Sweet! Thanks Bryan
Wow, guys it is weekend here and I did not really expect a response so soon, let alone another version to test 
Here you are already beating Picengraves support LOL
“Due to my setup I have to flip every image – that is no issue, but the placement often is.” -Can you just flip your stepper plug around?“It would be great to allow for an offset for the start of the engraving so you can engrave in the center of something bigger like a name plate.
Is it possible to add a box for a X and Y value offset from the origin?”
-I’m not sure I entirely understand what you’re saying. Can you not use the origin setting and pick the middle?
I was thinking of just flipping the stepper and to change the config but there is more to it.
The K40 machines have the origin and homing position in the top right corner, same for the limit switches.
One day I might re-build the machine and do to it to CNC specs with the movements and axis orientation.
Either I missed the origin setting somehow or fail to use it.
No matter what I do the image is always placed right in the origin position - but will try that again with new program version.
Maybe an example is better:
You have a plate of 20x20cm but want to place a 5x5cm engraving right in the center of the plate.
I might be able to mess around a bit with the origin setting but I see no way to be able to place the plate and get the engraving centered on it.
Worse with bigger parts of course.
Will get back in a bit after testing V6
Edit:
The start of the box is still messed up with the coordinates:
; Move to top left corner and begin box
G92 X0 Y8607064
Otherwise the code looks nice and clean now.
I can offest the engraving with M92 command after setting the laser to the referenced image coordinates.
For eaxmple a thicker piece of wood in the center of the table gets G01 X95 y85 followed by G92 X0 Y0 Z0.
So I first move the laser to the right position for the image, then set this position as the home reference.
Would be great to include this into the program instead of the
“; Move to top left corner and begin box
G92 X0 Y8607064”
This way the user can set the origin and no need to fix the decimal problem 
The test engraving on 3mm ply ended badly as in some areas the laser burned through the first layer, currently trying again on a piece of hardwood.
Only doing it in 0.2mm resolution this time to save some time.
If it works out ok I will try the other side with 0.1mm and pulsed laser operation.
So at least one image should be uploaded soon
Sneak peek of what else I’ve been working on tonight. Optimized rastering!!!
I’m going to be doing some testing on the code before I release it.
Hmmm,
things like this I actually do in a two step process:
First a slight vector cut of the outlines, then just a normal raster engraving for the infill.
Everything else I tried so far either ended in bad burn marks on the outlines or fine details missing.
Considering the speed of the vector tracing the extra time on the machine can be neglected.
You can see some example of this way here.
Which should also show that I am not Andrew, just too lazy to leave my normal Nick here so people follow me around.
Quite like it to be “the new guy” for once 
But I agree, if intergrated into your program it would give even more options and makes the use of other tools and programs almost a thing of the past.
One thing that would be a really nice addition though is a g-code sender from within the program.
Of course for spiled people like myself it would be great with a preview of the object in question on a plate with the size of the cutting area.
Since it is free to, maybe it is possible to “salvage” some code from Pronterface/Printrun?
Sending these massive amounts of code is a challenge in terms of right timing, checksums and so on, so somthing optimised for the few code bits we use here could improve stability during long engravings.
And since I now outed myself here, would you mind if I include your program and a link to here with my laser related Instructables?
I don’t see it as advertising if something good can get more attention especially not if it is so well supported freeware.
Edit:
Took the liberty to start some advertising here, feel free to leave a comment if you like.
Ok, half way through the second attempt of doing it right I gues I found out why it was so bad in the prvious runs 
Despite everything Marlin does not react at all to the decimal digits, only the stuff in front of the dot is recognised and used.
With the power levels now just at 6 and 8 it becomes quite obvious as there is only 3 shades of grey now and all detail lost.
Seems I now have to figure out a way to change the PWM handlin in Marlin from the default 0-255 to 0-1000 or better 10000 somehow.
Tried this already a few weeks ago and utterly failed as I was greeted with compiling errors LOL
Here is the last engraving, which was totally overburned.
Done on hardwood (has too much resin in it too) with the power levels between 6 and 16.
Before cleaning:
After cleaning:
As you can see there are also pinholes burnt into the wood, this happens almost every time the power changes or the head stops.
Seems to be a problem in the Marlin firmware too or at least related to the slow processing of all the code.
Bit longer and the other engaving is done for comparison.
And here we go:
Power between 6 and 8
As you can see there are no 256 shades of grey, just next to nothing for withe and three more - everything from 6 to 8 in solid numbers only
Guess I am back to the drawing board for now.
A while back one of the users on the forum concluded that piclaser performed better because it had the decimal power values.i dissected this wasn’t the case but i went ahead and made the change anyway.I’m still not sure if marlin can do anything with the decimals since your firmware isn’t really marlin.the turnkey tyranny firmware your using isn’t slightly modified.I’ve looked at it and it’s heavily modified in how it processes laser commands so the processing speed problem may possibly be there. I have zero speed issues running vanilla marlin firmware auth a diode laser. You also haven’t mentioned the TTL frequency of your laser driver. This is hardware related and will determine how quickly the power pulses can switch from one level to another.
