My laser engraver from 0 knowledge to 32.461% .... probably

[tweakie]

Hi Bobby,

Looking at your video it appears that you have a problem with your laser triggering circuit. It seems to be triggering late rather than at the exact moment of axis movement and it also appears to drop-out during axis movement.

Tweakie.

[BobbyW]

Hi Bobby,

Looking at your video it appears that you have a problem with your laser triggering circuit. It seems to be triggering late rather than at the exact moment of axis movement and it also appears to drop-out during axis movement.

Tweakie.

Yes i thinking on that too ' but on engrave no any problem at all. Also the problem seem's to be on motors tuning.
Because if i try on double size circle or ellipse, no any problem at all.
That is only at small pieces i try to cut , below 2 cm .
Trigger circuit i made is no way to have any uS delay ( and on that speed even mS don't affect the trigger ), i don't even use an integrator for that ,
the internal capacitance of source and wires are enough to have fine analog input. I use a good OP and a MOS optocupler.
Right now i don't even know where to start to debug , obviously no any experience with that  in the past.
I will made more test from 0 . Also ,  maybe trying to change the driver parameter's of the motors ?  ;D
Thanks

[tweakie]

If you .zip and post a copy of the Gcode which exhibits the problem (shown in your video) then I will run it here and at least rule out one possible cause.

Tweakie.

[BobbyW]

If you .zip and post a copy of the Gcode which exhibits the problem (shown in your video) then I will run it here and at least rule out one possible cause.

Tweakie.

I try to edit my post , you already answer .
EDIT later :
If i try to set the power very accurate , now i can do it on one single shape and no smaller than 2 cm . I can set to cut a paper without even to pass below ( cut just the paper and he didn't smoke , just cut)
If you notice in video , is a ellipse , but is 3 cm piece . If i draw the second one half of the size and i try to cut , the result will be like in my first video you watch.
I made a small vid to see what happen . To say simply , more smaller than 2 cm the piece will be , more inconsistent the power will be and also , if i mixt different shapes on different sizes
the power will be a mess .
https://youtu.be/U8SnTyCUAt0

And the code  is simply made by " vexx " . Nothing special , you can download what Art post some post below ( i'm at work , not home now  ::) )
Also if i try to download and run the code he attach , tle laser will burn hard the big circle and on small ones , dont even shoot.
I think my problem is acc and velocity setting .

Thanks again

[ArtF]

Hi Bobby:

  I can see a problem or two in your tests..

All your boxes show on takeoff the laser isn't shooting at the start..

This may explain small circles as they never get up to speed. So a couple things to discuss...

First, as the laser takes off on a move, the formula for power internally is

  max( minpwmsetting , (ActualSpeed / Feedrate) * power )

  So in the config, the minPWM sets the startup cutting power as the axis
starts to move if the axis is so slow the power computed is less than minPWM.
SO to set this approximately correct, make a square
in gcode. Cut it as you have and if there is no burn at the start of each
corner increase the setting Planner/MinPWM until you see each corner
has been cut. Set you feedrate so your table actually hits it in the velocity DRO,
(or close anyway, it doesnt have to be perfect..but if you see 1589 is the fastest
it goes, set the feedrate to 1600 or so. )

  The second problem is that feedrate and jerk are related. Jerk is defined
as the amount of mechanical jerk the machine can do, or how much jerk
occurs during a move.  As its lowered, the axis will move slower. As its raised,
the axis will move faster.If this is too low, you will not be able to move
fast enough on small segment programs. Another consideration is
look ahead. If a circle is made of hundreds of Gcode lines, the program
can only see ahead by the setting of Lookahead. Change this to
300 or so and it can see and thus plan for higher speeds.

If you run a program without laser on, see what max feedrate is achived in
that small circle , the math is that a small circle may never hit the feedrate
you desire, so power will never hit it as well. For example, if your feedrate
is set to 5000, and you have 2 circles, one large and one very small.

  The feedrate in the large may hit 4000 ( for example) but the small
circle may only hie 1000. If so , the power of the small circle will be 1/4th
the power of the large. In this case the feedrate for the program should
be set to 1000. (Or a feedrate for the small circle of 1000 and for the large
4000.)

