🔭 astrophysics

CHARA Array Delay Lines: Upgrades, Performance and Future Directions

This paper details the comprehensive modernization of the CHARA Array's optical delay lines from a legacy VME architecture to a hybrid FPGA and Linux-based system, which maintains high-precision tracking performance while enabling future capabilities such as extended baselines and advanced observing modes.

Original authors: Narsireddy Anugu, Nils H. Turner, Theo A. ten Brummelaar, Gail H. Schaefer, Philippe Bério, Christopher D. Farrington, Becky Flores, Douglas R. Gies, Stefan Kraus, Edgar R. Ligon, Olli Majoinen, John

Published 2026-02-19

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Narsireddy Anugu, Nils H. Turner, Theo A. ten Brummelaar, Gail H. Schaefer, Philippe Bério, Christopher D. Farrington, Becky Flores, Douglas R. Gies, Stefan Kraus, Edgar R. Ligon, Olli Majoinen, John D. Monnier, Denis Mourard, Nicholas J. Scott, Norman L. Vargas

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to take a super-sharp photograph of a tiny, distant star using six giant telescopes working together. To get a picture this clear, you have to combine the light from all six telescopes into one beam, like merging six rivers into a single stream.

But here's the problem: The Earth is spinning.

As the Earth turns, the distance the light has to travel from each telescope to the center changes constantly. It's like trying to merge six rivers flowing at different speeds; if you don't adjust the channels perfectly, the water (light) won't mix, and you'll just get a muddy mess instead of a clear stream.

This is where the CHARA Array's Delay Lines come in. They are the "traffic controllers" of the telescope system.

The Old System: The Aging Conductor

For over 20 years, CHARA used an old control system to manage these delay lines. Think of it like a conductor leading an orchestra who was using a paper score from the 1990s. It worked fine for a long time, but:

The paper was getting brittle (parts were hard to find).
If a musician missed a beat, the conductor couldn't fix it fast enough.
Eventually, the whole orchestra had to stop while they waited for a replacement part to be shipped from a factory.

The Upgrade: The High-Tech Conductor

In 2021, the team decided to replace the old conductor with a modern, high-tech AI system. They swapped out the old computer architecture for a new hybrid system using FPGAs (super-fast, specialized chips) and Linux (a modern operating system).

Think of this new system as a high-speed drone delivery network replacing a slow mail service.

Speed: It can react in milliseconds.
Precision: It can adjust the light path with the accuracy of a hair's width (actually, about 12 nanometers—that's thinner than a single virus!).
Reliability: It doesn't need to wait for parts to be shipped; it can diagnose and fix its own minor glitches.

The Bumps in the Road (Commissioning)

When they first turned on the new system, it wasn't perfect. It was like buying a brand-new sports car that had a few software bugs.

The "Jittery" Clock: The system's internal clock was slightly "jittery," like a drummer who can't keep a steady beat. This caused the delay lines to make sudden, jerky movements, blurring the star's image. The team fixed this by writing a software patch to smooth out the rhythm.
The Vibration: One of the telescopes (S1) was losing image quality because a power supply for a cable puller was vibrating the floor, like a washing machine on the spin cycle next to a delicate painting. They moved the power supply to the ground, and the vibrations stopped.

The Result: Crystal Clear Views

After fixing these issues, the new system is a powerhouse.

The "Fringe Tracking": Imagine trying to balance a broom on your finger. The atmosphere is like a gusty wind trying to knock the broom over. The new delay lines are so fast and precise that they can adjust the broom's position 100 to 130 times every second, keeping it perfectly balanced.
The Outcome: Astronomers can now see details on stars that were previously invisible. They can map the surface of dying stars, watch binary stars dance around each other, and even look for dust disks where new planets might be forming.

What's Next?

The team isn't stopping here. They are planning to:

Extend the Reach: Make the telescopes work together over even longer distances (up to 1 km), which would act like a giant magnifying glass, allowing them to see even smaller details.
Double Vision: Look at two stars at once—one bright one to stabilize the view, and a faint one to study.
Block the Bright Light: Use "nulling" to block out the blinding light of a star so they can see the tiny, dim planets orbiting it, much like putting your hand over a streetlamp to see the stars behind it.

In short: The CHARA Array took its aging, reliable but slow control system, gave it a massive upgrade to a modern, lightning-fast brain, and fixed a few glitches along the way. The result is a telescope array that can see the universe with unprecedented clarity, opening the door to discovering new worlds and understanding how stars live and die.

1. Problem Statement

The CHARA Array, a long-baseline optical/infrared interferometer at Mount Wilson Observatory, relies on a delay line system to equalize optical path lengths (OPL) for its six telescopes as the Earth rotates and to compensate for atmospheric turbulence.

Legacy System Limitations: The original delay line control system, operational for over two decades, was based on a VME architecture with PowerPC processors running the VxWorks real-time operating system.
Operational Risks: The aging hardware made acquiring replacement parts increasingly difficult, leading to frequent downtime for repairs.
Performance Gaps: While the legacy system was reliable, it lacked the modularity and modern telemetry capabilities required for next-generation observing modes (e.g., dual-field interferometry, nulling) and high-precision fringe tracking across a broad wavelength range (R to K bands).
Scientific Requirements: To maintain fringe visibility contrast losses $\le$ 5% (specifically in the R-band), the system requires cart tracking errors ( $\epsilon$ ) $\le$ 20 nm RMS and a control bandwidth capable of correcting atmospheric fluctuations faster than the coherence time ( $\tau_0 \approx 6$ ms).

