CHARA Array Delay Lines: Upgrades, Performance and Future Directions

Abstract

Long baseline optical and infrared interferometric arrays achieve high angular resolution and enable detailed astrophysical measurements. Interferometers have enabled observations of stars at various stages of evolution, as well as studies of binary stars, circumstellar disks, and active galactic nuclei. The CHARA Array is a long-baseline interferometric array at the Mount Wilson Observatory, USA. At the core of CHARA operations are the delay lines, which equalize the optical path length for all telescopes as the Earth rotates and compensate for optical path variations induced by atmospheric turbulence. We report recent upgrades and performance of the CHARA Array optical delay lines for high-precision interferometric observations. The legacy system had been operational for over two decades, and it was increasingly difficult to acquire replacement parts. Beginning in mid-2021, the control system underwent a major upgrade, replacing the aging VME-based architecture with a modern hybrid FPGA and Linux-based system; this modernization continued through the end of 2024. We describe hardware/software changes, the servo architecture, and lab/on-sky performance. The upgraded system achieves residual delay line cart tracking errors of $\sim12$~nm, the same level as the legacy system, and a control bandwidth of 100-130~Hz, allowing fringe tracking across the R, H, and K bands. Initial commissioning revealed key issues such as metrology time-tick jitter and vibration-induced visibility loss, which were diagnosed and resolved. We note ongoing and future efforts to extend baselines up to 1~km and support advanced observing modes such as dual-field interferometry and nulling. This paper is a reference for current and future use of the CHARA Array and for next-generation instrument design.

Figures (17)

• Figure 1: Top: Overview of the CHARA Array. The six $1\,\mathrm{m}$ telescopes, named S1, S2, W1, W2, E1, and E2, are arranged in a Y-shaped configuration. Vacuum pipes transport starlight from the telescopes to the central beam combiner laboratory. Bottom: Photograph of the delay line system showing six optical carts that move on rails to equalize the optical path lengths for each telescope.
• Figure 2: Schematic of the CHARA beam train feeding the SPICA (visible), MIRC-X (near-IR; SPICA-FT), and MYSTIC (near-IR) beam combiners. Only one telescope is shown. In the SPICA+MIRC-X+MYSTIC mode, the telescope dichroic (TelDic) directs 20% of the visible light to the AO/WFS and transmits the remaining visible light together with the near-IR beam to the beam-combiner laboratory. After the common CHARA delay lines, laboratory dichroics (Lab-dic) separate the visible and near-IR beams to the SPICA and MIRC-X/MYSTIC beam-combiner tables. MIRC-X provides fringe-tracking corrections (SPICA-FT) applied through the common delay lines, while SPICA and MYSTIC record science data; internal delay stages compensate chromatic dispersion and non-common-path optical path differences. A longitudinal dispersion corrector located between the delay lines and the laboratory dichroics is not shown. The diagram shows the functional architecture; the full six-telescope lab layout is omitted.
• Figure 3: Schematic diagram of a delay line cart with a cat's eye retro-reflector used at CHARA. The telescope incoming beam enters the cat's eye system and reflects off one side of a paraboloid mirror and then focuses onto a small secondary flat mirror, which redirects the beam back through the second side of paraboloid. The result is a re-collimated output beam traveling parallel to the input. The secondary flat mirror is mounted on a high-bandwidth PZT stage to enable fine optical path-length adjustments. The PZT, Voice Coil 1 and Voice Coil 2 and stepper motor work in a cascaded servo system.
• Figure 4: CHARA OPLE control system architecture and hardware layout. The left panel shows the physical hardware rack for the E1 and E2 delay lines, which includes FPGA-based metrology systems, Linux control computers, and PZT and voice coil voltage amplifiers (see Sec. \ref{['sec_architecture']}). The right panel presents the control logic. At the core of the system is the time critical Metrology FPGA, which receives absolute laser-based position measurements and high-precision clock signals Sturmann1998 ($16\,\mathrm{M}\mathrm{Hz}$ and $1\,\mathrm{Hz}$). It operates two control loops: (i) a $5\,\mathrm{k}\mathrm{Hz}$ PZT servo loop, and (ii) a $1\,\mathrm{k}\mathrm{Hz}$ loop that handles slower tasks such as target position updates and the transmission of telemetry data—including PZT positions and cart positions—to the Linux computer. The soft real-time Linux server (running OPLECtrlQt5) acts as the supervisor and controls the voice coils and stepper motors based on PZT offloads and eddy current sensors — at $1\,\mathrm{k}\mathrm{Hz}$. The ople server and opletab graphical user interface, operated by the CHARA telescope operator, handle cart selection, homing, baseline solution activation, and monitoring. The ople server sends baseline solution model, accounting for sidereal motion and array geometry, which are applied to the delay line motion at speeds of up to $8\,\mathrm{m}\mathrm{m}\mathbin{/}\mathrm{s}$. The fringe tracker (e.g., SPICA-FT, MIRC-X, or MYSTIC) sends atmospheric residual OPD offsets directly to the metrology FPGA, which are corrected by the PZT servo.
• Figure 5: Nested (cascaded) OPLE tracking servos. The metrology error drives the PZT loop (fast), which is offloaded sequentially to VC1 (optics cart centering), VC2 (stepper interface centering), and the stepper motor (coarse positioning). See Sec. \ref{['sec_architecture']} for details.