Bad accuracy with my cnc machine

[masinecc]

Ok guys, i really need help with this. Hope some more experienced could give me some advises what to do to get more accuracy with my machine. Here i am sending you a picture from the tests i did.CIRCLES -(1) lower left 10.47-10.70,(2) lower right 10.50-10.60 ,(3)center 9.76-10.90,(4) upper left 10.25-10.40,(5) upper right 10.12-10.70
Every next circle is deeper from the previous one 1-st one 5.4mm-last one gets up to 9.5mm(program is 5mm depth for all cirles, i check my table is parallel )
SQUARES:
1-st square (lower left corner) x=10.35 , y=10.39
2-nd square(lower right corner) x=10.38, y=10.40
3-th square (center) x=10.50, ( this square doesn look like square like the 1-st and the 2-nd one, x=10.50 lower and upper side of the square, Y-left=9.97, Y-right =10.43)
4-th square(upper left corner) x=10.49, Y-left =9.97, Y-right=10.50
5-th square(upper right) i didnt make it coz the tool went too deep in the wood and i was afraid not to broke the tool, but i ques the results would be similar to the 4-th and 5-th.

[routercnc]

The easiest way to get that level of distortion is to have a weak machine, especially in one axis.

This can be difficult to resolve because even if you use a very low feedrate (which is not great long term anyway) you can get pulling of the bit in the material if the machine frame stiffness is low.

If you lean your weight on the spindle can you deflect it?

I don't think it is lost steps etc.

That would be my first guess anyway. What machine is it? Can you post pictures of the machine?

[masinecc]

Yes, true, my x-axis is not rigid enough,, its too flexing. Thats what i think, thats where the problem comes from. I already ordered a new sbr 16mm rails. Here are some pictures from my machine http://www.mycncuk.com/attachment.php?attachmentid=15177&stc=1http://www.mycncuk.com/attachment.php?attachmentid=15178&stc=1http://www.mycncuk.com/attachment.php?attachmentid=15179&stc=1http://www.mycncuk.com/attachment.php?attachmentid=15180&stc=1

[Boyan Silyavski]

I do a test 50mm square and 25mm circle inside, profiled from the outside. This dimensiones are easier to measure. I use phenolic paper board as it does not change dimensions after the cut like MDF, wood and some plastics do.

The way you did the test tells me really nothing.
-may be you did not measure it properly as its very small piece
-the MDf is getting bigger just 1 min after the cut.
-such small pieces can show small deviation which rapidly could translate to much bigger when a 10cm piece is cut.


Obviously your machine far from strong. But even that could be compensated to some extend in Mach 3 via the step calibration
/example - http://www.instructables.com/id/Automatic-Axis-Calibration-Using-Mach-3/ or search google and read the Mach3 manual.

Now the problem with compensation such a machine lies that every time you change depth of cut and material cut you will have again to do that. I have small belt driven machine where i have 3 different screens with different calibration, one for plastic, one for wood and one for aluminum. That all to compensate the belt stretch for different materials.


Why i do a test circle and a square. The circle is obvious, the square is to help identify the specific problem. Some times the square will be Ok and the circle not. Funny, yes. That means the plunge rate is too much for the weak machine and the depth of cut is way to big.

http://www.mycncuk.com/attachment.php?attachmentid=12773&d=1405355037

http://www.mycncuk.com/attachment.php?attachmentid=12776&d=1405355150

[routercnc]

SBR16 rails will be an improvement and for wood cutting will be OK. BUT, the gantry and frame of the machine also needs significant upgrading. You will not gain anything just by changing the rails.

Have a look at the build logs. Here is an example of my wood cutting machine:
http://www.mycncuk.com/threads/6988-routercnc_MK3

Here is one for cutting aluminium:
http://www.mycncuk.com/threads/2148-Comments-sought-on-new-build-A-CNC-Router-for-RC-Gliders-and-Planes/page24

[masinecc]

And what is ur explanation for plunging further and further into the material with every next circle? Could it be bad cracked CAM program( i use Aspire 3.0) ??! My shielded cables are not grounded yet, could this cause a missing steps?!

[Boyan Silyavski]

Bad cracked program/Aspire/ if it works ok first and then at certain point start doing it. Or mach3 needs fresh reinstall.

[masinecc]

I ll reinstal both, and see what happens.

