Abstract
The Dark Energy Spectroscopic Instrument (DESI) is under construction to measure the expansion history of the Universe using the Baryon Acoustic Oscillation technique. The spectra of 35 million galaxies and quasars over 14000 sq deg will be measured during the life of the experiment. A new prime focus corrector for the KPNO Mayall telescope will deliver light to 5000 fiber optic positioners. The fibers in turn feed ten broad-band spectrographs. We will describe the production and manufacturing processes developed for the 5000 fiber positioner robots mounted on the focal plane of the Mayall telescope.
Introduction
The DESI project represents a significant advancement in observational cosmology, requiring a massive "swarm" of 5,000 robotic fiber positioners to map the universe in 3D. The primary engineering challenge lies in the density and precision required: 5,000 robots must be packed into a focal plane with only 10.4 mm between centers. Each robot must move its fiber to a target within 5 micrometers (RMS) in under 45 seconds and survive over 100,000 moves throughout the mission. This paper details the transition from complex R&D prototypes to an industrial-scale production line capable of manufacturing these high-reliability robots at a rate of 180 units per week.
Mechanical Design and Industrialized Assembly
Each positioner utilizes two 4mm DC brushless gear motors to control two rotational axes, designated as theta (θ) and phi (φ). The system operates on an "open-loop" principle; rather than using internal encoders, the robots rely on a "Fiber View Camera" at the telescope to provide optical feedback for iterative positioning.
To manage the assembly of over 200,000 individual parts, production was centralized at the University of Michigan using a custom SQL database and an iPad-based "electronic traveler" system. This allowed technicians to log every assembly step and verify tolerances in real-time. A critical "gating" strategy was implemented, where electronic sign-offs prevented any sub-assembly from advancing to the next stage unless it passed specific quality control (QC) checks. This rigorous tracking improved mechanical assembly yields from under 50% to over 80%.
Advanced Diagnostics and Quality Control
Because the positioners are largely held together by epoxy joints, they are essentially non-repairable once completed. This necessitated stringent pre-assembly QC, such as plasma cleaning all parts to ensure maximum bond strength. To inspect the internal state of completed units, the team employed high-resolution 3D Computerized Tomography (CT) scans.
CT scanning proved invaluable for "forensic" analysis of intermittent motor issues that were invisible to the naked eye. For example, scans revealed how motor cables were interacting with internal components, leading to modifications in cable braiding to prevent interference.
Key Results and Engineering Insights
The production line successfully achieved a peak rate of 50 assemblies per day. Beyond throughput, the project yielded significant mechanical insights. One notable discovery involved the trade-off between rigidity and compliance: the team found that removing an extra bearing from the motor—originally intended to reduce runout—actually increased the robot's lifetime. This change allowed the drive shaft to be more "compliant," better compensating for minor mechanical misalignments that otherwise caused premature failure in over-constrained systems.
Final performance metrics were robust:
- Yields: 80% mechanical assembly yield, with 95% of those units passing final positioning accuracy tests.
- Reliability: Weibull analysis of lifetime test data (up to 1.6 million moves) projects a 98% survival rate after 100,000 moves, far exceeding the 90% science requirement.
- Efficiency: Optimization reduced the labor time per positioner from over two hours to approximately 45 minutes.
Conclusion
The DESI fiber positioner production successfully demonstrated that aerospace-grade tolerances can be maintained in a high-volume, industrialized manufacturing environment. By combining custom digital tracking, advanced non-destructive testing like CT scanning, and iterative mechanical refinement, the team delivered the 5,000 robots necessary to begin the next great survey of the dark universe.
Summary updated