Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: SMA on July 03, 2011, 10:01:27 PM
-
Hey Guys,
As I have discovered chasing down lost steps is very difficult. Worked through mechanical issues and ended up cleaning up some minor binding issues etc. I have quadruple checked mechanical stuff. Even swapped Y and Z to eliminate a bad drive/transmission and issue still stands.
Z loses steps. Measure in real world it adds up to 3mm. This is over a variety of sizes of files. In this case I am cutting surfboards and there are more lines of code for a board that is 6' than 7'. No matter the size of the file there is a 3mm discrepancy at the home position from start of file to end of file.
This would not normally be an issue if the discrepency was throughout the board but it all happens during the cross over portion of the file where one side of the board is complete and the tool moves to the other side. The command is xo yo zo f1000. Then the machine proceeds to cut other side of board. Only in this case z does not come to 0 in real world instead it comes up about 2.5mm short. This is an issue since one side of the board ends up thinner than the other by, you guessed it, 2.5mm. This is not the end of the world but is irritating.
I slowed the cross over F com to 100 and watched to see where z gets to and it comes up 2.5mm short xo yo zo. Although the total mm off at home is 3mm but losing .5mm over one whole side is normal. Just seems like it is a crossover issue.
I have slowed motors settings all the way to 50% and still get a 3mm discrepancy. I am missing something? Is it a settings issue.
There are two files to a surfboard. A top and a bottom. To make things even more confusing the issue only arises with the top file. The bottom file only loses a mm max. When cutting the bottom file Z comes up about a mm short of z=0.
I have included a spreadsheet that includes my settings and the different speeds that I have tried running the drives at. I have what I call a z test file. It just simply moves Z up and down so may times. I measure position at beginning and again at end of file relative to a fixture that is fixed in place and get a .5mm diff everytime I run the file. Two cycles of the file equals 1mm. This is the case no matter the F com or Motor settings.
My issue is this. How can there be the same 3mm discrepancy over different number of lines of codes if it is step loss? How can I run at all these different speeds and motor settings and get the same .5mm for all speeds since step loss can usually be overcome by slowing machine.
sorry for the book
Stefan
-
Stefan,
Returning to the surfboard - when you move from one side of the board to the other (with Z raised clear of the work) does the Z axis 'tick' steps during this crossover move ?
Tweakie.
-
Hey Tweakie,
Thanks for the reply. I am not sure what you mean by ticking. These motors are microstepping sort of "humm" in my mind. I will run the file again and see if I can make out a ticking noise.
S
-
Well I had a look around for what this ticking might sound like. It sounds like it is pretty obvious and so I would say no to ticking. In my Gcode generator I can control the F command for the crossover. When I run it low, like F200, all I can hear is the rail bearings quietly rolling away and the even humm of the motor. As well I did swap Y and Z but it made no difference. Hopefully I am answering the question?
Thanks for the reply
S
-
OK, At what point during the crossover move do you suspect that the 'extra steps or lost steps' are being generated ?
Perhaps if we know when they are occurring a solution may be easier to find.
Tweakie.
-
And this is the crux of the matter. It is a long file at about 6000 lines of code. As it cuts it is very difficult to make out where the steps may be lost. I found a couple of "rough" spots where some binding was occurring but they have been rectified and now all axis move very smoothly. I am almost 100% sure that the lost steps are not due to resistance. I am not even sure that it is due to lost steps.
I thought someone might spot something in the motor settings or the file itself that I have not. When you look at a problem to long you start to just stare and don't change perspective enough to look at it from a different angle.
What I have observed is that when Z comes up to "zero" on the crossover it does not get to "zero" but rather comes up 3mm short. I can observe the cross over maneuver at F100 and watch as Z climbs up but comes short of the mark by 3mm and plunges back down to start the other side. My mind is that the crossover is the issue. There is nothing hard stopping it.
S
-
Maybe a soft limit. Not familiar with them as I don't have any set up. I know the parameters of the machine, can see it, and design well within its limits
I am pretty new, only had the machine a month, and so a bit green. I learn fast though and I am willing to take the time to read.
Here is a screen shot anyway.
S
-
Sorry to keep plaguing you with questions.
What I have observed is that when Z comes up to "zero" on the crossover it does not get to "zero" but rather comes up 3mm short.
Is that 3mm short on the Mach screen Z Axis DRO or 3mm short in physical distance when the DRO correctly reads 0.0000 ?
Tweakie.
-
The DRO reads 0 but I have a laser level and a bit of measuring tape that the laser level indicates on. When I home the machine I note the postion of the laser on the tape and then as it crosses over I watch closely, with a slow setting like F100, and see that it does not reach the same level as the laser indicated at the beginning. I have attached a photo of this rig. Works well for seeing how many mm's out at the end of a file. Y is usually off about .5 to 1mm. Perfect in the surfboard world !
S
-
Questions are good. Make me think about all that I am doing.
-
Thanks. I just hate having to ask tons of questions without offering some form of suggestion.
