Sometimes during startup, the motor LED in the backpack becomes red and blinks

[traversaro]

Description of the failure

Sometimes during startup, the motor led in the backpack becomes red and blinks.

Detailed conditions and logs of the failure

It happened once today and once yesterday. Unfortunately I do not have logs of what happened, but apparently when the motor LED in the backpack becomes red and blinks the motor power is not erogated.

Workaround

The issue can be avoided by deactivating both head motor control boards, i.e. all head joints (in the local configuration files icub_all.xml and icub_all_inertials.xml). As a consequence we cannot do torque control on the robot anymore since the IMU frame orientation cannot be propagated through the neck kinematic chain (requires reading the neck joint encoders through the motor control boards).

[traversaro]

To stop the LED motor blinking and actually turn it to green, you have to keep it pressed for 1 or 2 seconds (not sure).

[francesco-romano]

On the iCubGenova02 this happened with spikes in the power supply. cc @maggia80

[traversaro]

It happened again once today.

[nunoguedelha]

It happened twice again this morning, and counting..

[MrAndrea]

Hi @traversaro and @francesco-romano when the motor led blinks red, it means that an overcurrent occurred. (btw: the CPU led has same behaviour). If this seldom happens, just press for few seconds the button and forget it. But if it often happens there is a problem to understand, so in this case call us: it could be a problem with the power supply or, worse, problems on boards or cables.

[nunoguedelha]

@MrAndrea, The issue is occurring very often now, at least 1 over 2 yarprobotinterface run. Could you take a look this afternoon. The robot is free.

[nunoguedelha]

@MrAndrea, the issue does not occur if we don't activate the motor control board for head joints 2 to 5, i.e. the neck yaw and the eyes. So for now, as a work around, I deactivated both head motor control boards, so all head joints (through the local configuration files), since anyway, the neck chain cannot be fully controlled. @traversaro, I will add this workaround to the initial description of the problem.

[nunoguedelha]

Fixed by @MrAndrea by increasing the over-current limit to 6 amps on the robot main power board. The robot interface now starts with all joints activated (after removing the workaround) and without the over-current failure. I suggest we keep using the robot one or two days before closing the issue, for checking this doesn't occur anymore.

[traversaro]

Something similar happened today on iCubGenova02, three weeks after the installation of the new head. @gabrielenava is monitoring the situation and will open a new issue if necessary.

[MrAndrea]

@arrenglover and @francesco-romano, I have updated the firmware on the backpack in order to increase the over-current limit in the same way of the green iCub. Try and let me know if the issue is fixed.

[arrenglover]

During experiments I noticed the purple robot would no longer fault during startup of yarprobotinterface. However when all joints were under control, and I also run gaze tracking software to move the head, the firmware limit in the backpack would again be triggered. To run gaze tracking (e.g. something like the red-ball demo) I needed to idle other joints such that the total current was less.