Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: stephanisk on September 11, 2017, 07:49:01 PM
-
Hi ,
I have something that has been bothering me for months now.
When I am milling splines and circular movements like helix ramping my spindle always moves kind of like in shocks and not in a smooth movement which leaves marks on the finished parts, also it slows the process time. It is like when he mills splines after every line of g3 code the spindle pauses for half a second to continue the next movement. It drives me crazy.
In settings i have read ahead 200 lines on, so why doesn't he know what the next movement is going to be.
(Servo motors, leadshine drivers, )
I really hope someone has an answer to this, or is this an inherent problem with mach3 that a smooth movement in helix ramping without hesitations isn't possible? Cause than I have to look for another system. But I like mach3 so I hope there is a solutions to this.
Thanks in advance all of you
-
Hi,
do you have backlash compensation in operation?
Craig
-
Hi Craig,
Where would I find this? This is a setting I don't know where to find.
Stephane
-
Stephane,
You will find it in Config>Backlash.That is where you define the backlash, speed , and if you want to enabled it.
RICH
-
And are you in CV mode?
You should be in CV mode, with ALL CV options turned off, except "Stop CV on angles". Turn off all CV options in General Config, and CV Distance and CV Feedrate on the Settings page.
Make sure you are seeing G64 in the status line when the code is running.
-
Hi,
I will be looking at what you guys told me tomorrow morning. I wasn't able to boot up the machine today. To busy, I really hope one of these answers will solve this problem for me. Thanks already, I will update tomorrow
-
Hi,
I was thinking that backlash, or at least backlash compensation could cause the problem you describe.
The machine will momentarily halt and 'back up' an axis that reverses direction during an arc say.
Backlash compensation would ideally be off and and of course the acutual backlash should be zero or close
it.
Craig
-
I have tried all of the things mentioned and I got it to jerk a little less. By upping in config -> backlash the speed from 20 to 100% (everything else is on O and backlash enabled is not vinked. Also changed CV for angles to 90 it was on 6 before, all other CV options are off. But the hesitation is still there in arcs. ( For example milling letters in baroque lettertypes)
What else could be helping out so my machine makes 1 smooth motion and doesn't slow and speeds up at every line of g3 and G2.
Hope you guys have some other suggestions.
Also the backlash compensation mentioned above... Where do I find these settings? Cause haven't found any backlash compensation settings anywhere. 2nd thing. When I activate CV feedrate and set to 100 it makes it jerk more.
Thx
-
Hi,
turn backlash compensation off. Turn CV on.
From the manual:
-
sorry, misclick, from the manual:
5.6.3 Configure Backlash
Mach3 will attempt to compensate for
backlash in axis drive mechanisms by
attempting to approach each required
coordinate from the same direction. While
this is useful in applications like drilling or
boring, it cannot overcome problems with the
machine in continuous cutting.
Again from the manual:
6.2.15.3 CV Mode/Angular Limit
This LED is lit when the system is running in "Constant Velocity" mode. This will give
smoother and faster operation than "Exact stop" mode but may cause some rounding at
sharp corners depending on the speed of the axis drives. Even when the system is in CV
mode a corner with a change of direction more acute than the value given in the Angular
Limit DRO will be performed as if Exact Stop was selected. Full details of this are given
under Constant Velocity in chapter 10.
And Chapter 10 as promised:
10.1.16 Path Control Modes
The machining system may be put into any one of two path control modes: (1) exact stop
mode, (2) constant velocity mode. In exact stop mode, the machine stops briefly at the end
of each programmed move. In constant velocity mode, sharp corners of the path may be
rounded slightly so that the feed rate may be kept up. These modes are to allow the user to
control the compromise involved in turning corners because a real machine has a finite
acceleration due to the inertia of its mechanism.
Exact stop does what it says. The machine will come to rest at each change of direction and
the tool will therefore precisely follow the commanded path.
Constant velocity will overlap acceleration in the new direction with deceleration in the
current one in order to keep the commanded feedrate. This implies a rounding of any corner
but faster and smoother cutting. This is particularly important in routing and plasma cutting
The lower the acceleration of the machine axes, the greater will be the radius of the rounded
corner.
In Plasma mode (set on Configure Logic dialog) the system attempts to optimise corner
motion for plasma cutting by a proprietary algorithm.
It is also possible to define an limiting angle so that changes in direction of more than this
angle will always be treated as Exact Stop even though Constant Velocity is selected. This
allows gentle corners to be smoother but avoids excessive rounding of sharp corners even
on machines with low acceleration on one or more axes. This feature is enabled in the
Configure Logic dialog and the limiting angle is set by a DRO. This setting will probably
need to be chosen experimentally depending on the characteristics of the machine tool and,
perhaps, the toolpath of an individual job
If you haven't got the manual go to the Artsoft site and get it. Manuals work best when they're read, usually several times!
Craig
-
2nd thing. When I activate CV feedrate and set to 100 it makes it jerk more.
Which is why I told you to turn it off.
-
I tested it and yes I have put if off. But still not a smooth curve
-
Hi stephanisk,
Could you .zip and post a copy of your problem Gcode ?
Tweakie.
-
Manuals work best when they're read, usually several times!
Better make that 'Many many times' for Mach3!
If you are using subroutines there may be a slight hesitation between subroutine calls - but it is quite slight.
Sounds like a CV problem to me, OR possibly flex in the framework.
