[GAK2010]

The title is an assumption so maybe this should be posted in another area. Basic problem is that my home built system has started exhibit lost steps on the x axis and the slaved A axis but to differing amounts. I've trying to resolve this for 5 days now, so I am hoping that somebody here can make suggestions of things I haven't looked at. Here's the background.

My system has Nema 23 steppers (380 oz in) on all axes. This is a gantry router table with X an A driving the gantry. Mach 3 (most recent version) is the software and is talking to a Gecko G540. This system has been working for years fairly well (although I did have to replace the computer about 6 months ago) until last Friday. While machining a part the x stepper started to lose steps (detected audibly, visually and by vibration). Fortunately I could stop it before it damaged the part. This happened while it was cutting about 22 inches in a straight x- direction. The gantry was racked with the end closest to the x drive off in the x+ direction by more than 0.1 inch. This was rough cutting at 80 ipm which has not been a challenge in the past. I've cut the same material at double that without problems, but just don't like running that fast. The other side (A drive side) was off by a couple of mils (possibly within the precision of how I measured this).

So I realigned it and set the z zero high enough that it wouldn't cut the part. I restarted the program but reduced the speed to 40 ipm. The x drive sounded worse and indeed was offset again in a positive direction by about 0.030 inches. A side was again within 2 mils. The system moved perhaps all of 10 inches before this error was induced.

I rechecked the alignment of the racks/ pinons/ motors etc. The only thing that seemed a bit out was that the gear meshing into the x rack was about 0.3 degrees out from the rack teeth. I realigned that to within 0.1 degree (electronic gauge showed 0.0 misalignment but who really knows how accurate it is.)

From this point I did a series of 7 cycles of movements in x back and forth for about 22 inches each and then would look at the offsets from the zeros (referenced to the edge of the part I was cutting). I did this at speeds from 100 ipm down to 40 ipm. The offsets at the x side ranged from about 30 mils to about 0.060 (always in the positive direction.) The other side had an offset up to 15 mils (in the minus direction!) but the others were around 0 +/- 3 mils. There wasn't any apparent trend. Between each run I realigned the gantry and established the location of zero (ie edge of my part near to the x and a axis, at y values of 2 and 27 inches respectively). I should mention that the offsets are determined from the difference in the Mach DRO' readings for the original zero and the final DRO location to engage the gauge block at the end of the part. The gantry was always moved toward the gauge block from positive x to negative x. Hopefully that should remove any backlash contribution. While I can't be certain that Mach's DROs are absolutely correct, there were visible gaps between the tool and gauge so I am certain something is amiss.

At this point I presumed the problem was correlated to the x axis and played with a variety of tensions to hold the drive gear into the rack. This really didn't seem to have any effect until it got so loose the gear hopped out of the rack.

I also changed out the x drive stepper motor and then it's cable. Neither had a positive influence. Now I'm thinking that maybe it is the Gecko's x driver. I swapped out the x and a axis cables at the the gecko system. I left the direction unchanged in Mach so the system moved in the opposite direction physically. I expected that the problem I saw on the x would have moved to the a axis. Life couldn't be that simple of course. In the following data plus offset means that the tool did not return all the way to the guage (zero for that particular y value) while a negative offset means it went beyond it. (remember that the system is now physically moving in the opposite direction that Mach is reporting.

7 cycles at 40 ipm (y= 2.00) x offset = 0.0221 (y= 27.000) x offset= -0.0185
7 cycles at 80 ipm ( y= 2.00) x offset = -0.0536 (y= 27.000) x offset= -0.0304

I had previously confirmed that both X and A had identical motor tuning values, but decided to drop the acceleration value in Mach from a whopping 2 to a very slow 1 to see if that helped. This is painfully slow. A three toed tree sloth could dodge a change in direction of this machine without breaking a sweat. The results are as follows (x and a cables still swapped).

