@Bonne Wilce I didn’t save the gcodes, but can do on monday. I’m not anymore in the office.
I used “combine pixels” so the grayscale is not jittering at 100mm/s, but the small letters was also jittering.
Would be nice if we could set up a bigger comparison study with some more people.
@cprezzi as suggested, the image is converted to 1-0’s in laserdraw and the L signal is just turning the pixel on or off. I have looked at the internals of the LPS and L does nothing to power just enables the PWM setting from the pot.
I have a goal to look at the composite inside the LPS this weekend if I get honey-do’s done.
Can you post your photo above as jpg, for some reason when I click on it the top is clipped and can’t blow it up.
I would love to see us come up with a test file (in gcode so that there is no conversion or loading problems) and print that on every machine and creating a library.
We can also use that test pattern for troubleshooting machines.
@donkjr The picture above is jpg. When i click it, it opens full screen and I can download it as jpg.
@cprezzi quick question concerning your Moshi test… Was the image dithered prior to running your test?
I’ll post some things tomorrow @cprezzi around 40 mm/s we can raster without jitter. Our machine does not have a pot so we have it all digital. The tickle is set to between 0.5 and 0.7 trying to find a good ballance. Need to check what amp that is wrote it down somewhere.
The issue is a co2 laser is too much power so we need the speed to compansate :s
@Bonne Wilce so are you perhaps saying that the lowest power at which the laser ionizes is still to much power to engrave, so speed it up to reduce the surface exposure? Mmmmm
Indeed. Its all about power density dithering works because you are reducing the density of where the laser fires.
Laser web smoothie changes the power of the laser. I’m not 100 percent sure if the Chinese boards do this? I know the ones i visited local to me can’t adjust their power while engraving.
If we can just fix the jittering and raster at decent speeds i don’t think we will have any issues. But this is a hot topic to say the least 
@Bonne Wilce I’m pretty sure that the K40 cannot control pixel power because there is only one connection “L” to the LPS which is an enable not a PWM or analog input. In smoothie configs we are controlling pixel power and on-off with one control signal.
I had not thought of the possibility that the laser at its lowest effective ionization level has to much power for engraving and in our configuration will saturate the PWM control from smoothie.
This may be the most useful reason to separate pixel on-off from power control. Or, as you say achieve mechanical stability at higher speeds.
Certainly dithering does not have this problem as it is on-off patterns:
I think the LPS has just one control loop which combines/integrates the voltage of the pot with the pwm on period. The question is, how fast the loop response is. A slow loop will limit the pwm frequency, which also limits the max. feed.
The PWM frequency should be faster or equal as the stepping frequency.
If the head moves too fast, the laser makes a line during a pwm on period, instead of summarizing the energy in one point. That would explain why we don’t get much levels of gray.
@cprezzi the PWM period ideally needs multiple cycles of power control (pwm periods) per a pixel time.
So you want to increase the PWM frequency. However the PWM period cannot be so short that when engraving the smallest DF (5%) creates a pulse that is so fast the laser cannot ionize-extinguish in a pixel time.
This is the impossible optimization of PWM and granularity of grey shades that I refer to using a single control signal.
If you speed up the machine the pixel time in the controller should be proportionally reduced so the size of a pixel stays the same? Therefore a pixel should not change size. Correspondingly you would need to increase power to keep exposure the same.
Certainly you can overrun the controller with to high a gantry speed, I don’t know what that is for smoothie.
However no matter what speed the controller tells the laser to turn on-off the laser can only respond so fast. The data I have seen suggest the ionization-to light output time could be as large as 300us. If that is true its impressive we are getting the results we are.
When considered in isolation the limit of the lasers response suggests that slowing down should help. However if there is this overpowering effect @Bonne Wilce suggests where we cannot get the power low enough then speeding up helps lower the exposure power but is the wrong way to go for laser response… a catch 22
Now you know why “my head has been hurting for so many months”.
@donkjr As I wrote early in this thread, I do get the same effect of overpowering like bonne. To overcome that, either I need to speed up, which is limited by the usb communication and smoothies protocoll, or i need to additionally reduce the power with the PWM signal, which also reduces grayscale resulution. See the dilemmma?
I think it’s worth testing dithering.
On our lasersaur in the FabLab @Martin_Renold has written a pulseraster branch (https://github.com/martinxyz/LasaurApp) that does dithering, which works very well and the result looks great. The pixdel density is so good that you don’t realize the pixels from a few cm distance.
