Y Axis offset itself while running

Hello,

I am back again with a problem on my CNC build that I cannot solve by myself

When I travel my Y axis backward and forward, each time it travel it offset itself a little bit. I noticed it when trying to cut my table, when I saw that the back of my table has a "stair" for each depth pass.
So, I've try to run a simple GCODE test that consist of marking two point on Y, then running back and forth 10 times on the Y axis and then mark again these two points.

And after 10 travel I have an offset of about 2 mm in both direction.. See the picture. 

My Y axis INI settings :

[AXIS_Y]
MAX_VELOCITY = 25.0
MAX_ACCELERATION = 350.0
MIN_LIMIT = 0.000
MAX_LIMIT = 265.0

[JOINT_1]
TYPE = LINEAR
HOME = 10.0
MIN_LIMIT = -1.000
MAX_LIMIT = 265.0
MAX_VELOCITY = 25.0
MAX_ACCELERATION = 350.0
STEPGEN_MAXACCEL = 480.0
SCALE = 200.0
FERROR = 1
MIN_FERROR = .25
HOME_OFFSET = -1.000000
HOME_SEARCH_VEL = -5.000000
HOME_LATCH_VEL = 2.439024
HOME_IGNORE_LIMITS = YES
HOME_SEQUENCE = 1

My Y axis HAL settings :

setp stepgen.1.position-scale [JOINT_1]SCALE
setp stepgen.1.steplen 1
setp stepgen.1.stepspace 0
setp stepgen.1.dirhold 35000
setp stepgen.1.dirsetup 30000
setp stepgen.1.maxaccel [JOINT_1]STEPGEN_MAXACCEL
net ypos-cmd joint.1.motor-pos-cmd => stepgen.1.position-cmd
net ypos-fb stepgen.1.position-fb => joint.1.motor-pos-fb
net ystep <= stepgen.1.step
net ydir <= stepgen.1.dir
net yenable joint.1.amp-enable-out => stepgen.1.enable
net min-home-y => joint.1.home-sw-in
net min-home-y => joint.1.neg-lim-sw-in

The little GCODE test I did :

%
(1001)

(GOTO POINT 1)
G90
G0 X175 Y10 Z3
G1 Z0.2 F120

(PAUSE)
M0

(GOTO POINT 2)
G90
G0 X175 Y240 Z3
G1 Z0.2 F120

(PAUSE)
M0

(TEST)
G0 Z10
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000
G1 Y10 F1000
G1 Y240 F1000

(GOTO POINT 1)
G90
G0 X175 Y10 Z3
G1 Z0.2 F120

(PAUSE)
M0

(GOTO POINT 2)
G90
G0 X175 Y240 Z3
G1 Z0.2 F120

(PAUSE)
M0

(END)
G0 X340 Y250 Z50
%

Thank in advance for your help.

andypugh wrote:

This is generally either acceleration (or velocity) limits too high, step time too short or something physical slipping in the mechanical transmission.

First mark across all the pulley-to-shaft and coupling-to-shaft interfaces with a marker pen so that you can tell if they are slipping.

In the HAL file look for the parport.0.reset-time parameter. Double it and then re-try your test.

Then if that doesn't work, put reset-time back to where it was and try halving the max_velocity and max_acceleration in the INI file.

See if anything helps or indicates anything.
 

Hello, thank you for the advices but unfortunately, it's still did not work.

Here is all I have try tonight :

- Made the test with 20 back and forth travels instead of 10 => The offset was like 3-4mm at the end instead of about 2 mm.
- Made the test with 40 back and forth travels => The offset was about 6 mm at the end. My measurement is not precise but it indicate that the more it travel the more it offset. I suppose it exclude totally backlash kind of issues.
- Reducing the acceleration and velocity by half => No effect
- Doubling the parport.0.reset-time parameter => No effect
- Marking pulley, motor shaft and lead screw => Nothing slip
- Tried to set my encoder on 800ppr and so adapting my INI according => No effect
- Tried to set my encoder on "falling pulse" instead of default "rising pulse" => No effect

The only thing I still have to try tomorrow is to switch the stepper driver of my X and Y axis as the X axis didn't suffer this problem.. In case, the driver would be faulty. My motors are closed loop Nema 24 with CL57T driver powered with 48VDC.

Any others advices is really welcome because I out of ideas after that.

Thank a lot.

tommylight wrote:

Do you get latency warnings when starting LinuxCNC?
 

Nope. Moreover, I only issue this with my Y axis.

The only thing I still have to try tomorrow is to switch the stepper driver of my X and Y axis as the X axis didn't suffer this problem.. In case, the driver would be faulty. My motors are closed loop Nema 24 with CL57T driver powered with 48VDC.

If they are closed-loop steppers then it is possible that the encoder it slipping on the motor shaft. 
If the problem moves with the motor then it could be that. 
If it moves with the driver it is (probably) the driver.

I also have a spare motor, I will try it tomorow if it's not the driver. Is it common for a brand new motor to have a failed encoder ? 

The CL-series stepper drives are programmed from the factory with 4000 counts before they alarm out due to following error.

Before you do anything else, obtain a programming adapter and download the tuning software.  Change the following error counts in the drive to ~50 (or a similar tight value).
 

Hello all,

Today, I've try to switch X's and Y's CL57T driver, and the issue is still the same on my Y Axis. So, I've put a new motor on my Y axis and unfortunately.. problem is still there also.

I did my back and forth test on my X axis just to prove myself that I am no fool, and X axis has no problem..

So, to completely exclude a bearing issue, I did my test on Y axis with 80 back and forth travels. Result at the end, 18mm of offset !

My next thinking was that the belt is slipping on one of the pulley without me being able to notice it ?
Unfortunately I cannot mark pulley to each other because the screw's pulley a slightly smaller than the motor's pulley (cheap chinese stuff). Thus, my pulleys doesn't turn synchroniously. I have had to put my ratio at 205 to get the correct measurement but that's not a real issue ATM.

I then decided to put back my Y axis ratio at 200 (1 motor turn for 5 mm), marked the motor shaft, and made my initial back and forth travel test.
At the end the mark looked exactly where it was supposed to be but I had my 2mm offset. Now, I think I can defenitively exclude a software problem. If software/encoder/driver was the problem, my mark should have been off of about 1/2 turn.

I am left with the belt slipping theory, but it's already super tense. I will order new pulley, otherwise I really don't know what else to try.

Thank a lot for your support.

@spumco : I need to buy a RS232->USB adapter to be able to check driver settings. But anyway, it should work like other axis, they are all three with stock settings (except for resolution).

Ok, I am an idiot ..

After having disassembled my Y axis, I have found the issue. The pulley on the screw has a 3mm pitch instead of 2mm like all the others ones.. So for sure the belt was slipping all over..

Problem will be solved as soon as I get a new one.

Sorry for having taken your times for such a stupid mistake.