Leo, I’m not sure if your forum status allows you to edit your older posts. If you don’t mind I can edit your first post on this thread with the github link to try and make it easy to find the most recent version? -I’m about to do my first cncburn, making sure I had the most recent version!-
Don’t mind at all. Best place for it.
What is your feed rate here? If your laser is that powerful then you need to pick up the speed as far as the hardware limitations of your machine will allow you to. If you’re charring the wood at level 8 out of 255 then your laser may not even handle pwm input at all. A level of 255 is equal to 5v on Arduino so that means 51 power levels per 1 volt. A power level of 8 is only equal to less than .2 volts then! If you want to try something that may improve your results without much effort then try running the PWM output from Marlin through a resistor voltage divider that’ll bring it down to the low levels you need. This way you’ll get the full 255 step resolution from the Arduino PWM pin scaled down to a much lower level. Truthfully, I doubt it’ll work though because something just doesn’t seem right about your firmware/hardware set-up.
Vicious1-
Is there anyway I could persuade you to delete many of my posts in this thread? I blogged all over it not realizing I could reply to specific threads, meaning i overlooked the reply to the right of message. Dont want to ruin my chances of help in this forum or look like an idiot in posting like a crazy person.
Leo and Bryan
I absolutely love this software, however I have had much better luck with the previous version 4. Once I figured out how to offset the power levels and how to overcome the missed steps in y axis movement. In version 4 I use 10-12% wood at 5000 feed, which uses 0-28 or 30 depending on 10 or 12% in wood setting, but still amazing results on the few pieces I’ve ran. In version 4 the only changes I would have liked due to the power of my laser is the ability to use faster feed settings, for example if I set it at 7000 it still only moves very slowly at a predetermined feed rate, but one I would see more common with a feed of 1500. The feature I miss most when in image2gcode is being able to crop the image, but I dont mind editing image else where then open in image2gcode.
I have tested the newest version 6 and at 35% under wood setting that puts max power of my laser at its max of 89. My laser goes from 0-100, but again I have found a way around everything by using lower settings. The only downside is the lack of intensity in any given in image, meaning most in here have 0-255 while I have 0-30. I will upload a few of the images generated via image2gcode but they will be completed with version 4. Oh I did notice when playing with version 6 that it would not allow for dither at all, I would click it but then it would revert back to greyscale?
Sorry again for blogging all over the place and freaking out due to my lack of understanding in all this stuff. I got in way over my head and damn near drowned until finding this forum and also help from a friend.
Ok, first off:
At level 6 the laser starts to work, so far I have not used anything above 35 on wood as otherwise there is just too much burning.
The laser usually operates at 60kHz and PWM control is working just fine except that it only works in full increments.
Tested it with the stock standard Marlin on my 3D printer and confirm that Marlin is unable to handle digits.
Even before changing the crappy controller PWM was working fine, just only through the digital input on the panel - from 0 to 100% in 0.01% steps - confirmed with my oscilloscope.
Those values are also what I used in the firmware as the default settings for pulsed operation - yes the pulse control works fine too.
I had some engravings done with GRBL, the version for the Ramps 1.4 board.
There the PWM control goes from 0 to 1000 and the engraved results have not been that bad, although I gave up on GRBL before finetuning the settings.
So the hardware is quite capable, I only need to find the right stuff to transfer commands into motion.
Speed:
The above was done with the speed, acceleration and jerk set to max values but it dit not reall change from the one before done at just 1200mm/m.
I assume that due to the firmware forced pause when handling M3 commands as a single command instead of within the G1 movement the pinholes happen.
Grbl was much faster here but had too many other limitations that right now I neither have the time nor will to look into.
Marlin I know a bit GRBL properly for my machine would mean massive hardware changes, by massive I mean the need to take the entire machine apart to fix enstops and motors to be conform with grbl.
I did change the code again to include the laser commands into the G1 commands but it only slightly improved performance.
The massive amounts of tiny movement commands flood the look ahed buffer as well as the serial buffer.
If I find some time and the right mood I will try to change the firmware from the defaults to using more memory for the buffering and serial commands.
So far I am on the lower end of the scale using the same values that worked good on my 3D printer.
But I don’t think that 32kb more would make a massive difference.
So I will do the next test with a simple Gcode sender if I find one that like me
My persoal favourite would of course be to change the Marlin firmware so that the PWM stuff for the laser can go from 0-1000 instead of just 255.
10000 would be better, especially for more sensitive materials and I think it would not make much difference in terms of the coding.
Only problem is that I have no clue where to even look.
In some reference I found that the internal clock signal is used and that instead of the fixed 255 a user defined value can be used, but of course nothing about how to do this LOL
I am quite reluctant to go back from an otherwise perfectly working system to GRBL just to be able to do 8bit engravings -there has to be another way…
By the way, the guys at Piclaser said that a CO2 laser is only compatible with their CNC plus laser stuff and there just for bitmap engravings.
If I want to do bitmap engravings I just do them through Inkscape free of charge 
Last one for today:
I don’t know about the standard firmwares you guys use for the laser but Marlin supports Base64 encoding for raster engravings.