[masinecc]

And what are ur advices for upgrading my machine.Beside changing the rails what else would help improving the machine?

[JAZZCNC]

And what is ur explanation for plunging further and further into the material with every next circle? Could it be bad cracked CAM program( i use Aspire 3.0) ??! My shielded cables are not grounded yet, could this cause a missing steps?!

This won't be copied software or Mach3 fault. It will be one of 5 things.
#1 Mechanical issue so slipping coupler or something.
#2 Electrical noise from spindle affecting signals.
#3 Incorrect setup of Steps per milimeter
#4 Pulse signal is set wrong in drive so changing on wrong side so losing 1 step on every directiuon change.
#5 Low power parallel port voltage so not enough voltage differential between high and low signal causing lost steps.

My suggestion is first check all mechanical connections and wire connections. Then identify if noise by running G-code in air without spindle running. Measure starting height and check when finished.
If ok then check the Active edge set in the drive.

[davegrennan]

In your original pic of the cut parts the notch where the depth changes is a classic sign of machine flexure. I had the same thing with my machine till I beefed it up. If I were a betting man, I'd say your gantry sides are flexing and also the z axis is either flexing itself or flexing where its attached. I suggest that you examine each park in turn while cutting and identify the problem. As JAZZ mentioned, check the couplings too.

[masinecc]

This won't be copied software or Mach3 fault. It will be one of 5 things.
#1 Mechanical issue so slipping coupler or something.
#2 Electrical noise from spindle affecting signals.
#3 Incorrect setup of Steps per milimeter
#4 Pulse signal is set wrong in drive so changing on wrong side so losing 1 step on every directiuon change.
#5 Low power parallel port voltage so not enough voltage differential between high and low signal causing lost steps.

My suggestion is first check all mechanical connections and wire connections. Then identify if noise by running G-code in air without spindle running. Measure starting height and check when finished.
If ok then check the Active edge set in the drive.

I have already check all the couplings, they seem to be good. How do i check if pulse signal is set wrong in drive? Yes there is some noise while running in air, some strange noise like clicking. Starting point is not the same, its like 3-4 mm deeper when the program finished. What does active edge means, what is it used for?
Thanks for your help.

[JAZZCNC]

How do i check if pulse signal is set wrong in drive?

Depending on your drive it will have either a switch to change the Active edge or software setting. If not then it will be fixed and the documentation should state the rising or falling edge.


Yes there is some noise while running in air, some strange noise like clicking. Starting point is not the same, its like 3-4 mm deeper when the program finished.

I don't mean noise has in sound.! . . . I mean Electrical interference noise EMF from other components around like Spindles etc. Looking at the condition of your wiring with motor wires unshielded and probably poorly grounded control box then it's a Huge chance this is your problem.!


What does active edge means, what is it used for?

This means the point when the drive issues the Step signal. It can be the rising or the falling edge of the signal Pulse. (also some controllers look for the Step on certain edge.)
If your drive issues the Step pulse on the falling edge of the pulse and your controller is expecting seeing it on the rising edge you'll have timing issues. Often when active edge is wrong you'll loose 1 step or micro step for every direction change.

One common mistake made is that people will wire the motor phases wrong. ie: Put A+ to A- on the drive connection. Then when the motor spins in wrong direction they just change the Active hi-low state in Mach motor outputs and it spins in correct direction.!! . . . BUT this now means the control (mach3) is looking for the Step pulse on a different edge and if this edge is the opposite edge to drive then you could have timing issues resulting in loosing position.
The correct way to change an axis direction if needed is using the Reverse option in Home and Limits.

If you wire the motor phases correctly to the drive. ie: A+ on motor to A+ on drive then 99% of time the motors will spin in correct direction and will match the Active edge set in the drive. This is often the rising edge but some drives like the Geckos do use the falling edge in which case if the motor phases are set correct then you would set the Active step edge in motor tuning to match. ie Step low active ticked.

Another common cause of positional lose with parallel port revolves around timing issues and the Kernal speed set in Mach3.?
The pulses created only have a certain time interval between changing state and the higher you have the Kernal speed the shorter interval. So if you set the Kernal speed high you shorten this time interval so combined with high motor tunign it's very easy to lose Steps. This is why it's recommended with parallel port to keep Kernal speed set at 25Khz for most stable performance. Every change up in Kernal speed increases chance of unstable system.