Very mysterious dropping steps on one Y axis motor...I'm starting to lose it!!

[Hatchett Job]

I'm very aware that I'm a newbie here and I should probably enter the forum a bit more gently but I have a problem I could really do with some help with and it's a biggie.
I'm having an absolute nightmare trying to debug my new machine, I've been at it for seven days straight and I'm losing the will to live.

It's a pretty standard 8x4' setup.
2.2Kw spindle with VFD.
SM 86HT115 Steppers on Y axis with DM860e controllers on 48v.
SM 60STH100 on X axis with DM542 on 36v.
SM 57STH76 on Z axis with DM452 on 36v.
Generic Mach 3 controller board.

My right side Y axis is continuously losing steps no matter what I do, up to 2-4mm over the course of running for 20mins or so at between 45 and 70mm/sec.
I'm running two separate inputs for the Y axis motors.
So far I have checked all motor tuning is correct.
Checked the test tool paths I'm using are correct.
Replaced the controller board for another one.
Swapped over the two motor controllers, fault stays on the right motor.
Physically swapped over the two motors, fault stays on right.
Checked the rack pitch on both sides was identical.
Checked shielding on everything is properly earthed to the chassis.
Twisted the motor controller power supply wires.
Tried multiple PCs, fault stays on the right.
Ran the machine with the spindle off, fault stays on the right.
Ran the machine with the X axis powered down, fault stays on the right.

The only thing I've just noticed is one of the cables on the two Y axis motors is a different make to the other so I'll sort that out on Monday.
I've also been told that resonance in the body of the machine could be the problem but I've no idea what to do about that.
Feels like I'm clutching at straws now.

[m_c]

Do the motors run in the same, or opposite directions?

I'm guessing that if you're using a racks, then the motors will physically run in opposite directions.
In which case, how has that been achieved?
By the wiring having been swapped for one of the motors coils, or by changing the polarity of the direction wiring to the driver?

If it's been done by swapping the direction wiring polarity, then the step/dir timing will likely be out of sync, which will result in the driver moving two steps out of position when the direction is changed (but only when the direction is changed in one way, the opposite direction change will be correct).


The simplest test for loss of position, is to setup a DTI against the axis in question. Then run a simple g-code program that simply moves the axis back and forth a short distance 50 or a 100 times, and see if the axis returns to it's original starting point against the DTI.

[dazp1976]

Why use 86mm motors?.
48v is low for that size. Could be dropping out.
You have DM860 drivers so use them.
72vdc, no less.

[JAZZCNC]

M-C has probably given you the correct answer to your issue in that it's a step timing issue because even though you swapped motors and drives you will have kept the same Step settings in Mach3 and the drives will both use the same pulse edge setup meaning the issue will remain on that side.

Without getting too technical what's happening is that the STEP pulse is given in front of the DIR pulse so any steps given before the direction change are lost.
The cure is easy just change the Active state of the STEP pin in Mach3 just for that side,

The test is like M_C describes with small moves back n forth (bounce test we call it, see vid), 10mm is more than enough but I would do more than 100 as it's easier to see then. Notice on the video the indicator doesn't get touched until it's finished and goes back to zero slowly for the last move so the indicator doesn't get knocked.

This test is worth doing on all your axis because it doesn't always show, esp on a Z-axis that doesn't change direction very often ie Profile cutting and pocketing where it just goes to depth and stays there until finished. Often problems like this only come to light when someone tries to do 3D or engraving work with lots of Z-moves.

Dazp1976 point is valid also because 48V is low for a Nema 34 motor so you could be dropping steps depending on how you have the machine setup, esp on a rack n pinion setup which is very inefficient compared to ball-screw machines and have a lot higher friction to deal with. At higher feeds and with a heavy gantry on an 8x4 machine, you could easily be losing steps.
Try running slowly and see if you are still losing steps, this will tell you straight away if you are dropping any due to lack of power/friction.




https://www.youtube.com/watch?v=5JpPZGSYpG0

[Hatchett Job]

Why use 86mm motors?.
48v is low for that size. Could be dropping out.
You have DM860 drivers so use them.
72vdc, no less.

I don't know much about this.
I got all the components as a bundle from a local CNC building company.
The 48v power supply is what came in it, should I change it to 72v??

[Hatchett Job]

This is brilliant! Thank you guys
I have been checking the discrepancy by measuring the distance from the end of the track to the carriage when the toolpath has finished.
The discrepancy increases the more complicated the toolpath is so that would back up your advice.
Also, yesterday I ran the machine with the right hand motor running as a slave to the left and there was no run out.
I've not noticed any difference when changing the speeds.
I will change the Active state and report back.

[dazp1976]

I don't know much about this.
I got all the components as a bundle from a local CNC building company.
The 48v power supply is what came in it, should I change it to 72v??

Yep.
I'd move the 48v to the smaller DM452/542, ditch the 36v (keep it for bench testing other things maybe), and install a new 72v onto the DM860.
I usually buy these types of items on Aliexpress, pretty easily available.
This should improve any velocity anxiety when it comes to losing steps.

