Engraving issues with LO DSP

Discussions and help on this commercial controller.

Moderator: twehr

Engraving issues with LO DSP

Postby naPS » Fri Aug 19, 2011 2:35 am

So, I got the LO DSP in yesterday, and made up a cable to interface to the v2 interface board today to test it out. Overall first impressions are... meh. The documentation is absolutely horrific for these things, which makes me a sad panda. Regardless, the controller definitely seems workable.

I've got it up and running and doing everything it's supposed to be doing. It's just not doing the majority of it very well. I'm 99% sure it's operator error, but you never know. I've got the homing, direction and table sizing all pretty much correct. I've got the units set to mm/seec. The first thing I'm finding confusing is that the units don't appear to be standard. The pulse per unit is an interesting way to do things, and I've got it dialed in so I can cut 50x50mm boxes in vector all day long with 100% precision. So I know that part is at least good to go. The thing that I'm somewhat confused about is the rest of the speeds. When it says acceleration, and speed, and whatnot in all the different locations (x and y axis setup, etc.) is that all in mm/sec? Or is it in pulses / unit? It's confusing, and the values in the setup documentation seem extremely high for what I've been able to reliably use with Mach3.

I've been using the engraving plugin quite a bit with Mach3 @ 375 ipm in both X and Y at around 20-30% laser power (as set by the pot) and been getting excellent results on most surfaces. I can only assume that I should be able to do the same with the DSP, at least mechanically I know it's possible. And, from what I've seen from whmeade, that's about right. 200mm/sec = 472 ipm, which is about 100 ipm faster than what I've been engraving at, but should be doable. I've also been using a .001" scan gap in the Mach3 plugin, which is approximately the size of the laser dot, and it's been working very well. That should mean that my scan gap for engraving in the DSP should be (.001 *25.4) = .0254mm. I see whmeade using .1 and .09mm, which should be fine for most of what I'm doing as well. I just can't get that to work. Any time I use a scan gap while engraving of below .2, It gets skewed in the Y direction pretty badly. Once I get up around .2 and higher, the problem goes away, but the resolution suffers slightly. I tried fiddling around with the speed and acceleration of both axis, but it didn't change what was actually happening during the engrave. The only thing that had a noticeable effect on the skewing was the scan gap setting.

Here's a picture of what I mean. The four circles were engraved at 100% laser power for the top row, 50% laser power for the bottom row, with both rows @ 200 mm/sec. The first dot in each row is a scan gap of .1, and the scan gap increases by .05 for each circle you move to the right, with the last one being .25. You can see that the first circle with a scan gap of .1 is very skewed, and starting with the ones with scan gaps of .15 they're pretty good, with the best looking ones at .2.

TJK_7586.JPG


The second issue I'm having with the engraving is that it's giving me strictly monochrome results, with no shades of grey at all. It's either basically black, or white. I'm able to get several shades of grey using the Mach3 plugin, so I know it's possible. Anyone have any insight on this? I'm sure the PWM functionality works, as I'm able to engrave boxes at different laser powers, and they are different darknesses in the wood. Although, it seems like I'm using much higher power levels than most everyone else that's engraving, so that's a little weird. I typically use 20-30% power with the Mach3 plugin with great results, but I'm using 100% power with the DSP with mediocre, totally monochrome, results.

I'm wondering if the length of the cable from the DSP to the v2 interface board is too long. I used a 10' cable, figuring it would probably be fine, since the cable that runs from my parallel port on my computer to the interface board is 15' and works fine. I'm going to shorten it significantly tomorrow, and see if that makes any changes. If that doesn't work, I'll start some more troubleshooting from there. But for now, does anyone have any insight to this?

Also, what settings do folks use when you import photos? Do you leave all three of the tick-boxes checked when bringing in images? What settings are you using for the actual layer properties screen? Can I get a screenshot of someone's setup?

