Random X Axis movement - what are possible sources

I am starting to experience random jumps in the movement of my x axis which not only causes a rapid movement of the cutting tool, but the mill doe not register this rapid move so the entire x axis coordinate system is shifted after the rapid move. My system is a 3 axis mill with servos, differential encoders and running 5i25/7i77 boards. The system was initially set up 4 years ago and until recently the system worked great.

This happens on multiple programs, but the last program I am running machines a 4x6 array of rectangles with each rectangle being about 1" x 0.5 inches. The program runs four separate milling operations on each of the 4x6 parts (i.e. tool changes) using four different tools. The program has been run dozens of times over the last several years with no problems. My issue is that while the mill is machining one of the individual rectangular profiles, during any x axis G1 move, the mill can randomly accelerate for about 0.25 inch in the x direction, and then go back to normal. This can happen during any of the machining operations, but only occurs in the x direction. This can also happen when the mill is making a G3 arc move, but the shift still only occurs in the x direction. This rapid move in the x direction (sometimes positive, sometimes negative) results in a permanent shift to the mill's coordinate system in the x direction. Not only does the mill accelerate and then go back to normal movement, but during this acceleration the mill does not seem to realize that it did this rapid movement so the entire x axis coordinate system will shift about 1/8 inch so now that the mill goes back to its normal path, but the entire pattern is now 1/8 of an inch off in the x direction. When the mill does this, no axis following errors appear so the mill / linuxcnc does not know anything is wrong. This only happens on the x axis and not all the time. The program can go through three of the machining operations fine, and then shift on the fourth operation, or like yesterday it shifted during the second operation when the mill was surfacing each of the 4x6 parts using a circular spiraling pattern. I have tried things like slowing down the cutting speeds in case something is not running fast enough, but I still have problems. Like I said, the machine has been running for years without any problems and there is no other software installed or running on the machine. I have run the program several times without any tools, so no material is cut and the program completed without any errors; but when I reran the program with tools, the errors returned. This is probably more just random occurrence of this error because I don't see how having the tools do actual cutting would make a difference since the cuts are fairly light (i.e. aluminum)

As I understand the electronics of the servo motor, servo amp, and 7i77 board setup, the servo moves based upon a voltage signal supplied from the 7i77 to the amp. A momentary voltage spike in this signal being applied to the amp would cause such an accelerated movement, but it would not explain why the system does not appear to realize that a rapid move had occurred. When I say a rapid move, I am not talking about a G0 command being issued. This rapid movement occurs in the middle of the machine executing a simple G1 move. (i.e. the mill cutting from x 2.0 to x 3.0 for example) This has also happened during G3 arcs and it may have also happened during G0 rapid moves, but it is hard to tell since the G) rapid move does not leave a trail (i.e. a cutting path).

I do not understand how the mill can make this rapid move (i.e. shift) and not know it happened. Typically if the mill makes an erroneous move, linuxcnc reports a following error. The error has to be something that is common to both the movement of the x axis (something associated with the servo and possibly the servo amp/driver) as well as the encoder because the machine does not realize it made this rapid shift.

Things I can do are as follows:
1) I can order a replacement servo amplifier board. I probably should have a spare anyway, but would a troublesome board account for both the rapid movement and the fact that the system does not know it just make a rapid move and now the entire x coordinate is shifted?
2) I can switch some hardware to try to isolate the problem. I can try to switch the amp driver boards between the x and y axis to see if the shift moves to the y axis.
3) since I only have 4 axes, I can move the x axis to one of the other empty encoder connectors on the 7i77 board if that makes sense, but could something go wrong on the 7i77 board that would occasionally affect both the movement of the servo and the reading coming back from the encoders?
4) I can reload linuxcnc, but is there really a common point in linuxcnc that would cause such an error?
5) I can replace the encoder and/or servo, but I do not see how a failure of either component would cause this type of random error.
6) I can reload all sorts of software, but is there really any piece of software that would automatically change after years of working and only cause these random shifts in the x axis on an occasional basis?


Does anyone have a suggestion of where I should start looking for the cause of this error?

My guess would be to look at the X axis encoder and X axis encoder wiring

Yes, Peter is right on. One thing you can do is put the X encoder velocity on Halscope and wait for another error to show up. Any sudden jumps in velocity are physically impossible, and therefore due to problems reading the encoder. It could be a defective encoder, a dirty contact on the encoder connectors or a broken wire in the encoder cable.

I had a problem where a dirty contact was reducing the DC supply voltage to the encoder to the point that it was losing counts. It didn't cause big jumps, but a small error such that machined features were not in the right position. All I had to do what reseat the connector.

Jon