So, Your Z is physically short on dimension when coming back to zero and Mach is OK. I would guess at two possibilities;
1) Missed steps and the cause being that the motor does not have enough torque to raise the axis under all conditions. This could be stiffness in the axis (which you have already eliminated) or the motor drive electronics (drop in voltage from the PSU, insufficient current setting, etc).
2) Something physically loose (motor coupling, leadscrew nut, etc.)
Don't know if this helps.
Tweakie.
-
Hey Tweakie,
This machine is belt driven. Had some new belts come today and replaced them on z and X. Checked the movement of Z and it is free and clear. Slides easily fully up and down. I am pretty sure it is not mechanical within the motor since i swapped z and y with no change. Exactly 3mm short of zero.
I am not sure how to test PSU. I did try changing the Dir Pulse to 5 from 2 but it did not make a difference. Kinda at a loss at this point. Very strange how it is always 3mm short of z=0.
Thanks for the help though !
S
-
I can appreciate your problem and the difficulty your having.
I had a heck of a time trying to use my mill to do precise 3D machining and could not figure where those steps went or where and when it was skipping.
Like your code there were small increments of XYZ moves and changing from exact stop to CV or any other settings just wouldn't help.
Drove me nuts since the machine otherwise worked well for all the other stuff i have ever done with it.
So here is what i did:
Rebuilt the X,Y,Z axis ....new motors of greater torque, new thrust bearings, and meticulous adjustment of the ball screw nut and thrust bearings.
Now no backlash to deal with and feel very confident on the mechanics and controller side of things.
Now after 10000 lines or more of code at much higher feed rates and rapids, the out of position is .001" at the end of the program.
I think the major cure on my end was the new stepper motors which could deliver the acceleration and velocities required by the program.
Of course the other mods and adjustments provide for repeatable accuracy.
Even though i have the same motors,same gear ratio, same type / make of ball screws, same type belts and timing gears each axis has a different steps per unit
value ( maybe somewhat small but still different ) and I don't find that unusual.
Not saying you need to do the above, but, 3d maching places a higher demand on the system.
RICH
-
Hey Rich,
Thanks for the input.
My machine is brand new. Has less than 4 hours of machining on it and the builder has advised that the motors or over kill for what we do with the machine.
I feel like it is a setting in Mach. Some sort of offset that I don't know about. Lost steps should change with motor speed setting adjustments but even speeding the machine up, both speed and rate of acceleration, leads to the same result. I think I am missing something in settings?
S
-
I understand your frustration S but your machine should follow the instructions it is given by Mach and be in exactly the position the Mach DRO says it is. When there is a discrepancy, as in your case, then for some reason steps have been lost (or gained). Gained steps are often caused by electrical noise pickup but steps gained in one direction can appear as steps lost when the axis travels in the opposite direction. Lost steps are often caused by insufficient drive torque or mechanical problems.
As your machine is belt driven, could the belt be slipping or jumping a tooth on the pulley ?.
Have you checked the tightness of all the grub screws associated with the drive train to the Z Axis ?.
Tweakie.
-
I am new to this but I agree that a machine does nothing without instruction and then only what it is told. Learned that pretty quickly and how true it is. It is strange how it does not achieve Z0 when crossing over but what gets me is that not matter then lines of code it loses the same amount of mm's (I am not sure how many steps this would equal) everytime. You would think more lines of code means bigger discrepancy in result. Coupled with motor setting changes you would see some significant changes. This is just not the case. Always 3mm off at home.
I just replaced the belt yesterday, just to be sure and continue to eliminate all possibility of mechanical issue. Swapped Y and Z and in that process made sure the motor was well secured to the Z axis and as well that the Gear was well aligned to the idlers. I loosened the belt a bit to reduce binding at idlers and conversely I tightened a bit more to see if slippage was the issue. Idlers spin freely and true and belts are snug.
It is kinda like an odyssey without really going anywhere :)
-
It is kinda like an odyssey without really going anywhere
You are right there.
What are the steps per mm that you have set in motor tuning for the Z axis ?
Tweakie.
-
Just answered my own question ;D
OK just looked at your .xml and the steps per mm for your Z Axis are only 83.333. Your velocity setting for this axis is probably way too high for this ratio thereby reducing the available motor torque considerably.
By comparison, for my Z Axis I use a velocity of 1500 but my steps per mm are over 500 and that represents a reduction gear ratio of around 6:1 over your 83.333.
I think you should reduce your Z Axis velocity to around 1000 or less and try again to see if the error still persists.
Tweakie.
-
Hey Tweakie,
Thanks for posting again. I'll try anything to get his figured.
I ran the file through and same issue. I slowed z down to 1000. No change. Still 3mm over the file and z still does not make it back to 0 on the crossover.
Weird and weirder.
I am reading like crazy about MACH to see if there is a bad setting since I keep getting the same result. 3mm.
Pulled all belts off one more time to be sure there are no binds and things are all good.
S
-
And yes I did save the settings and even opened up setting tab to be sure they changed. You don't need to reboot Mach for Motor Setting Changes do you?
-
And yes I did save the settings and even opened up setting tab to be sure they changed. You don't need to reboot Mach for Motor Setting Changes do you?
