affect engraving time. . " (Epilog manual)

What does it mean? It basically means what I've been saying. There's alot of variables in our driver/firmware/motion control design that effects engraving time. . .

Peck, I was stepping back a bit here in my question. First, let's agree that "engraving speed" is not the same as "engraving time". Engraving speed is the speed of the carriage. Engraving time is the time it takes to complete the job.

I understand what they are saying in the first sentence of the Epilog statement. I don't understand the second sentence, where they attempt to support the decision to make the set %speed vs. actual speed non-linear. That is what I wanted explained; maybe I wasn't clear . . .

You are suggesting that it is physics that made the %speed vs. actual carriage speed inherently non-linear but I don't agree here. The non-linearity of the "speed function" is not a result of physics, but a conscious decision of the designers. The speed command is "mapped" to the motion control circuit in a prescribed way. When a "25% speed" command is sent to the driver circuit, it has to know the appropriate current to supply to the motor. It could have been mapped so the end result was linear, non-linear, parabolic, exponential, step-function or whatever relation they chose. In the end, they chose some type of non-linear relationship, because they thought it would help the user in some other way (the second statement indicates that non-linearity is very useful). But that is not a clear explanation to me. How is it useful to the user?

LaserPro seems to have chosen a non-linear/step relationship; it cuts off above 30% speed (i.e you can't command a vector speed above 30%. If you try, it will use 30% anyway.) I suspect ULS has a step function cut-off too similar to LaserPro. Epilog doesn't seem to use the "cutoff method" and I think Epilog's method is more logical. But it means that machine settings will be quite different. I can elaborate if anyone is interested.

If my car radio analogy doesn't work as a non-porportional % scale example, please refer to your original posting . . .

I spent a long time trying to figure out what you were meaning here but am still puzzled. If you are saying "well, your LaserPro is non-linear like the Epilog", then I agree. That was my point - that Epilog was similar to LaserPro. Probably some other machines are too. But I am missing something here. By indicating that the Epilog (or LaserPro) behaved non-linearly was not intended to be a criticism of the products. (Frustration, maybe, because I can't find the explanation.) Both were observations, and I just wanted some insight as to why a designer would choose that route.

I thought it would be more logical if it was programmed in a linear manner. If this is incorrect, I would like to have someone tell me why linear is not a good way to program the machine (e.g 10% speed is half of 20% speed). If the Epilog or LaserPro decision was made to "help me" as a user, then the designers should be able to say exactly HOW the non-linearity is helping me - because so far I have no information that shows how it is advantageous.

What is my objective with the quest for linearity between the set %speed and actual carriage speed? I think it would be advantageous for everyone if we could better predict results, both for a particular machine and within the group of users. My hunch is that true linearity of the speed function would help make the laser performance more predictable and transferable. (If a particular material itself is non-linear, so be it. Then the user has to learn the characteristics of the material.) Everyone says that "settings lists" are not useful. Is there anything that can be done in the future to make the systems more comparable and predictable? Or should we just accept that we can't compare and can't predict and that is the way it will always be?

Rob Bosworth

06-27-2007, 10:55 AM

Another thing to consider is the size of the structure that is being moved. There has to be a difference in acceleration and deceleration needed to move a shorter and lighter motion system than to drive a larger and heavier assembly. Then what happens when that larger assy. is stopped suddenly and sent off in a new direction. If it is done too suddenly, you might see "ringing" or vibration in that assy. Ever noticed a little wavey line near the corners of a rectangle that you are laser cutting. Some of this comes from the bouncing.

An interesting note. A number of years ago, we had a customer who owned machines manufactured from 1992 through 2005. His small 17" X 11" machine, 16.5"/ sec. would do a very detailed vector engraving job in 6 minutes and 13 seconds. As newer machines came on the market, he bought them to increase his throughput. He bought a machine, 32" X 18" that traveled ~ 90" per second. (And his newer machines had more laser power to take advantage of the higher feeds and speeds.) The fastest he could get these newer big machines to do the same engraving was just under 14 minutes. It made no sense to any of us, except maybe the systems' manufacture. They finally admitted that they had to "dumb" down the machine to keep the X axis arm from "ringing". So they slowed the acceleration and deceleration down, to minimize the harmonics that showed up in these bigger machines. These machines still moved at 90"/ second, but they took much longer to get there and much longer to slow down.

Another "black science" that we have experienced is how manufactures come up with their travel speeds. I worked for a company that made 5' X 10' and larger machines with up to 9 axis of movement. When we started, we advertized the machine as having lineal travel speeds of up to 300"/ minute. Within a couple of years, we were advertizing lineal travel speeds of over 1000"/ minute. Nothing had changed, except for the way they measured the travel speeds. The use to program the machine to move a certain distance, time the movement, then calculate out the distance traveled and how long it took to cover that distance. Then they found out that other manufactures were doing a similar test, only they were adding other axis movements into it, and the formula gave up some spectacular numbers. Something to do with doing a 45 degree movement, then extrapolating the numbers to show the movement actually moving through two 90 degree legs. More distance traveled, theoretically, at the same amount of time. I am not sure how our laser systems manufactures are calculating their numbers, but these machines are certainly fast.

