Epilog Legend 32 vectoring problem. Any ideas?

[Bill George]

I suppose something like that is possible but I can't really think of how. The image appears fine in the program. The way I typically do these types of plaques is set up vector lines in RhinoCAD, then fill the portions to be shaded with a solid hatch, set those hatches to the desired color for rastering, and assign which vector lines to be printed or not. I usually vector most of the lines for better contrast. Then I print the file directly in Rhino.

Andrew when your troubleshooting, anything is possible and spent most of my life troubleshooting and repairing. I might add I was paid very well for my skills. The preview tells you nothing. Try another file because Scott has a point.

[James Rambo]

I have the same machine but 60 watt.
mine was manufactured in 2002. I have had a couple problems over the years. I called Epilog expecting to need parts and found they are excellent with customer and tech service. I had a similar problem they said to remove the wire strip on top of the gantry and flip it over. it plugs in both end either orientation. They sent me a new one even though it worked flipped over.

[Andrew Stow]

I have the same machine but 60 watt.
mine was manufactured in 2002. I have had a couple problems over the years. I called Epilog expecting to need parts and found they are excellent with customer and tech service. I had a similar problem they said to remove the wire strip on top of the gantry and flip it over. it plugs in both end either orientation. They sent me a new one even though it worked flipped over.

Thanks, I will try that next!

I had already removed the old encoder. Got the new encoder today, so I went ahead and installed that.

It's fairly simple and straightforward using the datasheet supplied by Mouser. Only tricky part is making sure your allen wrench is actually turning the set screw on the collar of the encoder disc. Before you snap the top half of the encoder onto the base, make sure your allen wrench is engages by flipping the encoder over so you can see the encoder disc and collar and see the set screw advancing or retreating. Mine came with the correct allen wrench (.035").

So once I had the new encoder on, I ran some older files to (sort of) test if the problem was being caused by a corrupt file. Two different files worked ok. Not perfectly, but vectors have never completely perfectly matched the edges of all rastered portions with 100% accuracy. But it was still pretty good and about like I've expected from this machine over the years.

So then I went back to previously saved versions of the file I was having trouble with... and boom, same problem as before. In exactly the same spots. Rhino has an "incremental save" function and every last iteration of the image had the same problem!

I guess my next step will be James Rambo's suggestion.

And failing THAT, I'm going to try to manually redraw the lines from scratch, only using the previous lines as a guide (no snap-tos), then rehatch the portions to be shaded and save it all on a completely different usb drive and see if that fixes it.

Thanks for all the help, guys. And yes, Epilog tech support is great. I really cannot complain too much about this machine, either. About the only thing it's missing that I really want is a vacuum hold down table. It's never had a recharge and I bought it at a government auction a few years ago and have used the fool out of it since then. No telling how many hours on the tube. In an un-airconditioned shop in South Texas, no less. Epilog stuff is expensive but it is extremely sturdy and well-made. At least the old legend 6000 series is, anyways.

edited to add: It's definitely not file-specific. Looking more closely, I see some trouble spots on these other images I ran that show the same problem. Just not quite as dramatically.

[Andrew Stow]

I have the same machine but 60 watt.
mine was manufactured in 2002. I have had a couple problems over the years. I called Epilog expecting to need parts and found they are excellent with customer and tech service. I had a similar problem they said to remove the wire strip on top of the gantry and flip it over. it plugs in both end either orientation. They sent me a new one even though it worked flipped over.

When you say flipped over, do you mean swapping it end for end? It won't fit simply rolling it over.

If end for end, what is it plugged into down on the optics side and how do you remove it? It looks like I'll have to remove the plate that the lens/mirror assembly attaches to, but I'm afraid if I do that, the mirrors will lose alignment.

[Kev Williams]

Since the encoder didn't solve the problem, I wonder if your Epilog is getting funky info from your Rhino? The basis for my wondering, is that every so often when I open a DXF file I created my self in Corel or Gravostyle, upon opening the files in the opposite program I'll get offset curves and flat-spots on circles-- sometimes other anomalies, but they're rare. And this NEVER FAILS to happen: whenever I export from Gravostyle as a DXF, my EZcad (and ONLY Ezcad, it's fine in Corel!) will import it in 'pregnant' format...

As in Corel:
pmcorel.jpg
Exported from Croel as dxf opened in Gravostyle:
pmg5.jpg

Exported from Gravostyle as dxf and opened in EZcad:

pmez.jpg

I think this is a 'polylines/lines & arcs' thing, which I don't much understand, but regardless, EZcad arcs many of the straight lines quite radially! But the same dxf file opens just fine in Corel...