Thanks fellas.
naPS
 
Posts: 202
Joined: Sat Apr 09, 2011 1:53 am

Re: Engraving issues with LO DSP

Postby twehr » Fri Aug 19, 2011 1:23 pm

naPS wrote:So, I got the LO DSP in yesterday, and made up a cable to interface to the v2 interface board today to test it out. Overall first impressions are... meh. The documentation is absolutely horrific for these things, which makes me a sad panda. Regardless, the controller definitely seems workable.


I hope you are talking about the LO documentation and not the installation document I wrote and has been used by several people very successfully for installation. If you did not use it, you probably spent too much time trying to figure things out on your own.

naPS wrote:I've got it up and running and doing everything it's supposed to be doing. It's just not doing the majority of it very well. I'm 99% sure it's operator error, but you never know. I've got the homing, direction and table sizing all pretty much correct. I've got the units set to mm/seec. The first thing I'm finding confusing is that the units don't appear to be standard. The pulse per unit is an interesting way to do things, and I've got it dialed in so I can cut 50x50mm boxes in vector all day long with 100% precision. So I know that part is at least good to go. The thing that I'm somewhat confused about is the rest of the speeds. When it says acceleration, and speed, and whatnot in all the different locations (x and y axis setup, etc.) is that all in mm/sec? Or is it in pulses / unit? It's confusing, and the values in the setup documentation seem extremely high for what I've been able to reliably use with Mach3.


If you set it up for mm/s then everything (except for the pulse units) is based on mm/s or, in the case of acceleration, mm/s2.

As for high speeds, I can run quite well up to 700 mm/s, do I don't think you are misreading or misunderstanding anything there. The max speed that your particular machine can run will be determined by the mechanical features (how loose, tight, smooth, etc.) and the drives you are using along with the micro-step setting. If you are not getting the results you think you should, try adjusting the micro-steps AND simultaneously adjusting the pulse units accordingly (either halving or doubling to compensate for changes in micro-steps).

naPS wrote: I've also been using a .001" scan gap in the Mach3 plugin, which is approximately the size of the laser dot, and it's been working very well. That should mean that my scan gap for engraving in the DSP should be (.001 *25.4) = .0254mm. I see whmeade using .1 and .09mm, which should be fine for most of what I'm doing as well.


The scan gap is how you control the effective DPI of your output. For scanned images (photos) it should be set to match the DPI of the photo image - i.e. a 254 DPI photo should be set at .10 mm scan gap. For 300 DPI, you should set it at .09 mm. Mismatching your scan gap and the actual photo DPI will likely cause you to have skipped lines or double engraved lines in the output. They will be visually unacceptable in many cases.

When doing filled vector engraving, you are free to set the scan gap to anything you want. I generally leave mine at the default .10 mm. Bill Meade, on the other hand, likes using .09 for a slightly higher resolution (300 vs 254). You can go do to the .02 or .03 for really high resolution but doubt you will find it worth while.

Even though your laser may be mechanically capable of a 1000 DPI resolution, you will likely never need that high of output. The time necessary to do 1000 scan lines per image inch is 4 times as long as it takes to do 254 scan lines per image inch. For MOST things, the difference in true output will be unnoticeable.

naPS wrote: I just can't get that to work. Any time I use a scan gap while engraving of below .2, It gets skewed in the Y direction pretty badly. Once I get up around .2 and higher, the problem goes away, but the resolution suffers slightly. I tried fiddling around with the speed and acceleration of both axis, but it didn't change what was actually happening during the engrave. The only thing that had a noticeable effect on the skewing was the scan gap setting.


If you are having issues with the Y axis on scanned images or vectors not coming out at true height, you likely have a combination of micro-steps and pulse units that does not work well for your particular drives. Read my thread http://buildlog.net/forum/viewtopic.php?f=13&t=584&p=4141&hilit=missing#p4083 where it is discussed at length.

