MMT Primary Mirror Actuator Test Stand Upgrade D. Clark February 2010

The Existing Test Stand

Problems with the Test Stand A. Lack of documentation on hardware and software used in the test system. B. Poor wiring practice throughout, particularly in the load-cell amplifier outputs. C. Bad wiring and signal integrity in the connections to the VME chassis. D. The power supplies are improperly mounted and have exposed AC wiring on the connections. E. The VME computer is extremely outdated and has (nearly) irreplaceable hardware. The CPU board for just one example uses an end-of-life Motorola 68000 series processor and cannot be replaced, other than with on-hand spares (that were bought from eBay!). F. Calibration tests take excessive amounts of time to complete due to the need to settle and filter the noisy force monitor signals in the system. G. Calibration software uses a primitive console-based format that is not connected to the MMT MySQL database. H. Actuator calibration results are used only for a volts/pound slope measurement correction, and could be extended to include offsets for a more complete y = mx + b calculation for use in the calibrated force command set in the primary mirror support software. I. The test stand software was written in C using VxWorks RTOS, and support is problematic due to Tom Trebisky’s retirement. J. Spare parts are not available for any of the electronics. Indeed, the Acromag 9330 VME card for analog input is the spare for the primary mirror cell computer.

Upgraded Test Stand Features Use commodity PC with Fedora and GUI for actuator calibration Repackage electronics and power supplies; protect hardware from environmental issues Replace un-spared electronics with new design with low-noise performance Clean up all cabling and harness wiring Use new data acquisition hardware that is affordable, maintainable, and “industrial strength” Add health-checking hardware to ensure system is in working order Implement better data archiving by use of the MMTO MySQL database

Custom Electronics A loopback interface, controlled by the PC, allows the analog outputs to be connected to the analog inputs to ensure the hardware is working. Replacement load cell amplifiers are intended to improve the low-noise performance and packaging aspects of the system.

Motivation for use of EtherCAT Avoid expensive proprietary hardware Avoid need for driver development Minimize “homebrew” electronic design Electrically-isolated signaling High data throughput Use commodity PC with standard Fedora Avoid writing boutique software Self-contained project for evaluating new hardware

Further EtherCat Resources EtherCat Technology Group: www.ethercat.org EtherCat Master (GPL Linux): www.etherlab.org Simple Open Ethercat Master (SOEM) http://soem.berlios.de/ (TU Eindhoven) Beckhoff Automation www.beckhoff.com

EtherCat Master as Etherlab.org

Ethercat Master using SOEM A project of TU Eindhoven GPLv2 license Very simple implementation using RAW sockets in C Not as advanced as Etherlab, some desirable features are missing Unclear how process data are offered to application

EtherCat Master via Etherlab.org’s software Requires configured kernel source Can use generic network driver so no new kernel module is necessary for NIC driver Need to have kernel module ethercat.ko, a vanilla Fedora installation won’t work Need to tell ethercat which NIC to use (customize etc/sysconfig/ethercat) /etc/init.d/ethercat start – syslog will note any messages from the master

Test Stand Upgrade Tasks Strip existing equipment from the stand Clean, prep and paint the stand Add 19” rack rail to accept electronics Repackage power supplies Build enclosure for EtherCat modules and loopback electronics Build new load cell amplifiers Integrate the new electronics and cabling Write new test stand software and test