My other problem is harder to replicate, and I don't have any screenshots, but that issue more resembles YOUR problem, mis-shaped curves and offset start/stop points... What puzzles me is that your bad file(s) look good on the screen. However, could be just the machine itself is getting an 'edited' version.

Do you have access to Corel? Or, is it possible to export files out of Rhino as a DXF or EPS? You can upload them here, and we can open them and see what happens. I'd be curious to upload your file, and if it looks good in Corel re-export it as a dxf, upload it here and have you try THAT file...

[Andrew Stow]

Since the encoder didn't solve the problem, I wonder if your Epilog is getting funky info from your Rhino? The basis for my wondering, is that every so often when I open a DXF file I created my self in Corel or Gravostyle, upon opening the files in the opposite program I'll get offset curves and flat-spots on circles-- sometimes other anomalies, but they're rare. And this NEVER FAILS to happen: whenever I export from Gravostyle as a DXF, my EZcad (and ONLY Ezcad, it's fine in Corel!) will import it in 'pregnant' format...

As in Corel:
pmcorel.jpg
Exported from Croel as dxf opened in Gravostyle:
pmg5.jpg

Exported from Gravostyle as dxf and opened in EZcad:

pmez.jpg

I think this is a 'polylines/lines & arcs' thing, which I don't much understand, but regardless, EZcad arcs many of the straight lines quite radially! But the same dxf file opens just fine in Corel...

My other problem is harder to replicate, and I don't have any screenshots, but that issue more resembles YOUR problem, mis-shaped curves and offset start/stop points... What puzzles me is that your bad file(s) look good on the screen. However, could be just the machine itself is getting an 'edited' version.

Do you have access to Corel? Or, is it possible to export files out of Rhino as a DXF or EPS? You can upload them here, and we can open them and see what happens. I'd be curious to upload your file, and if it looks good in Corel re-export it as a dxf, upload it here and have you try THAT file...

I've seen the problem you're talking about. Usually happens if I try to save something as a DXF in the wrong type. Like polylines vs 2004 polylines or something.

Yeah, I don't know if it's a problem with Rhino or what. It only just started doing it! Maybe there's some settings in Rhino I inadvertently changed? Or maybe my install of Rhino is corrupt?

I tried Rambo's suggestion but no dice. It's definitely a Y axis (flying lens/mirror assembly axis) problem. The MOST trouble it has is in vector mode tho.

I will try exporting the file where the problem is most noticeable and uploading it here.

[Kev Williams]

I've had broken wires to my stepper motors happen with 2 machines, and both machines twice- If your machine was messing up randomly I'd think something like that could be wrong, but yours isn't random. Aside from a software issue, the only other thing I can think of is a bad stepper. I've replaced a couple of 'tired' steppers over the years, but they've never been servo's, just basic steppers. But, if a stepper or servo has a bad spot inside for whatever reason, it sounds logical to me that it may repeat bad activity.

However, there's an easy way to test for that too, which you may have done already, which is simply to engrave the suspect file in a different location, such as move the engraving layout 1/2" down and to the right, that would move any bad spot the stepper is running to a different place on the layout... if the bad spot changes places, bad motor, if the bad spot remains in the same place even after moving the layout, then it stands to reason the stepper is doing what it thinks it's supposed to do

[Andrew Stow]

I've had broken wires to my stepper motors happen with 2 machines, and both machines twice- If your machine was messing up randomly I'd think something like that could be wrong, but yours isn't random. Aside from a software issue, the only other thing I can think of is a bad stepper. I've replaced a couple of 'tired' steppers over the years, but they've never been servo's, just basic steppers. But, if a stepper or servo has a bad spot inside for whatever reason, it sounds logical to me that it may repeat bad activity.

However, there's an easy way to test for that too, which you may have done already, which is simply to engrave the suspect file in a different location, such as move the engraving layout 1/2" down and to the right, that would move any bad spot the stepper is running to a different place on the layout... if the bad spot changes places, bad motor, if the bad spot remains in the same place even after moving the layout, then it stands to reason the stepper is doing what it thinks it's supposed to do

Thanks Kev, I tried your suggestion. And this time, I watched it through the whole program.