EDIT - I just re-read your post. If you are getting skewing (not compression) then it is likely that you are dropping some x axis steps. Could still be the say kind of issue, but be sure to look closely at your x axis. Specifically, decrease your acceleration a little and/or decrease your start speed (mine is at 10, I think).

naPS wrote:
The second issue I'm having with the engraving is that it's giving me strictly monochrome results, with no shades of grey at all. It's either basically black, or white. I'm able to get several shades of grey using the Mach3 plugin, so I know it's possible. Anyone have any insight on this?


The results ANY laser will give you on engraving is 1st a function of the quality of the image you are asking it to engrave. That being said, it sounds to me that you MAY have skipped an important step in importing your photos to PHCad. AFTER you import the photo, you still have to tell it to dither the image (menu function). If you don't do that, it will do a strictly histogram based output - any pixel greater than 50% gray is BLACK and any pixel less than 50% gray is WHITE. Usually, this is not the effect you want. Make sure you dither the image after import and before downloading to the laser.

--------------OR----------------

The best solution is to do your own dithering in Photoshop (using the Gold Method plugin) or in CorelPaint (a manual process) or PhotoGrav (best but most expensive solution). Then just import the image into PHCad or send the image to PHCad from CorelDraw. You will get much better results than if you try to use the dithering from PHCad.

naPS wrote:Also, what settings do folks use when you import photos? Do you leave all three of the tick-boxes checked when bringing in images? What settings are you using for the actual layer properties screen? Can I get a screenshot of someone's setup?


I think this is covered in my answer above, but I obviously do not have any screen shots for you. I'll try to do a blog posting comparing the various methods this weekend (I hope) over at http://DIYLaser.Blogspot.com. Anyone not currently subscribed to that may want to do so. I am trying to post useful information and insights here on BuildLog.net, but am trying to do more in-depth or less "buildish" articles over there.
tim
--
"The answer is usually easy and obvious once you know what it is." tw

DIYLaser Blog
SemiHomemadeTools.com
twehr
 
Posts: 439
Joined: Mon May 24, 2010 4:49 pm

Re: Engraving issues with LO DSP

Postby naPS » Fri Aug 19, 2011 4:27 pm

twehr wrote:I hope you are talking about the LO documentation and not the installation document I wrote and has been used by several people very successfully for installation. If you did not use it, you probably spent too much time trying to figure things out on your own.


Correct. I'm talking about the documentation for the controller in general. I've downloaded the manuals from several manufacturers of lasers that all use this same controller, and they're all garbage. They all tell you what you can set each parameter to, but not what function the parameters actually perform. It's great that it tells me to set my Pulses per Unit at whatever value, but I would like to know what "Unit" I'm trying to set my pulses to. I did find one set of instructions that set the "Unit" was in picometers, and one that said it was in nanometers. So I have a feeling that's fairly firmware specific. It sucks not knowing what unit I'm trying to set the pulses for, since I could easily calculate out the actual number of pulses per whatever, enter that value, and then begin any troubleshooting activity from there if my cuts aren't coming out correct.

The setup document you came up with is great, and got me running in about 2 seconds.

twehr wrote:As for high speeds, I can run quite well up to 700 mm/s, do I don't think you are misreading or misunderstanding anything there. The max speed that your particular machine can run will be determined by the mechanical features (how loose, tight, smooth, etc.) and the drives you are using along with the micro-step setting. If you are not getting the results you think you should, try adjusting the micro-steps AND simultaneously adjusting the pulse units accordingly (either halving or doubling to compensate for changes in micro-steps).


Yup. Got that down. I can make perfect vector cuts with it, no problem. Again, knowing what the actual unit was would be helpful, but in the long run, I guess it doesn't matter that much as long as it cuts 50mm when I tell it to cut 50mm.

twehr wrote:The scan gap is how you control the effective DPI of your output. For scanned images (photos) it should be set to match the DPI of the photo image - i.e. a 254 DPI photo should be set at .10 mm scan gap. For 300 DPI, you should set it at .09 mm. Mismatching your scan gap and the actual photo DPI will likely cause you to have skipped lines or double engraved lines in the output. They will be visually unacceptable in many cases.

