Inconsistent engraving problem ?

[SScnc]

I'm having a problem that is puzzling me, as you can see in the pics, the top half has horizontal lines but the lower part does not. I've read Tim's article here: http://diylaser.midwestlaserarts.com/20 ... n-gap.html regarding scan gap but I don't understand why part of the engraving is rough and part isn't.

I know the wood is crappy quality but I was just testing on it. The metal shown is powder coated aluminum and the horizontal lines are still there. I will be engraving on cast acrylic and powder coated metal and need to figure out how to get a smooth finish.

What's causing this and what can I do about it ?

Thanks

You can click on the pics to enlarge them a couple of times.
X Swing
Speed = 80
Power = 95
Gap = .05

[lasersafe1]

Please explain the power 95%, but images show 5 ,3, and 2 mA. Something doesn't fit.
Also does it do this in both unidirectional and x swing?
There is no need to ever take the Y scan gap below 0.1mm, so prep your image at 254dpi.

[SScnc]

Regarding the power levels, my machine has an over ride pot on it, before running the job I press "LASER" and adjust the pot while watching the ammeter.

I've only tried it in X Swing so far. I use Corel Draw X4 and to be honest, I don't know how to change the DPI when doing the artwork.

[SScnc]

WOW, is this forum now dead or is this just such a strange phenomenon that no one has ever seen it ???

[lasersafe1]

Forum not dead. Waiting for your answer from my last question. You owed us a few more tests to help you determine the true answer. I have seen this before, and in my experience it seemed to be due to a tube that was nearing end of life.

[twehr]

Sorry, I had not seen this thread until today, or I might have jumped in a little earlier.

Looking at your first example, I see an issue that I have had forever. Specifically, the varying density bands (some wide, some narrow). It is pretty obvious that this implies a variation in the power being delivered to the material. So the question is, "Why does the power level vary?" Unfortunately, I have never found a real answer, and certainly no fix, but I can make some observations based on your example as well as the hundreds of dollars of material wasted running tests trying to eliminate the phenomena.

John (lasersafe1) and I have discussed this many times. What I see is that you (we) get pretty good results when the image being engraved requires few on/off cycles in a single scan line. As the number of on/offs increase, the consistency of the output decreases. This leads me to believe that the tube is not cycling power levels quickly enough to keep up with the demands of the image.

Notice the rather broad section in the middle of your first image; it looks pretty good. Above that (and possibly below - not sure if the below is due to this problem of just a variation in the material), there is an increased number of on/offs and the consistency suffers. John (lasersafe1) and I have discussed this for hours and come to no verifiable conclusion about the cause. Here are some possibilities, however:

1. Controller is not able to keep up with the cycling speed - especially on fine detail photographs.

2. Power supply is sluggish on getting up to full (specified) power level.

3. Tube is sluggish getting up to full (specified) power level.

4. Material has inconsistent reactivity to the laser (inconsistent coating thickness, etc).

5. Coolant flow is inadequate to maintain a consistent tube temperature.

Certainly there may be other reasons as well, but let's start with these and see where we get.

I believe we can disregard possibility 1 right away. I only used the LightObject DSP, but the problem has been present while using 3 different controllers/models/versions. It is, I suppose, still possible that the issue is there since it is the same programmer for all 3 I tried, but having done many tests with the o'scope looking at the DSP output signals, I believe the controller is not the best place to spend our time looking.

Possibilities 2 is probably not it either. I have also used 3 different PSs with no affect on the results. Of course, most PSs are made by the same company in China, so a design fault there could have a broad-reaching effect.

Possibility 4 is where I was betting for a long time. The reason I say that is that I have yet to be able to do quality work on LaserTiles, where I know they take a standard tile and coat it with an otherwise clear coating that turns dark gray with exposure to the laser. I spent over $300 dollars on LaserTiles alone, during my testing. (BTW - Doing large wall murals in laser tiles is the whole reason I got into lasers in the first place, but to date I have been unsuccessful at achieving high quality results.) However, the effect is noticeable on other materials, as well, like coated metals. Even in wood, where charring tends to hide it, you can look closely and see the problem. So I believe possibility 4 is also a bust.

Possibility 5 was also my target for a long time. I have redesigned my cooling system several times and the results do not change enough to say that cooling is is the real cause.

That only leaves us with possibility 3. Like the PSs and cooling systems, I have tried several different tubes - actually it is 5 to be exact. The results are the same. But it is still a possibility - here is why. I believe (and I know I will be instantly corrected if wrong), the nature of the tubes we use does not lend itself to consistent rapid on/offs.

A. The tube's power rises and decays with each cycle. The higher the power level needed to achieve changes in the material, the longer it will take to rise and decay to/from that power level. The shorter the cycles (narrower spaces between ons/offs), the less time the tube has to rise fully or decay fully.

B. The longer the tube is on, at any power level, the higher the temperature in the tube; as the temp changes, so does the power output.

C. RF-excited lasers don't seem to have this issue. They are either on or off with little or no power/temp latency - they are essentially binary. Glass tubes, on the other hand, have an analog component to them during the rise/decay of power.

I have tested with varying the power level and found that the lower the power level, the less severe the phenomena. This makes sense. The time to cycle from off to 50% to off again is shorter than the time to cycle from off to 90% to off again. Unfortunately, reducing power requires a proportional reduction in speed. The reduction in speed can influence the temp of the tube as each cycle is longer, so reducing the power is not the ultimate answer.