It rasters flawlessly. In raster mode, it copied the image exactly as it is in the program.
When it started vectoring, that's when it had problems. So it is definitely just a vectoring issue.


What I did for this latest test is I left the logo portion in it's original position without altering it, then copied it and moved the copy to the top left corner and rotated it 90 degrees. The errors have moved! Note that the vectoring on the "S" and "E" follow the proper line now and the outline around "cola" is much improved. So this tells me it probably isn't a corrupt image file.

While watching it vector the outline of the "E" of the logo in the original position (bottom logo), I noticed it would start the vector line on the legs of the E halfway in the middle of one side of the square and then go around. But when it got to the last turn, it didn't quite travel far enough in the X-axis (horizontal) before making the turn to complete the box. The effect is a box that looks like a tetris piece. Which would seem to isolate the problem with the X-axis. I was 99% sure it was a problem with X-axis but this definitely confirms it.



I tried exporting as DXF to upload here, but I cannot get it small enough to fit in the sawmillcreek size limits. Smallest I could get it was about 200kb with the hatches deleted saved as 2004polylines dxf.

20180630_153509.jpg

Going by the logic you've expressed Kev, I'm leaning towards a problem still existing with the machine. Maybe it's a problem with the servo itself like you suggested? I sure hope not. 300 bucks from epilog. I'm not sure I want to sink another 300 bucks in a machine this old. I tried looking for a substitute using the part numbers on the motor by couldn't find anything exact. Possibly a substitute could be used but I'd have to know the voltage, bolt pattern, degrees per step (or whatever the equivalent value is for servos), number of wires, etc...

Worst case at this point is I won't be able to use the vector mode anymore with any reliability. Which stinks, as the vector mode is really nice for doing 3D text or just generally increasing contrast between the edges of graphics and the unmarked wood.

What I don't get is, if it's a problem with the motor, wouldn't it have a problem both rastering and vectoring?

[Kev Williams]

When vectoring, the steppers are constantly tracking the changing XY coordinates, when rastering there's only 4 coordinates: extreme right X, next Y, extreme left X, next Y... this pattern repeats until the bounding box changes, then a new pattern starts with 2 new X and 1 Y coordinate. The full sweep of the laser means the stepper will simply bypass its bad spots.

Based on your pics, it appears the erratic engraving is occurring in the X axis. And if the TEXAS was also rastered (it appears so), then it IS doing it while rastering too- the notches at the top of the T, bottom of the E, bottom of the S...

But here's the thing, what the stepper is doing has nothing to do with the laser's firing, and those notches wouldn't be there if the laser had fired like it should, instead it mimicked the vector problem exactly..

Ergo-- since the steppers aren't going to the right places, AND the laser isn't always firing in the right places, my money says it's NOT a mechanical issue, but rather a 'data' issue; software problem, or could be a cable connection problem, controller problem, loose plug connection on the control board or elsewhere. And over the years, bad connections have been the cause of 90% of any problems I've ever had...

[Andrew Stow]

When vectoring, the steppers are constantly tracking the changing XY coordinates, when rastering there's only 4 coordinates: extreme right X, next Y, extreme left X, next Y... this pattern repeats until the bounding box changes, then a new pattern starts with 2 new X and 1 Y coordinate. The full sweep of the laser means the stepper will simply bypass its bad spots.

Based on your pics, it appears the erratic engraving is occurring in the X axis. And if the TEXAS was also rastered (it appears so), then it IS doing it while rastering too- the notches at the top of the T, bottom of the E, bottom of the S...

But here's the thing, what the stepper is doing has nothing to do with the laser's firing, and those notches wouldn't be there if the laser had fired like it should, instead it mimicked the vector problem exactly..

Ergo-- since the steppers aren't going to the right places, AND the laser isn't always firing in the right places, my money says it's NOT a mechanical issue, but rather a 'data' issue; software problem, or could be a cable connection problem, controller problem, loose plug connection on the control board or elsewhere. And over the years, bad connections have been the cause of 90% of any problems I've ever had...

I thought it might be the raster too at first, but I watched it run the program. Those notches on T, E, and S were caused solely by vectoring. When it rastered, the image was flawless. The notches is from new lines caused during vectoring. I'll do the image again in raster only so you can see what I'm talking about.

Good idea checking the connections on the board! Will try unplugging and replugging some wires on the mainboard as well.

[Andrew Stow]

Checked connections on main board... appears to all be in order.

