ABS When the Machine Tool Moves Randomly

ABS is the abbreviation most controls use for absolute. In a positioning context it describes a mode where axis position is referenced to a fixed machine zero rather than to wherever the axis happens to be. When an operator says the machine moves randomly, the first question is whether the control is still in absolute mode or has been switched to incremental, sometimes labeled INC or relative.

The second meaning of ABS sits in the parameter table. On many controls, parameter #0 holds a unit bit that tells the control whether the stored values are metric or inch. Change that bit and every position value in memory is reinterpreted. An axis that was sitting at 100.000 mm suddenly reads as 100.000 in, and the next commanded move sends it roughly 2.54 times further than the program intended. That is what looks like random motion.

Absolute positioning also depends on the reference return. A machine that has not been homed, or that lost its encoder zero after a battery change, will place the part wherever the control thinks zero is. The motion is not random in the mechanical sense. The control is executing a valid move against a wrong origin.

Keep the two apart in your notes: a mode or unit problem changes the command, a reference problem changes the origin. The fix is different for each, and swapping parts before you know which one you have wastes a shift.

The classic failure goes like this. Someone opens the parameter search screen and presses the NO.SCR softkey instead of the intended key. The screen returns INPUT, and parameter #0 is edited. The unit bit flips between metric and inch. From that moment every axis behaves as if it were scaled, and the operator sees the tool travel to positions that do not match the drawing.

On a lathe this shows up fast. A 50 mm shoulder becomes a 50 in shoulder, and the slide runs into the tailstock or the chuck before the operator can hit feed hold. On a mill the effect is milder per move but still dangerous on long traverses. The machine is not losing steps. It is doing exactly what the wrong unit says.

Check the unit bit before anything else. Most controls display the active unit somewhere on the position screen, often as a small mm or inch flag near the axis readout. Compare that flag with the program header. A mismatch between the two is your answer.

Restore the bit, then re-reference the axes. Values that were written while the wrong unit was active may have been stored as converted numbers, so do not assume the original offsets survived. Verify the work offsets with a dial indicator before running a real part.

• 1
  Softkey pathNO.SCR opens a search, not an input. Confirm the cursor is on the parameter you want before writing.
• 2
  Unit flagWatch the mm or inch indicator on the position display, not the program header alone.
• 3
  After the fixRe-home every axis and re-check work offsets with an indicator.
• 4
  Write protectionEnable parameter write protect during normal production so a stray keystroke cannot edit #0.

Not every random move comes from the parameter table. Once you have confirmed the unit bit and the absolute mode are correct, the next question is whether the control issued the move or the drive made it. Read the command position and the actual position on the same screen. A gap that grows during a cut points to the drive, the encoder or the mechanical train. A gap that appears instantly between two blocks points to the program or the offsets.

Encoder feedback faults are the usual suspect on older machines. A contaminated scale, a loose coupling on the feedback shaft or a failing battery on an absolute encoder all produce position errors that the control reports as unexpected motion. The axis may jump at power-up, then behave normally for hours.

Backlash and lost motion look similar but behave differently. They appear when the axis reverses direction and stay consistent in size. Random motion does not repeat. Measure backlash with an indicator on a warm machine, then compare the number with the mechanical specification for that axis.

Thermal growth is the slow version of the same problem. A spindle or ballscrew that heats over a long run shifts the work zero gradually. The operator sees a drift that starts after two or three hours and resets overnight. Log the room temperature and the axis position at the start of each shift to confirm it.

Start with the cheapest check. Confirm the unit flag, the absolute or incremental mode and the active work offset on the position screen. Three numbers, thirty seconds, and they rule out the most common cause of what looks like random motion.

Next, run the axis in MDI with a single move of 10 mm and watch the actual position. Repeat in the opposite direction. A clean machine returns to the same count both times. Write the two numbers down. A difference of more than the axis specification tells you this is mechanical or feedback, not a parameter problem.

Then isolate the drive from the control. Command the same move from the drive test mode. If the axis behaves there, the command path is suspect. If it misbehaves in both, the fault is downstream of the control. This split saves hours because it tells you which cabinet to open.

Finally, look at history. Ask when the problem started, what was changed that day and whether the machine was moved, re-leveled or had a battery replaced. Random motion almost always has a trigger. Find the trigger and the diagnosis usually follows.

Only after these steps should you swap a servo drive or a scale. Parts swapping without a recorded baseline is how a two-hour job becomes a two-day one.

Parameter write protection is the single most effective measure. Enable it for daily production and lift it only for planned maintenance, with the change logged. Most controls also let you back up parameters to a memory card or over the network. Take a backup after every commissioning change and keep the date on the file name.

Operator training matters more than any setting. Teach the difference between a search screen and an input screen, and make it normal practice to press the position display after any softkey that writes data. A quick look at the unit flag catches the mistake before the first rapid move.

Schedule reference return checks and backlash measurements at a fixed interval. Trending the numbers over months shows a failing encoder or a worn ballscrew long before it produces scrap. A machine that measures stable for a year is unlikely to move randomly for no reason.

For shops that outsource machining, ask the supplier how they control parameter changes and how they record them. A documented backup routine is a fair question during a supplier audit.