Now, as I said, there are likely other reasons, as well, but I have (for my purposes, at least) eliminated the Controller, PS, Material, and Cooling systems. To me, everything points at the tube. I believe it is an inherent flaw in the glass tube design. If I had $2000+, I would buy a Chinese RF tube and try to prove it out, but I don't have the dollars to spend on a test that may not prove positive.

I know (and hope) many of you can say I am totally wrong and that you are getting perfect results with your glass tubes, but I know a lot of the effect is determined by the material and the image that is being engraved. I get good results sometimes. I just have never been able to get consistent enough results to be able to say I have found the answer.

If anyone has some useful input on other possible causes we would all love to hear them.

[Cre8ivdsgn]

There are a couple of other oddball possibilities, but a decision can be made rather easily:

1) Take whatever pulley circumference you have for the offending axis and multiply by two.
2) Now engrave a "gray" rectangle with the long axis parallel to the offending axis and a length equal to step one.
3) Look for any patterns that repeat twice.
4) Loosen the set screw on the drive pulley, move the pulley on the motor shaft, and retighten.
5) Rerun the engraving test. (Note, the test is worthless if you do not start at the same point every time!)

If the patterns change, the problem tracks to the pulley. Otherwise the problem tracks to either the motor, positioning system, or something else.

If you find any, they could be related to pulley issues (bore not centered with pitch diameter, or oversized bore) or they could be due to poorly manufactured stepper motors.

So if a motor is rated with a phase angle accuracy of +/-3% of one step (this is based on full steps per revolution), and you have a pitch diameter (yours probably differs from mine) of 1.452", you get roughly 0.007" per step and 3% of that is 0.0002". This is not something that will be too visible.

Swap X and Y motors and cables and see if the paterns remain the same. If they do remain the same, you are getting to the bearings or something potentially unrelated to X and Y. (Like if the Z is not perpendicular to X and Y in some way, changing the laster focus.) If the patterns change, you are potentially chasing a motor issue or possibly a combination of motor and pulley issue.

Also, I have seen several machines where the carriage attachment pulls one side of the belt out of parallel with the other. As the stage gets close to either end, the belt forms angles off of parallel that I think could easily account for additional positional error. These errors go to a minimum at the center of travel if X and Y are perpendicular to each other. I think you might expect to see bunched up lines near the extremes of travel and excessive spacing at the center of travel. I have always taken care to make the belts run parallel to each other on both sides of the pulley and so I haven't seen this on my own equipment. This is just an intuitive observation. I could be (and have been in the past!) way off base here!

I hope this makes some sense. The coffee wasn't so great this morning!

[lasersafe1]

Indeed it's tough. I would take a look at mechanics very carefully as the last post indicated. The oddness that Tim and I saw on the laser tiles seemed to me to only occur on the laser tiles because of the inconsistent depth of the laser sensitive glaze. Tim sent me tiles to try on my machine and the same oddness occurred. When I would run my test image on anodized aluminum or granite, it was perfect. I have not seen this problem to this extent in wood other than the occasional grain pattern showing through.

This is off topic, but still on topic... Perhaps someone on the forum has a nice rf laser and the LO DSP? Perhaps this person would be willing to test the white laser tile to see if the observations by Tim and I on his tile are related to tube lasers?

Another problem I see from the original post. I know my tube won't consistently fire down at 2 or 3 mA. Perhaps this is too low. Can you speed up the raster speed and go to higher currents of 8-10mA?

[BenJackson]

1. Controller is not able to keep up with the cycling speed - especially on fine detail photographs.

As far as I can tell the Epilog lasers use PPI when rastering as well. I thought it was to control power density, but maybe it answers your question about consistency: If the raster output is made of "spots" then there are no different on-times for different image areas. I suppose it might also eliminate the need for as much overscan in the rastering sweep.

I plan to try this in my LinuxCNC config and see how it goes. I started looking at this because I'm not 100% satisfied with how Rowmark LaserMAX is engraving with "stripes".

If the patterns change, the problem tracks to the pulley. Otherwise the problem tracks to either the motor, positioning system, or something else.

The original idler designs had a lot of wobble. My early vector cuts had annoying inconsistencies where slots in some areas of a cut would fit fine but be too small elsewhere. When I started raster engraving I made a ruler image and engraved it. The periodic pattern (wavelength equal to pulley diameter) was quite obvious. I ordered the new idlers from Bart and that cleared it up.

[cvoinescu]

I disagree with the possible explanations offered by twehr. The power seems consistent along X (where the laser switches on and off) and seems to vary along the Y. To me, it looks like either (a) some scan lines deliver more power and others less (so there's a low-frequency variation in the power of the laser that settles after a while), or (b) some scan lines are packed closer together and some spaced further apart, which would also vary the total amount of power delivered to the material. The latter could come from mechanical trouble, such as some stiffness or binding of the wheels of the gantry in that region, dirt on the rail, or maybe the gantry rubbing against something, or stiff cables offering non-uniform resistance.

Talking about twehr's possible explanations specifically, I think 1, 2 and 3 are not the case because they would cause banding or artifacts along X, not along Y. I guess a weird (< 1 Hz) instability of the power supply could cause the effect, but it seems unlikely. Bad contact that eventually heats up and improves? 4 is out because the banding is perfectly horizontal and not aligned to the grain of the material. Something like 5 could be the cause, though. If the coolant pump is on only while the laser is on, maybe there are bubbles or dirt in the coolant, which randomly disperse some of the laser's power, until the coolant circulates a little and flushes them out, which is when the situation improves. Still, the mechanical explanation (rough Y movement causing non-uniform scan line density in the banded region) seems the most likely to me.

Disclaimer: large grain of salt required due to my limited actual experience with lasers.