Problem about the stiffness setting about the experiments about TestStand

I assemble a test stand Kit mentioned by the paper .

Now my progress is doing the experiment about Quasi-static,using the Cartesian impedance controller the paper provided:

toruqe = JT(K(xd - x) - D$\dot{x}$)
where xd is foot end position in the local coordinate system of joint HFE.

And then i increase the stiffness K from 10N/m to 100N/m.
According to my test,10N/m,20N/m,50N/m is ok,the leg become more stiff when de K increase.

When i set the K=100N/m,the leg sudden jerk at start,and then the Udriver report a encoder error.

The paper have memtioned that the stiffness can be increased almost at 360N/m without damping,i don’t know how to achieve.

Does anybody have issues?

@Felix Grimminger

Every time the microdriver detects the encoder index pulse the sum of the pulses of the incremental track are checked - if the sum is not correct it means that encoder pulses were missed or too many were counted during one revolution - that triggers the encoder error message.

So if you encounter an encoder error there is a inconsistency of the the encoder signals.
It could be the connector, the wires or a damaged code wheel - we never had a damaged encoder so far.

You could check the encoder wires and the connector.
You could also check the codewheel for scratches and maybe replace it.
Make sure that the codewheel doesn’t touch the encoder when installed - scratches on the codewheel can cause those sorts of issues.

Too much play of the motor shaft assembly can also lead to scratches on the code wheel.
To test this hold the rotor, try to move it up and down axially and make sure that there is no (or only minimal) play and that the code wheel doesn’t touch the encoder surfaces at any time.

Good luck and best wishes.
Felix

Thanks for reply,Felix. What i want to ask is not the encoder error.If motor is considerable oscillation,i think this will cause encoder error.

I am doing the Quasi-static experiment mentioned in the paper An Open Torque-Controlled Modular Robot Architecture for Legged Locomotion Research

The problem i face is that the leg may jerk if i increase the stiffness parameter K more than 150N/m.

I am not sure if need to do extern compensation like gravity compenstion,or just a simple Cartesian impedance controller is totally OK.

I would recommend to investigate and fix the encoder problem.
We have not observed encoder errors caused by oscillations or jerks.
The encoders and the quadrature decoders on the motor driver board can handle violent vibrations and high accelerations without any issues.

I cannot comment on possible issues with your impedance controller.
Maybe somebody else can help?