Richard Rumancik

06-27-2007, 11:48 AM

Thanks for that info, Rob. It gives me a stronger argument that laser manufacturers should really spec TWO different speed numbers, one for rastering and one for vectoring. It appears that when they spec a maximum speed, it is with rastering only, in which case the carriage can achieve its maximum speed because (1) the motion is in a straight line (2) there is only one axis involved so there is minimal math required (3) there is less mass/inertia in moving the x-axis alone than in moving in an x-y direction and (4) overtravel is permitted in rastering (but not vectoring).

It is my observation that the older LaserPro, which specs 42 ips speed, vectors at about 11 or 12 ips maximum for simple rectangular shapes, and maybe 8 or 9 for circles. If it tried to vector at 42ips, the vibration, overtravel, ringing etc would cause trouble. I think it would be much easier technically to increase raster speed in a laser than vector speed. So they did this in the model you referred to but they could not get an improved vector speed.

By only quoting raster speed as the "speed" of the laser, it is easy to confuse the customer. Most people would have expected an improvement going form a 16.5ips to a 90ips machine. But these are both maximum raster speeds and have nothing to do with vectoring.

I have no reason to believe that the maximum speeds they are quoting during raster are not real. But if you take overtravel into account, you will probably find out that a 42ips machine really rasters at about 35ips or less, because of the turn-around time.

Rodne Gold

06-28-2007, 3:27 AM

Rob , Same happened to us with our large over head router cnc.engraver.

We bought a top of the range Tekcel and had to slow down dramatically over speeds we used on our Isels when engraving as the whole spindle/z- axis was so heavy that despite having servo motors and blindingly fast travdrsing rates , detailed engraving jobs were horrificaly bad at high speed due to inertia issues.
Richard , it gets worse re prediction - the Spirit actually has 4 levels of vector "quallity" and each of these affect speed. The bugbear with most lasers is the fact that when they scan on the X axis , the use the same speed when engraving as scanning - ie they dont speed up to max when there is white space , hence we use the cluster function a lot , that is , tell the laser to complete an element on the left and then do one on the right instead of doing both those elements in a single "pass"
This works well when there is a lot of space between the 2 elements.
I dont think vector cutting can be speeded up regardless of velocities and ramping as most materials can only be cut at a certain rate. If the motion system of the machine cant actually deal with the fastest rate of cut and you have to slow it down cos of quality issues , welll , that's another issue. What does work well is the full optimisiation routine where te laser works out the shortest pathes over the whole job , especialy useful in cases where you doing vector cutting and the laser is jumping from one thing to another.

At the end of it , when buying machiner we apply the 1/2/x2 rule. Halve advertised claims re speed and thruput and double advertised running costs , if the purchase still make sense , go for it.

Dean Carpenter

06-28-2007, 3:59 AM

Rodne Gold

06-28-2007, 6:22 AM

The problem is that advertised speeds dont equate to real world speeds. Or if they do , the output at the rated speed is well nigh unuseable.
Apart from that , I have never seen a G figure advertised for lasers.

Richard Rumancik

06-28-2007, 10:25 AM

Richard , it gets worse re prediction - the Spirit actually has 4 levels of vector "quallity" and each of these affect speed.

Rodne, do you have any idea of what parameter they are adjusting when you select different "quality levels". I don't mind the idea in principle as long as they say what is being adjusted when you change quality levels.

What I don't like is when manufacturers design the machines up to make it "easier" for the user, but then lock the user out of the settings. For example, the LaserPro (as you noted earlier) changes to CW below a setting of S3. That is, it ignores the ppi setting and turns the laser on continuously (at least, that is my understanding). I would like the option of deciding some things on my own. Defaults and pre-set conditions are acceptable as long as they allow the user to override them when they want. Also, I like to see some of the technical reasons behind certain decisions. I know not everybody is interested or cares, but for people who want to understand the laser better a "theory of operation" for the laser system (preferably written by the designers) would be a nice touch.

Richard Rumancik

06-28-2007, 10:48 AM

To determine the true speed of the machine, as a rule of thumb look at two things:

1) Top raster speed in metres or inches per second
2) acceleration: expressed in G-force

The higher the numbers the faster the machine.

These numbers will tell part of the story. But as Rodne noted, nobody in the laser world specs G-forces (maybe they do for machine tools.) Higher speed, as has been noted, doesn't necessarily mean higher throughput. If a manufacturer increases maximum speed but not acceleration, then the overtravel will be longer and there will be more lost time on turn-around. If they do increase acceleration, they may find problems with settling/vibration/ringing and not be able to maintain the original turn-around time as anticipated.

From what I have seen recently the easiest way to compare two lasers (for rastering) is to compare the time it takes to raster a solid black filled rectangle. Of course, this is simplistic, as it doesn't really allow for much of a quality comparison. (Maybe a detailed graphic with a heavy black border would be better.) But my point is that you probably need to compare overall throughput as opposed to just the numbers. The raster test will show how the max speed and max acceleration (G's) work together.