Coupling Stiffness Servo Tuning: Why an Axis Refuses to Tune
Every so often an axis refuses to tune and nobody can say why. The encoder is good, the motor is more than big enough, the mechanical assembly was built to drawing, and yet the gain cannot be raised past a certain point without the whole thing starting to sing.
Nine times out of ten we find the same thing in that situation. Someone has fitted a forgiving coupling to an axis that needed a stiff one, and the drive is now talking to a spring. That is the heart of every coupling stiffness servo tuning problem we get asked about.
Why coupling stiffness servo tuning questions go together
A servo loop closes around everything between the motor rotor and the thing being positioned. The drive commands current, the motor produces torque, and the torque travels through the coupling, the screw and the nut before anything moves at the far end.
If the coupling twists under load, the motor turns and the load does not follow immediately. The encoder - sitting on the motor - reports movement that the tool has not made yet. The drive sees the error and responds, but by the time the load catches up there is an overshoot to correct.
That extra spring in the loop shows up as a lower frequency at which the axis becomes unstable. Raise the gain and the axis oscillates sooner than it would with a stiffer joint. Lower the gain to keep it quiet, and you lose the responsiveness the machine was bought for.
How it shows up on a real machine
The symptoms are consistent once you know what to listen for.
A following-error trace that has a visible lag on every acceleration - not a spike at the reversal, but a slow curve behind the command. That is wind-up, and a metal bellows or diaphragm coupling will halve it on the same axis.
Whining or ringing that appears as gain is raised, at a frequency that follows the axis rather than the spindle. Reduce the gain and it goes; increase it a little and it comes straight back.
A machine that holds size beautifully in one direction and drifts in the other. The coupling takes up its deflection one way, then has to be pushed back through zero before the load moves the other way.
Settling time that never quite meets the specification on a short move, though it is fine on a long one. Short moves spend a larger share of their time in the region where the spring matters.
| Symptom | Points to |
|---|---|
| Following error on every accel, both directions | Torsional softness somewhere in the train - coupling first suspect |
| Oscillation at a fixed frequency, either direction | Loop gain too high for the stiffness present; reduce gain or stiffen the joint |
| Size error that changes with direction of travel | Wind-up plus backlash, often a keyway or a worn spider |
| Rattle at reversal only | Lost motion in the hub fit or the key, not stiffness |
Testing it without instruments you do not have
You can find out a great deal by hand before you touch the tuning screen.
Clamp the axis, or hold it against a stop, and apply a small amount of torque to the motor shaft. On a stiff coupling you feel it go solid almost immediately. On a soft one there is a perceptible give, and it comes back the same amount each time. That give is the deflection the drive is fighting.
Compare the two ends of the same machine. If the X axis feels solid and the Y axis does not, and the couplings are different types, you have your answer without measuring anything.
One caution - do this with the drive disabled and be gentle. You are looking for a small amount of movement, not testing whether the coupling will hold your weight.
The trade nobody escapes:
There is no coupling with infinite stiffness that also absorbs misalignment. Stiffness and forgiveness sit at opposite ends of the same scale.
What you can do is make sure you are not paying for forgiveness you do not need. If the shafts are aligned well, you do not need a soft coupling, and fitting one costs you tuning margin on every move the axis makes.
What we would change, and in what order
Before swapping the coupling, check the cheap things. A clamp screw that is not tight enough lets the hub creep and that feels exactly like softness when you test it by hand. A spider that has hardened with age is stiffer but also less able to follow misalignment, and it can start transmitting vibration instead of absorbing it. A loose key rattles and adds lost motion that no amount of gain will remove.
Then look at alignment. If the coupling is soft because it is being bent to follow misaligned shafts, fixing the alignment lets you fit something stiffer - and that is the change that shows up on the tuning screen.
Only after those do we look at the coupling type. If the axis is a positioning axis with a tight tolerance, and the alignment is genuinely good, a bellows or diaphragm coupling is the right part, and whoever tunes the machine will feel the difference within a few minutes.
On the other hand, a machine that only ever rough-cuts and where the joints are hard to get to, keep the forgiving coupling. Tuning margin is worth less than a coupling nobody has to visit.
We machine bellows, diaphragm, jaw and rigid couplings in aluminium and steel, bored and keyed to the shaft sizes on your drawing. The range is on the shaft coupling range, and for positioning work the diaphragm coupling is the usual answer.
Tell us what the axis is doing, the shaft diameters and the alignment you can hold, through this form, and we will suggest the stiffest type your alignment will tolerate.