When doing filled vector engraving, you are free to set the scan gap to anything you want. I generally leave mine at the default .10 mm. Bill Meade, on the other hand, likes using .09 for a slightly higher resolution (300 vs 254). You can go do to the .02 or .03 for really high resolution but doubt you will find it worth while.

Even though your laser may be mechanically capable of a 1000 DPI resolution, you will likely never need that high of output. The time necessary to do 1000 scan lines per image inch is 4 times as long as it takes to do 254 scan lines per image inch. For MOST things, the difference in true output will be unnoticeable.


And therein lies the problem I'm seeing. I'm beginning to think this is a software issue, as I'm having issues in the X direction, only at scan gaps of about .14 and lower. .15 and up, it's great. But the scan gap should have no effect on my X movement, only on my Y movement. But, it's having a pretty interesting effect on my X movement. I've created another profile, and I'll chuck the picture up at the end of the message here.

twehr wrote:If you are having issues with the Y axis on scanned images or vectors not coming out at true height, you likely have a combination of micro-steps and pulse units that does not work well for your particular drives. Read my thread http://buildlog.net/forum/viewtopic.php?f=13&t=584&p=4141&hilit=missing#p4083 where it is discussed at length.

EDIT - I just re-read your post. If you are getting skewing (not compression) then it is likely that you are dropping some x axis steps. Could still be the say kind of issue, but be sure to look closely at your x axis. Specifically, decrease your acceleration a little and/or decrease your start speed (mine is at 10, I think).


So here's the rub, I'm not losing steps. I made a test profile that would check specifically for this, and I've confirmed that mechanically, I'm good. I'm pretty sure it's a software problem. What I did was make the same pattern I made before, the 8 circles, but I added 8 vector cut outlines for those circles that would cut last. The raster portion of the file populates the circles from the bottom left to bottom right, and then moves up a row, and does top left to top right. The vector operation was fairly random with the order of the circle profiles it cut. The results were fairly interesting. The bottom left circle is both skewed and out of position in raster, with the remaining circles on the bottom row not skewed, but out of position. The top row has the first circle skewed, but none of them out of position. All the vector cuts ended up exactly where they should have been. If I was losing steps, the vector cut circles would have been out of position as well.

The weird thing is - the top 3 circles on the right line up perfectly. If something was amiss, these ones should have suffered the most, but they're literally perfectly aligned. And since the raster profiles for the top and bottom rows are exactly the same, with laser power the only thing being changed, you would think that the results as far as the circle alignment would be repeated between the rows, but it's not. So strange.

twehr wrote:The results ANY laser will give you on engraving is 1st a function of the quality of the image you are asking it to engrave. That being said, it sounds to me that you MAY have skipped an important step in importing your photos to PHCad. AFTER you import the photo, you still have to tell it to dither the image (menu function). If you don't do that, it will do a strictly histogram based output - any pixel greater than 50% gray is BLACK and any pixel less than 50% gray is WHITE. Usually, this is not the effect you want. Make sure you dither the image after import and before downloading to the laser.


Yup, definitely skipped that step. I'll do that up, and give it another shot. I figured I had to be doing something wrong. Looks like I'm going to be getting familiar with that function in Photoshop.

twehr wrote:The best solution is to do your own dithering in Photoshop (using the Gold Method plugin) or in CorelPaint (a manual process) or PhotoGrav (best but most expensive solution). Then just import the image into PHCad or send the image to PHCad from CorelDraw. You will get much better results than if you try to use the dithering from PHCad.


Do I still have to do the dithering in PHCad when I import the image?