Dual motor axis are best ran with one as a slave.
Coils on both the motors should be wired identically to avoid inductance/resistance differences. Esp when it comes to 8 wire units.

[Hatchett Job]

So I tried running my test tool path with step low active selected on A axis (my right hand Y axis motor) and then on the Y axis.
Both times A axis gained steps, am I missing something?



https://i.imgur.com/zgqEEJD.jpg

[JAZZCNC]

This is brilliant! Thank you guys
I have been checking the discrepancy by measuring the distance from the end of the track to the carriage when the toolpath has finished.
The discrepancy increases the more complicated the toolpath is so that would back up your advice.

You need to do this in a methodical order. First, you need to identify if you are losing steps because of pulse timing problems or tuning/power/mechanical problems.
Forget your toolpath and just do the bounce test first by making a simple G-code program to move the axis back n forth, this will tell you straight away if it's a pulse timing issue and if it's not then you can move on and look else were.




Also, yesterday I ran the machine with the right hand motor running as a slave to the left and there was no run out.

Ok, you are giving confusing messages now.? So are you saying you were not using slaving before.? If yes then how were you doing it.?

[Hatchett Job]

Thanks so much for helping me with this.
Here's where I'm at.
Like you said I made a test toolpath that jogs the Y axis up and down 100 times.
First I ran it with motor A polarity reversed and sharing motor Ys signal from the control board. it worked fine.
Next I ran it with motor A slaved off motor Y with motor A step low activated and it lost steps.
Then I ran as above with motor Y step low activated and it lost slightly more steps.

[JAZZCNC]

Thanks so much for helping me with this.
Here's where I'm at.
Like you said I made a test toolpath that jogs the Y axis up and down 100 times.
First I ran it with motor A polarity reversed and sharing motor Ys signal from the control board. it worked fine.
Next I ran it with motor A slaved off motor Y with motor A step low activated and it lost steps.
Then I ran as above with motor Y step low activated and it lost slightly more steps.

What controller are you using and what frequency is it running at.? If the frequency is too fast then the drive opto's might not be able to keep up so steps are getting lost.

[Hatchett Job]

This is the board I have.
Is the frequency the Kernal speed in ports and pins configuration?...if so it's 25,000hz
https://i.imgur.com/jwkxLQR.jpg
https://i.imgur.com/s6u85zB.jpg

[JAZZCNC]

This is the board I have.
Is the frequency the Kernal speed in ports and pins configuration?...if so it's 25,000hz

Ok so you are using the parallel port now I'm not saying that is your problem, thou it's highly possible, but what I am suggesting is you do yourself a favor and buy an external motion controller like the AXBB-E which uses an Ethernet connection. Combine this with more volts for the steppers and you'll transform the machine into a different animal.

It will be more reliable and faster, you'll also be able to dump Mach3 and go with UCCNC software which is much better.

Now back to your issue.!

Because you are using the parallel port some of this could be because you have the machine overturned, So give us the settings you have for the motor tuning and the micro-steps set on the drives for each axis.

Also, have you checked the STEPS PER is correct by measuring it moves the correct distance told.?

[Hatchett Job]

Yeah that Ethernet board does look like a better job alright.
Here's pics of my my motor tuning and the motor controller jumpers.
Yup I have the steps per distance set correctly.
That in itself has been an issue, when I use the feature in Mach 3 settings to do this the actual cut dimension is different so I have to make a cut and measure that.https://i.imgur.com/w4Tv4PX.jpg
https://i.imgur.com/qLgTRjx.jpg
https://i.imgur.com/wL6FBnZ.jpg

https://i.imgur.com/45FFX59.jpg

https://i.imgur.com/vF9r79D.jpg

[Cube3]

Many moons ago I used that board and had missing step issues. I put it down to the speed of the opto isolators, but couldn't say for certain.I then used a board from CNC4YOU which was a lot better and worked ok in Mach3 and had 10v PWM for the VFD.

I finally got fed up with Mach3 and bought the AXBB and wouldn't go back. I can now run my toolpaths without babysitting the machine and hovering over the emergency stop button. Very confident with UCCNC.

Sent from my SM-A715F using Tapatalk

[dazp1976]

Many moons ago I used that board and had missing step issues. I put it down to the speed of the opto isolators, but couldn't say for certain.I then used a board from CNC4YOU which was a lot better and worked ok in Mach3 and had 10v PWM for the VFD.
I finally got fed up with Mach3 and bought the AXBB and wouldn't go back. I can now run my toolpaths without babysitting the machine and hovering over the emergency stop button. Very confident with UCCNC.
Sent from my SM-A715F using Tapatalk

Same.
Although went the UC300eth and UCBB route.

I did however have zero issues with the 5 axis parallel board and mach3 at 35,000 khz.
Will be re-using it on uc300eth port#3 for switching some relays.

[Doddy]

Just a point of sanity - that board has no opto-isolated outputs to drive the stepper driver (which will have) - the buffer devices - 74HC series has a rise/fall time measured in 10's of nano-seconds and current source/sink in the order of 35mA - that's NOT the problem here.