  Jerk in a circle is computed as a function of its radius. The larger the
radius and mathematically the smaller the resultant jerk. Smaller circles
have larger jerk.


  So I would first look to the square and adjust the minPWM to make it
start that corner with power high enough to burn. If you turn off
DistCorrection ( button on the laser control), you will then get
the set laser power at all times during motion, which can be fine in cutting
but will cut harder when its slow due to planning motion.

  It takes some playing to get this right. Unlike normal CNC where motion
alone is enough to cut, Auggie needs to get the power to be properly proportional
to speed, which is why all the power settings.

Let me know how the squares corners come out with adjustments and we'll go
from there.

Thx
Art

[BobbyW]

If you .zip and post a copy of the Gcode which exhibits the problem (shown in your video) then I will run it here and at least rule out one possible cause.

Tweakie.

I got home and i start to play again> So i start to read again what Art say me , the entire tunning process .
I change my config from 3200 to 12800 steps , (i put the file below) and do it step by step and..... the result was the same
but at least , i got some knowledge ++  :-\
Anyway , i put below that code how you ask me , but i don't think that matter , the result are the same ,
i try a lot of programs to generate , same result , so..... i think i will keep digging 
Big square test as Art suggested
https://youtu.be/tSRGY6CA1iI
and movig test for tunning max speed and acceleration
https://youtu.be/2qOZfn9ruWk

Thanks

[BobbyW]

Hi Bobby:

SO to set this approximately correct, make a square
in gcode. Cut it as you have and if there is no burn at the start of each
corner increase the setting Planner/MinPWM until you see each corner
has been cut. Set you feedrate so your table actually hits it in the velocity DRO,
(or close anyway, it doesnt have to be perfect..but if you see 1589 is the fastest
it goes, set the feedrate to 1600 or so. )

Thx
Art

Yes , i attach some files in my previous post , we write in same time  (can see my test notes in that .txt)
So , i made the test how you say .
On beginning was the same , after i rise "jerk" to 6000000 (i was annoyed in that moment) and surprisely , he cut nice the square .
?? How much i can rise the jerk ? Exist a limit value for ?
Also i put min PWM at 20% , i have second power control at analog on min/max power and i set the power just to barely cut the paper.
I notice , the tool hit almost all time the feed set , excluding that small ellipse when decrease the power.
I put on max CV radius 0.10 (what is hes effect ?) i didn't notice any diff.
https://youtu.be/LU0qrUX1Pxw

//~~~~~~~~~~~~~~~~~~~
AND not at least , today i find out you are the creator of Mach 3 , man , a awesome work realy , i never used but , every time i look at one video about
machining , that Mach 3 remain in my mind. Again my intuition didn't fail . I have such a big respect for you .

Thanks for everything you do for as
Bobby

[ArtF]

Hi Bobby:

  Jerk can be a large number. I sometimes use 1000000 and sometimes 50,000,000,
depends on what Im doing and how it reacts. Setting Jerk very high basically reduces
smoothing, but some systems dont need smoothing so a high jerk is fine. The limit is
in the billions I think..

  That last video looked much better. If your doing an ellipse there are many small moves in
some code, so make sure your look ahead is high.. maybe 400 - 500. This helps the planner get
up to speed. Small circles calculate jerk a bit differently as they have to take radial vector
pressures into account. Its all a balance, but I think your getting there, the last video showed
your very close to what you need. Play around with jerk and feedrates and eventually it will
sink in how they affect each other.

Thx
Art

[tweakie]

Hi Bobby,

Your last video shows great improvement so you are definitely heading in the right direction.

Tweakie.

[BobbyW]

Hi ..
Thanks everyone for help. I figure out what " jerk " and other variable do . I manage to have constant power on
cutting almost on any shape i made . Now trying to script a "air button " for air assist , i found that library was already created
by @Ya-Nvr-No , and of course did't work , so . I start to look on you videos on youtube , and trying to see where are located all
variable in the system , i found some variable (including " AirAssist " ) variable , is duplicate in the my system.
That affect some how the Auggie funtionality or is just a visual bug ?
I put below a screenshot of that
Thanks
Bobby