Thanks for your help with this Tim!

edit : Oops! Forgot to attach the image. Here it is :

TJK_7587.JPG
naPS
 
Posts: 202
Joined: Sat Apr 09, 2011 1:53 am

Re: Engraving issues with LO DSP

Postby twehr » Fri Aug 19, 2011 5:04 pm

naPS wrote:And therein lies the problem I'm seeing. I'm beginning to think this is a software issue, as I'm having issues in the X direction, only at scan gaps of about .14 and lower. .15 and up, it's great. But the scan gap should have no effect on my X movement, only on my Y movement. But, it's having a pretty interesting effect on my X movement. I've created another profile, and I'll chuck the picture up at the end of the message here.
...
Do I still have to do the dithering in PHCad when I import the image?


This is very similar to the behavior I had in the Y direction. For me, there was a definite speed at which it stopped/started working. Adjusting the micro-steps and pulse width allowed me to change where that speed break was. I don't know that it is a software bug as much as it is an issue with the speed at which the data needs to be generated/sent. Changing the steps/pulse unit will increase or decrease the amount of data that needs to be generated/sent.

No - you do NOT have to dither after importing into PHCad if you have already done it elsewhere.
tim
--
"The answer is usually easy and obvious once you know what it is." tw

DIYLaser Blog
SemiHomemadeTools.com
twehr
 
Posts: 439
Joined: Mon May 24, 2010 4:49 pm

Re: Engraving issues with LO DSP

Postby naPS » Fri Aug 19, 2011 5:52 pm

twehr wrote:This is very similar to the behavior I had in the Y direction. For me, there was a definite speed at which it stopped/started working. Adjusting the micro-steps and pulse width allowed me to change where that speed break was. I don't know that it is a software bug as much as it is an issue with the speed at which the data needs to be generated/sent. Changing the steps/pulse unit will increase or decrease the amount of data that needs to be generated/sent.


Hmm, there might be something to this - maybe the controller just can't keep up with the number of pulses necessary to maintain a .1 scan gap, and the X speed. But you would think there would be distortion in Y as well if that was the case. Also, the circles are identical in speed and power, the only thing that changed was scan gap. So, it shouldn't have any effect on it, especially since there is so much less movement in Y than X. It's strange, for sure.

I'm not so sure the controller can't keep up though, since Bill is able to scan at .09 scan gap at 200-250 mm/sec. I wonder what pulses per unit he's at. I'm at 2000 steps per inch, I might try dropping that down to 1000 steps per inch on the drivers and see what effect that has on it.

And thanks for the dithering info - that did the trick. The image actually came out pretty good - it's almost as good as the Mach3 plugin was giving me, and with some tweaking, I'm sure I can improve it. Improving my scan gap will help that immensely.
naPS
 
Posts: 202
Joined: Sat Apr 09, 2011 1:53 am

Re: Engraving issues with LO DSP

Postby twehr » Fri Aug 19, 2011 6:00 pm

naPS wrote:But you would think there would be distortion in Y as well if that was the case. Also, the circles are identical in speed and power, the only thing that changed was scan gap.
...
And thanks for the dithering info - that did the trick. The image actually came out pretty good - it's almost as good as the Mach3 plugin was giving me, and with some tweaking, I'm sure I can improve it. Improving my scan gap will help that immensely.


Y is getting a lot fewer signals than is the X.

If you get good at creating your original artwork and dithering it well, you will beat Mach3 by a long way.
tim
--
"The answer is usually easy and obvious once you know what it is." tw

DIYLaser Blog
SemiHomemadeTools.com
twehr
 
Posts: 439
Joined: Mon May 24, 2010 4:49 pm

Re: Engraving issues with LO DSP

Postby naPS » Fri Aug 19, 2011 7:25 pm

twehr wrote:Y is getting a lot fewer signals than is the X.


That's exactly my point though. If it was a processing problem, it would occur in both axis. Processing is independent of axis, it's simply just calculating moves. If it's missing in X, it's sure to miss in Y occasionally as well, and this just isn't occurring, which is again leading me back to this being a software issue. And if it truly was a miss, it would carry through to the rest of the job, which it clearly isn't. With fewer steps in Y, it would magnify the problems in Y if there actually were any.

twehr wrote:If you get good at creating your original artwork and dithering it well, you will beat Mach3 by a long way.