Cheers
Roger
-
Hi Tweaky
Here you will find the G-code (in txt-file, this forum doesn't allow me to post it in .zip)
Greetings
Stephanisk
-
Hi Stephanisk,
I have just run your Gcode here (I had to reduce it to 0.25 scale to fit on my engraving machine) and my machine motion appears to be smooth between Gcode line transitions etc. so I think that rules out the Gcode as being the problem.
Could you attach your <profile> .xml file (copy and rename it to a unique name such as stephan.xml before attaching) then we can check your Mach3 settings.
Tweakie.
-
hi,
here is my mach3mill xml file. I hope this is the one with all the settings. I do believe it is :)
Thanks for all your help so far. Hope we find the problem.
regards,
Stephane
-
Hi Stephane,
Unfortunately I am unable to test your .xml settings - I don't have an XHC controller and your settings will not run under Mach3 parallel port control.
Could you try reducing your X and Y axis Velocity settings by 50% (in motor tuning) and setting the Acceleration number to at least 10% of the Velocity number (save the changes) - does this make any noticeable difference ?
Tweakie.
-
Anchor a dial indicator to the spindle and mount a smooth block on the table. Now measure how much flex you get when you put a load on the spindle of 1 - 2 kg, on each of X & Y axes. Having measured in one direction, try the other direction - you wiol get both flex and backlash of course.
Then repeat while stressing the table instead.
Cheers
Roger
-
Hi Roger,
Thanks for your input. What is it that you want to measure? The rigidity of the machine?
The whole machine weighs about 3000 kg. The bridge itself is about 1200kg. This is about as heavy duty as a flatbed router can be to be honest. U could mill some steel with this if u wanted too. So there is no actual flex when I put pressure on spindle or table. The dial says 0. ;)
Regards
Stephane
-
Hi Stephane
OK. Sometimes hard to tell whether we are dealing with a 3 ton machine or a Chinese aluminium desktop...
Next bright suggestion: can you dive into the bowels of the config data and work out what the single step distance is? Some maths required. On my machine it is about 0.8 microns.
Then try commanding moves in steps of about half that in both directions. We are looking for a loose pulley on the motor shaft here: you probably would not see it with the previous test. See how many microns you have to move in one direction after going in the opposite direction before the DI flickers. Yeah, boring stuff, but the results can be illuminating.
I remember one poor guy who found that the grub screw going into the dimple on the motor shaft was really loose. He soon fixed that!
Cheers
Roger
-
Hi,
I've done the same sort of thing on my mill.
The stepper/ planetary gearbox/ ballscrew resolution is 1um. The manufacturer of the planetary gearbox claims less than 2 arc min backlash equivalent
to 0.7um lash.
When I did the same measurement that Roger describes I got 4um lost motion? As it turns out I'm using some cheap aluminium shaft couplers with a spiral
slot cut into them. They wind and unwind a bit like a spring and there is my lost motion. I have retained those things because they are a bit weak, I've had
the occasional crash and the coupler twists itself up and shears rather than some other component breaking. While I didn't intend it they serve as a mechanical
fuse.
With the gearboxes and mechanical advantage of the screws the stall thrust of my axes is over 1500 N or 150kg thrust, enuf to crush things like hands and so
on. This machine is tiny, machining volume of 200mmx200mmx200mm, and yet its till got the grunt to injure badly. Was quite a surprise when I realised that.
As I've said before 'some surprises are surprisingly more surprising than others'.
Craig
-
some cheap aluminium shaft couplers with a spiral slot cut into them. They wind and unwind a bit like a spring and there is my lost motion. I have retained those things because they are a bit weak, I've had the occasional crash and the coupler twists itself up and shears rather than some other component breaking. While I didn't intend it they serve as a mechanical fuse.
Good analogy, good idea.
Yeah, belt reduction and ball screw drive - beats a Juicero easily.
Cheers
Roger
-
These last suggestions I am a little lost on. What exactly do you wante to measure?
My machine is made of servo with planetary reducer (gearbox) in front . Only my z axis is directly placed on ballscrew. This one I remounted once, there are no loose screws there. On the motors also every screw is supertight.
So I am a little lost here guys.
Stephane
-
Hi Stephane
We don't even know what your machine looks like, but we are trying to help anyhow. We can be reasonably sure that if we come up with 10 suggestions, 9 of them will be wrong.
What I was suggesting is a way of seeing how much backlash there is from the motor end. With a fine pitch ballscrew, well mounted, you might have zero backlash when measured from the mill table, but you could stll have backlash from the motor end.
Cheers
Roger
-
Ok, I see.
Backlash is really small, this I can tell you, measuring it, I wouldn't know how to start
-
Hi,
I don't hold Chinese made gear in high regard but I think in this case your concern is unfounded. Where Chinese made stuff falls down is that certain
features like backlash comp or probing don't work and the companies don't seem to be bothered to upgrade their plugins. Also their
documentation is often execrable most probably due to language and they don't seem overly concerned that their customers are struggling.
They do work tho...and any well constructed pulse stream is just as good as any other well constructed pulse stream be it Chinese constructed or
otherwise.
Craig
-
Ok, so we can take out the card as the reason. So we are back to the settings
-
Hi,
I hope I haven't put you wrong there...I use an ESS and so cannot say for sure that your board it not the cause of the complaint, just that I don't think
its likely.
Additionally with all those units in service do you not think that others would have commented if they had the same problem?
If you don't mind me asking but a 3000kg machine is pretty damn big and presumably must have steppers/servos and spindles to match and be worth
tens of thousands of dollars. If you are of the opinion that the controller is questionable why not buy an alternative like an ESS with a know reputation,
about $190 USD plus shipping sounds pretty minimal in comparison to you machine.
Can you post a pic or two?
Craig