Page 1 of 2

Help: Dynamic backlash! [PICS]

PostPosted: Wed Aug 15, 2012 2:59 am
by artofmystate
My laser has dynamic backlash and I tried everything I know to fix the issue without any luck; I am hoping I can explain the problem and have a hand at solving the issue.

The version of the buildlog laser 1.0. The controller is 608 AWC DSP. The stepper drivers are A4988 Pololus. Microstepping at 1/16. The test program is a 101mm x 1mm box which cycles five times. Testing the Y axis. (there is no space/jerk when running the program) To measure backlash I am using a dial indicator with the resolution in 1000ths of an inch. At the beginning of each cut cycle I set the indicator to zero.

Running a cut cycle 10 times at the cut speed of 10mm/s gives the following measurements when the cycle is complete (in order of testing, units in 1000ths of an inch): -7.5, -9, +10, +15, +3, -2, -15, +2, +2, -9.

Again, but this time the speed is set to 400mm/s: -3, +1, -8, -15, -8, +9, +7, +4, -7, -16.

The machine
laser_entire.jpg


The electronics bay
laser_electronics.jpg


The testing setup
laser_setup.jpg


DSP parameters
ss_2.png


DSP parameters
ss_1.png


Details:

Motors are NEMA 17, 0.9°, Dual Shaft - which are found in the makerslide store: http://store.makerslide.com/index.php?m ... ucts_id=30

The version of LaserCAD is 5.95 and the DPS firmware has been flashed to the latest version 120312V7.UG5 3-12-2012.

Video of the testing procedure: http://www.youtube.com/watch?v=wC0SbDPFAUw


Known issues:

The connecting rod of the Y motor has a whip. See video: http://www.youtube.com/watch?v=4mM7YiLfvyI and http://www.youtube.com/watch?v=uKdfdHCsXOk

Although the belts have a little bit of spring to them (.020") the machine settles back to zero when adding then removing pressure on the axis. The Y axis beam slightly rocks, when forced, about .010".


Step I have taken:

Updated firmware and software.

Increased the amperage to the motors from .67amps to 1.6amps with no difference to degree of backlash.

Added ground wire from stepper board to DSP.

Tried different stepper Pololu driver in different board slot.

Used 1/2, 1/4, 1/8 stepping modes. The backlash increased the most using half steps.

Cleaned the rails.

Tightened the belts to snuggness, not overly tight nor springy.

Tried various acceleration and start speed parameters, high acceleration increases backlash variance only slightly.

Removed all equipment attached to the mains line, the only equipment powered on the breaker is the laser and PC.

Checked the tightness of grub screws, rails, frame, pulleys, etc.

_________

From here I am at a loss. The backlash is not really bad when cutting out small parts, but when trying to cut full 16" x 20" sheets with the parts spaced at .050" apart the loss in accuracy is so great the machine cuts into other parts.

I have MACH3 running my CNC, I have yet to connect MACH3 to this machine.

I've read posts where other experience some similar issues had the steppers to spongy, please review my motor settings.

Also, I lightly tested the X axis with the same, yet smaller range in error, issue.

Anyone help is appreciated, thanks.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Wed Aug 15, 2012 6:41 am
by macona
Dynamic backlash? I have never heard of that.

You can adjust software till you are blue in the face and you will never make a difference. Backlash is purely mechanical.

You need to fix the whip problem. Put a bearing support in the middle of the shaft. Two evenly spaced supports would be better. Or use a thicker shaft.

You test where it goes back and forth is pointless. Bring the carriage towards you against the dial indicator and zero the indicator using most of the stroke. Command the carriage to go 1/2" away. How far it was supposed to move minus how far it actually moved is your backlash.

I like belts tight. Tight enough that they go twang when I pluck them. You are not going to break a belt from over tightness, at least not within limits.

Or are you talking about loosing steps?