Ran program again, raster only. The raster looks fine. 20180701_031232 (copy).jpg

Question: With the machine turned off (of course!) when you go to move the x or y axes back and forth, like for lubing the bearings or what have you, should there be much resistance on the X-axis? I notice there's kind of cyclical resistance on the x-Axis. As I move it from left to right it will approach a point of increasing resistance, over come it, then have more resistance again as I move it further right. I can't remember if it was like this prior to when I noticed this problem or not.

[Andrew Stow]

Looking on Moog's literature about the C13 series servo motors, looks like they do use graphite brushes. I wonder if a dying/worn out brush would cause this kind of problem. Hm.

[Kev Williams]

When my GCC (only machine I have with servos) is off, there's zero resistance. When ON and the gantry is unlocked, there's some resistance. But none when powered down--

Back to your pics

I've reproduced the previous pic- yellow arrows show the notches, which I assume was part of the raster engraving- The green arrows show two places on the S where the vector and raster don't line up, the bottom arc is slight, the inside arc is off quite a bit. But the missing color in the notches is why I presumed rastering is an issue too...
tx.jpg


And under close scrutiny, your newest pic does show a reasonably bad raster error, in the area where the star and the S intersect-- Corel paint's 'auto adjust' actually changed the color of the pic, but it highlights the issue, where the horizontal line of the star makes a 3-sided rectangle up into the S and is not engraved.
tx2.jpg
--It's a simple 2-coordinate error, take the 2 corner points out and it would be a straight line... again, this error (IMO) takes the steppers and other mechanicals out of the equation, because aside from the fact those 2 points shouldn't be there, the laser engraved it perfectly...

[Andrew Stow]

When my GCC (only machine I have with servos) is off, there's zero resistance. When ON and the gantry is unlocked, there's some resistance. But none when powered down--

Back to your pics

I've reproduced the previous pic- yellow arrows show the notches, which I assume was part of the raster engraving- The green arrows show two places on the S where the vector and raster don't line up, the bottom arc is slight, the inside arc is off quite a bit. But the missing color in the notches is why I presumed rastering is an issue too...
tx.jpg


And under close scrutiny, your newest pic does show a reasonably bad raster error, in the area where the star and the S intersect-- Corel paint's 'auto adjust' actually changed the color of the pic, but it highlights the issue, where the horizontal line of the star makes a 3-sided rectangle up into the S and is not engraved.
tx2.jpg
--It's a simple 2-coordinate error, take the 2 corner points out and it would be a straight line... again, this error (IMO) takes the steppers and other mechanicals out of the equation, because aside from the fact those 2 points shouldn't be there, the laser engraved it perfectly...

The yellow arrow errors in the first pic are all from vectoring. It looks like a raster error after the vectoring is done because of the thickness of the line left by the vector. Not sure what the blue arrow is in the second pic. Contrast wasn't great on that piece of cherry. Will try it again on a piece of anodized aluminum to get a better idea.

If it's not mechanical or motor, I'm just at a loss to figure out what this is. Kind of a bummer. I do appreciate everyone's help, tho.

As a last resort, I might take the motor off and apart and see if the motor bearings are having an issue or if one or more of the graphite brushes are worn down. I think I could safely replace bearings and brushes (done that more than a few times on my dillon brass trimmer). But anything else for motor repair is beyond me.

[Andrew Stow]

I did a little test. With the machine off, I moved the x axis back and forth and noted that the parts where it has "resistance" are when the pulley set screw is pointing at about the 8 o'clock position.

Then I put one of those test pieces back on the table, turned the machine on, and turned the diode pointer on. Then I put the machine into service mode and jogged the x axis over until the dot was pointing at the errors. In each case, the error was right about where the x axis pulley set screw was pointing to approximately 8 o'clock. Further, when jogging x axis back and forth, it didn't jog at a uniformly smooth rate. It would move smoothly until it got to those parts where the errors occurred during vector, then kinda skip ahead real fast.

MAYBE whatever it is that is happening isn't as likely to occur during raster because raster has enough velocity to get past these "bad spots" without appreciably slowing down and it only messes up during vector because vector mode jogs the motors slowly enough to have a problem getting past the "bad spots" at a uniform speed.

I'm really leaning toward something wrong with the motor. I've got a couple of small jobs to do for friends that I can get by on raster mode for now, but when that's all done I think I'm going to try taking the motor out and seeing what the brushes look like and how the bearings feel. If any of this rings any bells for any of ya'll please yell out. There's a ton of stuff I don't know.