Sounds like you did it right.
If slowing the Velocity to 1000 made no difference then it blows my theory out of the water.
If it was a problem with Mach itself then I am certain it would have also been discovered by others, after all thousands of professional and semi-professionals use Mach on a daily basis.
Could you perhaps post or email me a section of GCode where this problem happens so I can try to duplicate the same error here ?
Tweakie.
-
Hey Tweakie,
Ok so I attached the crossover with the four lines of code before and after. I feel like I am overlooking something, as I am sure many feel the same, and that it is something simple. I have dealt with the difficult things with no results.
To add to it all I cut a board this evening that was only out about 1.25mm. this is well within reason and would be fine if that is all it was ever out. The interesting thing here is that the crossover happened at the tail and not the nose of the board.
Either way here is the bit of code.
The search continues
Stefan
-
Brave try Stefan but for some reason the code did not attach. (try zipping the file perhaps).
Tweakie.
-
Hey Tweakie,
Classic case of forgot to attach.
S
-
SMA,
Confirm your steps per mm using a dial indicator and not the laser since that is not accurate enough for checking the step value.
You are consistant in the amount you are out and swapping axis implied that it is not a drive problem / signal path problem.
- Test that a simple movement of Z is consistant , post reversing the direction ...both up and down movement.
set your steps per, the movement should agree with DRO display ( not using the laser please.......)
- Now check for backlash / non movement on reversal
RICH
-
Hi Stefan,
Don't know if this helps but I couldn’t get anything, error wise, from your crossover file so went back to look at your earlier files.
I am trying to understand your “surfboard file top” and “surfboard file bottom” but the conventions used are rather unusual, at least for my machine.
You seem to have the X axis as the long axis and Z zero at the bottom of the work rather than the top. The final GCode lines of taking Z to zero, switching off the spindle then returning to X0, Y0 would break my cutter but you seem to have your Z axis reversed.
By convention +Z is usually upward (away from the work) and –Z is downward (towards the work).
I ran this file 4 times without error (just cutting air). I had to exchange the X and Y axis in the code so it would fit my machine table (the Y axis is usually the long axis by convention) and scale it down a bit but other than that it ran OK and my machine returned exactly to Z zero each time (using feeler gauges to check).
The GCode compilation is generally weird particularly with the jumping feed rates from F200 to F13000 for no good reason and there is a small Z axis glitch halfway along the long axis in both directions which I cannot understand so I would only score this 4 out of 10 for nice GCode to use.
Tweakie.
-
I checked to see that steps per mm is accurate and it is (not using the laser). What is interesting is that I changed steps per mm and ran file again and still was out about .5mm per cycle of my z test file which just moves z up and down 124mm. Change motor speeds, F commands to 500, step 5 dir 5 and again no change. Exactly .5mm. Very weird.4
This is minor now since my new problem is the machine just not working at all. I started a new post about this. Man it is like I, or the machine, are cursed.
S
-
Glad you have a sense of humour S ;D
Tweakie.
-
Hey Guys,
Still losing steps. As requested I am going to post details of my motors and a picture of my machine.
The issue remains the same. Z axis loses steps even when slowed to 50% of it's usual operating speed. Y loses exactly half as many steps as Z and X loses none. X is doing most of the work but it is not overcoming as many g's as Z (considering gravity).
The gearbox is a 5:1 Ratio
Ananheim Automation
GBPH-0601-NS-AA231-375
Back Lash 10 ARC Min
The Stepper Motor
60HS100DE06B
Gecko Drivers
The Mach Version is the latest Version R3.043.022
I am not sure about the power supply but I am finding out from the designer asap.
What is weird is that steps are lost no matter the motor settings. Slow it down and you get the same result as when you speed things up.
Everything mechanical has, at this point, been gone over far to many times to have missed something.
I am still of the mind that it is a setting issue but I am a babe in the woods. The reason I think this is due to the consistentcy of loss.
I have posted the file that I have been using to test Z. It just moves Z up and down several times. As posted before I end up with the same step loss not matter the Fcommands or speed settings of the motors.
sorry for the distance between posts. i was trying to get caught up and keep customers happy.
S
-
Hi Stefan,
A wild guess - your little Z Axis gearbox is jumping a tooth.
Tweakie.
-
Stefan,
see attached ???
RICH
-
Good Eye. That is an idler wheel but nice spotting.
I have come up with a new solution that will fix the issue that does not involve chasing steps anymore.
I have attached a bit of code that is the "crossover" portion of the surfboard file. As it is Z (Z=0.00) comes back up to home and then crosses over to cut opposite side of board. When it does this it comes within 9mm of the Z limit switch. My problem would be solved if instead it went all the way and touched off the limit switch then come back to home and then continue cutting. Problem solved.
Question is this. What kind of command should I use to make this happen. I have attached what it looks like now and if any of you fine fellows could instruct me as to what the string of code that would work best.
This remedy will allow Z to "find" it's place after one side of the board is complete since I am of the mind that it is the crossover that is the cause of the issue. I don't konw how or why but I will use coding as a solution for now.
S