[dazp1976]

Just a point of sanity - that board has no opto-isolated outputs to drive the stepper driver (which will have) - the buffer devices - 74HC series has a rise/fall time measured in 10's of nano-seconds and current source/sink in the order of 35mA - that's NOT the problem here.

That's a good point.
Only optos on the inputs.
I'm pointing to the 48v psu.
Might turn out he's trying to run 2* 86's on a single 350w.
Not happening.

Prev user prob had same problem.

[Hatchett Job]

You guys have started to lose me with the Opto isolators chat, can you put it in layman's terms?
Also the relevance of the frequency?
I'm good at making stuff but electrical engineering and computers are not my strong point.

[Doddy]

Apologies. This particular thread of conversation started with questions around signal integrity (pulse widths, and the like). That lead into a general concern if the breakout board (bob) might suffer in its design with the use of slow opto-isolators - devices that act like electronic switches, that would couple the stepper pulse from the computer to the stepper drivers. It’s a weakness in design that crops up from time to time and exhibits the problems you have witnessed. My post was simply to identify that these devices are not present on your bob, for the stepper driver signals, and that we shouldn’t get distracted with talks of changing bobs on the basis of an invalid assumption.

[dazp1976]

You guys have started to lose me with the Opto isolators chat, can you put it in layman's terms?
Also the relevance of the frequency?
I'm good at making stuff but electrical engineering and computers are not my strong point.

The 2 big black spider looking things are buffers.
The 5 little black things on the left with 4 legs are optocouplers.

The buffers are to smooth things out.
The optos can sometimes be slow and cause pulse problems. However, These boards don't use opto on the outputs so this is not your problem.

In a nutshell.
Optos allow 2 voltages to work together, it acts as a switch via a light diode.
So you can use say 24v limit switches into the opto and then that can signal out to your 5v parallel port without it blowing up.

Using the higher 24v on switches and the like helps to reduce false triggers as it is less susceptible to electronic interference from vfd's etc than a fully 5v system.

[m_c]

Something I'm still not clear on, is I'm assuming you're relying on the settings in Mach to reverse the direction of one motor?

I'd suggest you set both axes identically in Mach3, then reverse the wiring to one coil on one motor, which will cause the motors to rotate in opposite directions.
This way both drives should receive identical step/dir signals, so any direction polarity issues will affect both drives equally.

[JAZZCNC]

I'm with M_C on this, It's getting confusing because your not making it clear how you are running the motors.

Regards the pulse timing I didn't suggest the BOB was to blame but rather the fact it's using the parallel port, which is very often the cause of such issues. Often PC and BOB working together to screw things up. My suggestion to go with an external controller will virtually eliminate all the issues the PP suffers from and increases performance and reliability massively.

Without knowing more about the PSU I can't say if a lack of power is the problem or not like DazP keeps chirping on about, but it's one possibility I suppose thou the fact you say it does it at low feed rates and the fact you don't have the motors tuned overly high makes me suspect this is not the cause of your problem.

I'm still putting my 10 pence on the pulse timing, but you are not clear on how you are checking this or the results you are getting so the jury's still out here until your more clear.?

[Hatchett Job]

Apologies. This particular thread of conversation started with questions around signal integrity (pulse widths, and the like). That lead into a general concern if the breakout board (bob) might suffer in its design with the use of slow opto-isolators - devices that act like electronic switches, that would couple the stepper pulse from the computer to the stepper drivers. It’s a weakness in design that crops up from time to time and exhibits the problems you have witnessed. My post was simply to identify that these devices are not present on your bob, for the stepper driver signals, and that we shouldn’t get distracted with talks of changing bobs on the basis of an invalid assumption.

Not at all...thanks for the clarification.

[Hatchett Job]

Something I'm still not clear on, is I'm assuming you're relying on the settings in Mach to reverse the direction of one motor?

I'd suggest you set both axes identically in Mach3, then reverse the wiring to one coil on one motor, which will cause the motors to rotate in opposite directions.
This way both drives should receive identical step/dir signals, so any direction polarity issues will affect both drives equally.


This is exactly how I am currently running the machine but with both Y motors running from one output on the BOB.
If I give the motors separate outputs then Motor 'A' drops steps.

[Hatchett Job]

I'm with M_C on this, It's getting confusing because your not making it clear how you are running the motors.


Without knowing more about the PSU I can't say if a lack of power is the problem or not like DazP keeps chirping on about, but it's one possibility I suppose thou the fact you say it does it at low feed rates and the fact you don't have the motors tuned overly high makes me suspect this is not the cause of your problem.

I'm still putting my 10 pence on the pulse timing, but you are not clear on how you are checking this or the results you are getting so the jury's still out here until your more clear.?

I made a tool path that jogs up and down the Y axis 100 mm 100 times.
I measure with digital callipers the distance the bearing carriage ends up from the end of the rail at the end of the tool path on both sides.
When giving each motor separate signals form the BOB Motor A (right side) consistently looses steps finishing 1mm off home.
Changing the step low active doesn't make any difference.

When I give both motors the same signal form the BOB it works fine.
I had to start using the machine for a project so I'm using it like this for now, not ideal but not terrible either.