Well, not just for the laser stuff but there is so far no real use for it on a normal 3D printer…
What I was thinking is to comine all those G1 commands with no laser power change into a Base64 string.
If there is a power change and only a few movements after this till the next power or line change the encode line will just be short of the standard length.
This should allow for a code reduction of about 60% comapared to what the V6 version offers, plus a massive performance gain as the machine does not need to handle so many commands.
But I have no clue how hard it is to implement and if it works on all firmwares, GRBL of course sticks to CNC standards and won’t support it.
I wasn’t aware that grbl supported 10bit pwm on the spindle commands, maybe this change was made on the grbl for ramps version. Yes you can modify the firmware for more resolution on PWM but it does involve playing with timer prescalers. Before you do, make sure that the timer you’re modifying isn’t going to affect stepper pulses or any other time related functions that may be using the same timer. Plenty of timers available on the mega2560 but i think only two of them will support up to 16bit pwm. You’ll probably need to modify the firmware code that processes the fan/spindle commands too.It may just be using an analogWrite command for the pwm output and that command only supports8bit values.you’ll need to write to the registers directly if you are going to 10 or12 bit.
Thanks Guys for all the hard work!
Things are working pretty smooth on my end. I did a nice portrait as a trial but The subject didn’t want her picture up here…bummer. I am having some small glitches that didn’t show up with some different software but I want to do some more tests before I say anymore about it.
I did a little focusing script for repetier, I’ll do a separate thread for it.
I like the honeycomb fan grill and that cool vortex twist on the end of your housing.you gonna post that design? If glitches seem speed related then it may be my fault for adding a bunch of decimals that Marlins 8 bit pwm can’t even process. I’ll make them optional on next revision.
@Andrew ,
“In version 4 the only changes I would have liked due to the power of my laser is the ability to use faster feed settings, for example if I set it at 7000 it still only moves very slowly at a predetermined feed rate, but one I would see more common with a feed of 1500.”
-You may have hit the max feedrate that is built into Marlin. You can delve into the firmware (Configuration.h) and bump up the max allowed feedrate and see if that solves the issue.
“The feature I miss most when in image2gcode is being able to crop the image, but I dont mind editing image else where then open in image2gcode.”
-I agree, cropping seems like a nice feature. I don’t really want this to become an image editing tool because that’s just more to maintain and troubleshoot, but cropping seems pretty benign to add.
Unfortunately I do not have any experience/access to the K40 setup so you would have to look elsewhere for help on that.
“Oh I did notice when playing with version 6 that it would not allow for dither at all, I would click it but then it would revert back to greyscale?”
-Hmm, I will take a look into why that’s happening
Thanks, I can post the design but right now it’s pretty part specific, Heat sink, and board at least. It also has some flaws right now, hard to put on a new lens because the heat sink kinda floats in there. I think when the resistors get here I’ll do another revision on it when I test the voltage divider. Not sure what is going on really, maybe its the power supply. some of the lines start late or end late and cause artifacts in the prints. I also think the slow accelerations in my firmware are causing some of the stuttering, I guess it could be the decimals, is that from the servo pins we are using, if the divider works and we use the fan port is it going to be different?
You don’t need to answer any of that. I am super happy with the software you guys have adapted. I’ll stick to hardware and stay out of your way on the software.
I found some 25mm stainless shower curtain rods…time to finally start the mpcnc updates. So much easier when I can test both sizes…
I don’t think using the fan port will be any different. On your next burn try getting version 4.i think that is last pre- decimal version.If the stuttering goes away then I’ll definitely make the decimals optional. Using host software may slow things down too or your acceleration settings could be it too. I’m Sure you’ll meter that divider output before you connect your driver but it should work. I think there’s a FASTPWM definition that can be enabled on the fan pins. I’ve read some posts suggesting it improves laser grayscale definition but haven’t verified this.
The “Dithering” function has been fixed. The inherited code is full of typos and I fixed “dirthering” to dithering, and well, that messed up some other references. It is back to the typo-d version until I have time to scrub the entire thing.
I didn’t roll the version number so if you want dithering, just re-download it from the source.
Awrsome job once again Bryan! I’ve never even tried the dithering. I hope someone will give it a shot and post pics for comparison to raster. Good stuff!
Attached are three projects I created via image2gcode using cheap k40 laser with arduino+ramps 1.4 using a forked version of marlin modified for 40watt laser. I am super excited to have something finally working after months of headaches. The last image shows where my machine was sticking but printing now in diagonal the problem is solved again so freaking awesome.
Bryan, I will check the firmware for the predetermined feed rate max and min, but when I use another program such as inksckape and set feed at 6000 or more I have no issues. At that rate of feed it is super fast but still the firmware does allow it?
Also at a low feed rate say 250 or 300 and 100% power, my laser will come very close if not succeed cutting through 1/4 inch MDF, these images that I have attached were created at max power in image2gcode of 25.
Thanks again and you’ve done an awesome job with this software, I am very grateful.
Andrew
