E-Mail to Dipl. Ing. Bernd Jüchtern (Maxon Motor Corp.):
"Project description: As part of my Master's thesis "Concept development and testing of an in-vessel articulated arm for remote handling in ASDEX Upgrade - IVAR" at TUM, I am designing a robotic arm to inspect and maintain the plasma vessel of the ASDEX research reactor. I am commissioned and financially supported by the Max Planck Institute for Plasma Physics in Garching. The background to the project is that the experimental operation would have to be interrupted for up to three weeks if something unexpected "went wrong" (e.g. a lens fell into the vessel or a tungsten shingle came loose from the wall). At the moment, an inspection team has to be sent in, which means that the vessel has to be completely evacuated and baked out. The robot is to be inserted through an existing diagnostic port (hard requirement: max. diameter of links: 85mm) and then work in a prevacuum at 10e-3 mbar, which does not entail any heating up.
For this, of course, it is essential that the components both function in vacuum and outgas as little as possible to the ambient conditions (prevacuum, about 25°C). In order to reach all locations of the inner wall of the vessel in a semi-circular movement, about 5m must be bridged horizontally and, in order to reach the ceiling and the bottom of the vessel, about 1m vertically (-> a total of 6m unsupported).
Function of the robot: As a basic concept, I decided on a cable-actuated, cantilever arm, as drives directly at the joints of the robot would represent too much leverage for the structure. This robot has two types of joints, 20 in total: 12 that allow rotation in the horizontal plane (1 DOF, two ropes needed for control) and 8 that can also move in the vertical plane (2 DOF, three ropes needed for control). The 2DOF joints are introduced first, followed by the 1DOF joints. A final degree of freedom is achieved by the translatory insertion of the arm. The last bearing point is at the entry opening. Since the work order for the manipulator is limited to manoeuvring within a tokamak torus, I decided to combine four links each with Bowden cables into common controlled segments (29 DOFs become 8 DOFs) to reduce complexity and drive count. The ropes used are high-strength Dyneema ropes, which are attached to slides in an actuator box. These slides move translationally on a linear guided trapezoidal spindle (currently still being designed, probably TR10x2 or TR12x3 or TR14x3), which in turn will be driven by suitable drives.
Requirements for the drives:
- Operation in a high-purity pre-vacuum: 10e-3 mbar and no outgassing allowed.
- The rope and thus the actuator that is to move the second 2DOF segment in the vertical plane when it is fully extended is subjected to the greatest load. With a payload of two kilos and a dead weight of 1.7kg per segment as well as 0.3kg for the gripper, about 1.5kN are required to hold the link at a force application point by the rope at a distance of 0.038m from the neutral strand. In order to then also achieve an acceleration, 2kN can be assumed. With a trapezoidal thread pitch of 2mm (TR10x2) and an assumed efficiency of 0.7, a drive torque of about 0.446Nm is required at the spindle (at 3mm pitch correspondingly 0.668Nm). The speed is of secondary importance, but should be able to reach about 100/min.
- So a gearbox is needed - I was thinking of a GPX ceramic gearbox with about i=66:1 and an EC22. If possible, I'd also like a cheaper version (the DC-max was also shown in the selection programme, but I'm not sure if it's vacuum-compatible).
- An encoder is not absolutely necessary because the joints are to be equipped with angle sensors according to which they are controlled. A speed controller is therefore sufficient.
- In the end, 13 drives will be needed in the drive unit for the complete robot. In the limited time available for my Master's thesis, however, I will only be able to build a shortened demonstrator. Therefore, I need six complete drives for the time being."