2. Methodology and System Architecture

The paper details a comprehensive modernization of the delay line control system, completed between mid-2021 and the end of 2024.

Hardware Architecture:
- Hybrid FPGA/Linux System: The legacy VME system was replaced with a hybrid architecture. A dedicated Linux workstation controls each of the six delay line carts (modular design), while a timing-critical Metrology FPGA handles real-time servo loops.
- Actuators: The system utilizes a cascaded servo approach with four stages:
  1. PZT (Piezoelectric Transducer): High-speed, small-range correction (5 kHz loop).
  2. Voice Coil 1 (VC1): Keeps the PZT centered (1 kHz loop).
  3. Voice Coil 2 (VC2): Keeps the optics cart centered (1 kHz loop).
  4. Stepper Motor: Coarse positioning (100 Hz loop).
- Metrology: A 1.319 µm heterodyne laser interferometer provides absolute, unwrapped path-length measurements with nanometer-level accuracy.
Software and Control Logic:
- FPGA: Runs a 32-bit NIOS soft-core processor (100 MHz) handling the 5 kHz PZT servo loop and 1 kHz target updates. It receives timing synchronization (16 MHz/1 Hz) from a GPS-synchronized master clock.
- Linux Backend: Runs OPLECtrlQt5 software to manage voice coils, stepper motors, and telemetry. It acts as a supervisor, receiving baseline solutions and fringe-tracking offsets.
- Communication: A dedicated low-latency Ethernet connection links the fringe tracker (e.g., SPICA-FT) directly to the metrology FPGA, bypassing the main server to minimize latency.

3. Key Contributions and Commissioning Challenges

The authors document the transition from the legacy system to the new architecture, highlighting specific technical hurdles encountered during commissioning and their resolutions:

Time-Tick Jitter (Critical Issue): Early operations revealed systematic tracking errors of $\sim$ $\sim$ 80 nm RMS caused by "time-tick jitter" in the FPGA's internal 5 kHz loop. Discontinuities in the 40 kHz counter caused erroneous target position spikes proportional to cart velocity.
- Resolution: A software fix rounded tick values to multiples of 8, restoring synchronization and reducing errors to $<12$ nm.
Bandwidth Reduction: To mitigate the initial spikes, PID gains were conservatively tuned, inadvertently reducing system bandwidth and hindering fast fringe tracking.
- Resolution: Once the jitter was fixed, PID gains were re-optimized, restoring the full control bandwidth.
Mechanical Vibrations: Persistent visibility loss on the S1 telescope was traced to mechanical vibrations from a cable puller power supply mounted on the vacuum beam tube.
- Resolution: Relocating the power supply to the ground eliminated the vibration coupling, restoring fringe contrast.
Clock Synchronization: Intermittent boot-time synchronization failures between the FPGA and the OPLE computer caused large OPD deviations ( $\pm400$ $\pm 400$ nm).
- Resolution: A diagnostic routine was established to detect and reboot/sync affected units automatically.

4. Performance Results

Post-upgrade characterization demonstrates that the system meets and exceeds the technical requirements for high-precision interferometry:

Tracking Accuracy: The upgraded system achieves residual cart tracking errors of $\sim$ 12 nm RMS (ranging from 5–12 nm across the six carts), well within the 20 nm requirement for R-band observations.
Control Bandwidth: The system operates with a $-3$ dB bandwidth of 100–130 Hz, sufficient to suppress the majority of atmospheric piston variations.
Latency: The end-to-end latency (fringe tracker command to actuation) is approximately 2 ms. This allows for effective fringe tracking at rates up to 500 Hz (SPICA-FT), though phase lag becomes a limiting factor near the R-band coherence time ( $\tau_0 \approx 6$ ms).
Fringe Visibility:
- The delay lines contribute $\le$ 2% to fringe contrast degradation in the R-band and $<1\%$ in H/K bands.
- The system successfully supports simultaneous observations across R, H, and K bands using SPICA, MIRC-X, and MYSTIC instruments.
- High-contrast fringes with four resolved visibility nulls have been observed, demonstrating the system's capability for detailed imaging (e.g., limb-darkening measurements).

5. Significance and Future Directions

This paper serves as a critical reference for the CHARA Array's current operations and future instrument design:

Enabling Advanced Modes: The modernized, modular architecture supports future upgrades, including:
- Extended Baselines (CMAP): The system now supports cart speeds up to 20 mm/s, necessary for extending baselines to 1 km.
- Dual-Field Interferometry: Stable phase tracking allows for observing faint off-axis targets while tracking a bright reference star.
- Nulling Interferometry: The demonstrated stability ( $\sigma_{OPD} \approx 90-230$ nm) suggests feasibility for high-contrast nulling to detect exoplanets and exozodiacal dust.
Scientific Impact: The upgrades enable high-fidelity imaging of protoplanetary disks, binary stars, and evolved stars with unprecedented resolution ( $\sim$ 200 $\mu$ as).
Benchmarking: The performance metrics (tracking error and bandwidth) are comparable to or better than other leading facilities like the VLTI and Keck Interferometer, validating the hybrid FPGA/Linux approach for next-generation interferometric control systems.

In summary, the successful modernization of the CHARA delay lines has resolved legacy maintenance issues, restored and enhanced tracking performance, and laid the technical foundation for the next generation of high-resolution astrophysical observations.