Hmm, that will be interesting, because I've had some pretty stunning results using the Mach3 plugin. If it's better, I'd be very, very happy.
naPS
 
Posts: 202
Joined: Sat Apr 09, 2011 1:53 am

Re: Engraving issues with LO DSP

Postby trwalters001 » Fri Aug 19, 2011 7:56 pm

Hi All,

naPS wrote:I'm at 2000 steps per inch, I might try dropping that down to 1000 steps per inch on the drivers and see what effect that has on it.


My system is set up (mechanically) for 1000 spi, so I set up the DSP that way. I wasn't getting very good results until I dropped the DSP down to 500 spi...

I also get better results engraving if the bitmap size is some multiple of 250/500/1000.

Tim
trwalters001
 
Posts: 173
Joined: Sun Oct 24, 2010 10:08 pm
Location: Atlanta, GA

Re: Engraving issues with LO DSP

Postby J45on » Fri Aug 19, 2011 7:58 pm

Reading with great interest guys
I am also experiencing similar results :(
I am running Barts 200 step motors, light object PSD5042-2P drivers set to 8 micro steps
And a thunderlaser DSP

Would this also effect cutting as well ?
I cut a long box today and the fingers started to go slightly out of line,although they are spot on in corel draw

Image

Image
J45on
 
Posts: 258
Joined: Thu Mar 31, 2011 11:30 am
Location: Kent,united kingdom

This makes no sense...

Postby naPS » Sat Aug 20, 2011 3:17 am

In the immortal words of Ted "Theodore" Logan - Strange things are afud at the Circle K.

To do a quick test on if the steps per inch is affecting my results, I set the drivers to take me from the previous 2000 steps/inch to 1000 steps/inch. I fired up Mach3, did a couple of test cuts and confirmed that yes, X and Y are actually at 1000 steps/inch. I then fired up the DSP, and roughly halved the Pulses per Unit setting - thinking that if I took the number of steps / inch in half, that this should also go to about half. Made sense at the time. What didn't make sense is that when I fired up the laser using the DSP after making the changes and resetting it, the head started flying all over the place. The net result is that I ended up having to exactly DOUBLE the "Pulses per Unit" setting from ~12.7 to ~25.4, which makes absolutely no sense to me. Unless the Pulses per Unit is inverse of what I think it actually is, which is possible, and in this case, probable.

Anyhow - now that my Pulses per Unit were doubled, I ran the test pattern again, and received the expected result. With the number of pulses doubling, my X axis skewing was literally about twice as bad as before. This really leads me to believe that the DSP doesn't have the computational power to keep up with high resolution stepping - which isn't the end of the world with .001" precision, but as it stands right now the best I can do is about .002" precision, which is about half of what others have been able to get with the controller. The other strange thing is that the vector cutting at all speeds, powers, etc. is perfect. This again leads me back to believing that it's quite possible that it's a software issue with the way it's handling raster.

I'm going to change the stepping to different resolutions tonight to play around with it. I can get 500 - 16,000 steps / inch with the drivers that I have here, so I'm going to give that a shot.

One thing that hit me - is there a pulse length setting somewhere in the firmware? If set incorrectly, that could definitely be causing an issue, although again, it should affect both X and Y movement, so...

TJK_7588.JPG


Here's the result of the test ran twice at lower steps / inch, with the original test at the higher steps / inch to the left.

Edit : Okie dokie. I set the drivers to 500 steps/inch from 1,000 steps/inch, and the Pulses per Unit setting did indeed double. So it's an inverse relationship. I'd really, REALLY, like to know what the Pulse per Unit units actually are.
naPS
 
Posts: 202
Joined: Sat Apr 09, 2011 1:53 am

Next

Return to Commercial DSP CO2 Laser Engraving/ Cutter Controller

Who is online

Users browsing this forum: No registered users and 12 guests

cron