7 cycles at 80 ipm (y=2.000) x offset= -0.0226 (y=27.000, I forgot to write down the zero dro value so consider this suspect) +0.04357

7 cycle repeat at 80 ipm (y= 2.000) x offset= -0.0742 (y= 27.000) x offset = +0.0037

7 cycle repeat at 40 ipm (y= 2.000 ) x offset= -0.0037 ( y= 27.000) x offset= +0.0050

One might think the problem is approaching a solution, but the machine sounded and felt like it was missing steps. I was completely done spending time on this machine for today and while repeat this tomorrow at even slower speeds and then reswap the cables. But one more thing to mention.
While doing the last cycles at 40 ipm I noticed that the side of the gantry closest to the A drive was moving before the x axis. The A stepper is now driven off the X port out of the Gecko and that cable is 12 ft long. The cable going to the x stepper motor is 6 ft long. At first I wasn't convinced I was really seeing this as I was at an oblique angle and toward the x axis side of the the gantry rail (about 46 inches long). It is difficult to watch both sides in that position and the A axis motor is blocked by the gantry itself. When the cycle test was finished I was able to reposition the keyboard/ screen to sight down the gantry rail and could see clearly that the a side was moving just slightly before the x side. I repeated this multiple times just to be certain it was indeed happening.

Has anybody else ever seen a noticeable lag in pulses going to a slaved axis out of the G540? Clearly I'm not talking nanoseconds with these old eyes.
Speaking of pulses, the Mach3 pulse duration was 5 microsecond during all of these tests.

I don't believe in ghosts, but I am coming around to thinking there is a ghost now inhabiting this system. Any suggestions short of exorcism would be appreciated. I really don't want to plod along at 40 ipm or less for the rest of my life.

[bhurts]

I didn't see you mention checking the pulley and gears setscrews. I would check to see if you have something mechanically slipping before looking at the electronics. That means belts, pulleys, gears, etc...

Ben

[GAK2010]

Hi Ben,

I did check the side associated with the x drive multiple times. Before I replaced the motor yesterday they were for sure tight. On the other side I checked for slop by hand but didn't put the allen wrench to the set screw on that motor. I will do so today.

Thanks,

Greg

[GAK2010]

A quick update. I swapped the stepper cables back so that X is being driven from the X driver of the Gecko and A from the A driver. The A side continues to lead the x axis in movement. Now I am thinking there is something in the mechanical system. It wasn't a loose set screw or bolt as I've checked them all with the exception of a couple on the Z axis which are not accessible without a partial disassembly of that mechanism.

I'm not sure that this is the root cause of my problem but it is certainly a condition that shouldn't exist. I'm going to disassemble the both sides of the gantry and start from scratch to put it all back together.

I must have been a bad boy this year so I will be expecting nothing but coal in my stocking.

I'd still appreciate any other ideas somebody might have.

Greg

[dsnellen]

I had a similar problem and the answer was a slippage of the screw to motor coupler. The coupler had pulled off the motor shaft slightly and the set screws weren't gripping the motor shaft tightly. Repositioning the coupler and tightening the screw solved my problem.

Dave

[GAK2010]

Thanks for the suggestion Dave.

I finally got to the root of the problem but it did require a major tear down. At the end it was the linear bearing seizing on the x axis. It was almost accidental that I found it so I'm hoping it may help somebody else by mentioning. It was only by completely disconnecting the bearing for the structural support for that side of the gantry and then trying to slide the bearing out of the way that I realized it was way too stiff. I had to overnight the spares. Late today I finally got everything put back together and calibrated.

Noticing the slight hesitation on the X axis was the key. It's not really easy to see both ends of the gantry when they are separated by over 50 inches , so sighting down the gantry at every start up will be a new part of my process.

[dsnellen]

Greg,
Glad you got it fixed. Running a CNC machine does require excellent troubleshooting skills with logical thinking being at the top. It's what I call the big three: What is it doing? Why is it doing that? And how can it be fixed? Sometimes the answers are not always oblivious. And it can be extremely frustrating at times.

Dave

[Oleksii_Kyslyi]

Hi, GAK2010,

We want to add our 5 cents. For the future we recommend to use servo motors, because they have feedback feature. In your case with stepper motor a definition of a problem is like to work with black box. In a case of servo motor, because of feedback presence, it is simplier to define a cause of a problem and fix it. Yes, servo motors a more expensive, but it is simple to calculate, how much you should to pay more for servo in comparison with time in the future, which you will need to spend for fixinf of appeared problems.

---------------------------------------------------------------------------------------------
Oleksii Kyslyi Solutions for CNC machines Website: http://ww-cnc.com