Re: Help: Dynamic backlash! [PICS]

PostPosted: Wed Aug 15, 2012 5:30 pm
by mpeele
We had a simular problem. I would not call it backlash but positioning accuracy. You are gaining and loosing steps durring the motion.

When we were using the 1/4-20 rod coupler with set screws there was a lot of wobble. We removed it with one of these in the photo from DumpsterCNC.
2012-08-15 11.01.23.jpg

It's 1/4-20 one side and 5 mm on the other. I don't think it is causing the problem because when I added it positioning accuracy was not affected but the rod wobble stopped.

Next I would look at the acceleration configuration. Without and acceleration we do a calibration with full x and y travel at about 144 ipm( or 60.96 mm/s) using the optos trips to detect axis limits. We get at most +/- 1 step over the full range of Y axis motion and X axis is consist.

2012-08-15 11.01.23.jpg


Increasing the step rate to 91.44 mm/s I see +/- 1 step error in both the X and Y axis. Here acceleration would help both axis.

Now if I slow down the stepping rate we see even more error on both axis.

We use our own laser controller and are currently adding acceleration.

We use 18 tooth pulleys and Gecko 251 steppers driving the same motors with 24 volts. I'm not sure what the current is.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Tue Aug 21, 2012 12:36 am
by artofmystate
I will test backlash using the prescribed technique and post my results when I can. (I am college right now so my tinker time is limited)

I agree, it's likely to be 'positioning accuracy'.

The threaded rod will be replaced as soon as I can along with better couplers.

Thanks for the replies.

*edit*, I am also going to try MACH3 now that I found a cable I can rip apart. =)

Re: Help: Dynamic backlash! [PICS]

PostPosted: Tue Aug 21, 2012 6:32 am
by macona
First off microstepping increases resolution but not accuracy. There is no guarantee where the motor is between full steps. Microstepping is primarily for smoothing things out. So use the lowest setting you can.

How many volts are you running the motor on? Top speed is limited by voltage. If you dont have enough voltage you will loose steps. Try slowing things down. Also try increasing you acceleration and deceleration times.

Disconnect the motors from what they drive and put a marker on the shaft and program it to run back and forth for a while. Come back later and see if it is still returning to it's home position. If it is not then you may have a noise issue. You could also have an issue where the pulse length from the controller is too short and it is not catching every pulse. I dont know if there are settings for that.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Wed Aug 22, 2012 7:18 pm
by edmond22
I had a whipping problem on the Y axis drive rod too, solved it by replacing it with 8mm threaded bar in place of the 6mm one I was using.
The 1/4" or 6mm threaded rod is just too flexible over the almost 1m length.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Wed Aug 22, 2012 8:07 pm
by Enraged
if it's unsupported, it may be whipping. Nook has a good set of calculators for finding max speeds: http://www.nookindustries.com/ball/Ball ... ticalSpeed

It's for ballscrews, not threaded rod, but it might help get you in the ballpark.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Sat Aug 25, 2012 4:12 am
by artofmystate
I tested the system with MACH3... there is no backlash or position accuracy, at all!

I used the same speed/acc as set in the DSP, then I ramped everything up the most I can (stopping at the point that may cause damage) still no backlash/PA. 837inch/m speed and 200inch/m acc, the machine was jerking real bad, but the most positional accuracy I recorded was .0005"!

So, I am not sure what's the deal with the DSP. My guess is EMI; I don't know.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Sat Aug 25, 2012 9:00 am
by macona
Kind of sounds like you are not getting a strong enough signal to the stepper drives from the DSP. Are you using a current limiting resistor before the drive? It might be a good idea to put a scope on there to see whats going on.

Re: Help: Dynamic backlash! [PICS]

PostPosted: Sun Aug 26, 2012 12:25 am
by artofmystate
Although I don't currently have a scope I measured the voltage to be .30vDC for the low signals and 4.96 for high signals. From there I am not what to test. Perhaps a new PSU for the DSP and stepper driver board?