Prime-Cam: CCAT Submillimeter Instrument

• Prime-Cam is a modular submillimeter instrument featuring a 1.8 m cryogenic receiver with up to seven independent modules for versatile observations.
• It combines broadband polarization-sensitive imaging and spectrometer modules across 100–900 GHz to address cosmology, galactic polarization, and star formation.
• The design supports over 100,000 MKID detectors, promising a more than 10× increase in mapping speed compared to current facilities.

Searching arXiv for recent Prime-Cam / CCAT papers to ground the article. Prime-Cam is the first-generation science instrument for the Fred Young Submillimeter Telescope (FYST) at the CCAT Observatory: a 1.8-meter diameter cryogenic receiver designed to exploit FYST’s high-throughput crossed-Dragone optics and 5600 m Cerro Chajnantor site for wide-field and deep mapping between 100 and 900 GHz (Vavagiakis et al., 2022). In its planned configuration, Prime-Cam houses up to seven independently developed instrument modules, combining broadband polarization-sensitive imaging modules and imaging spectrometer modules across frequency windows between 210 GHz and 850 GHz, with microwave kinetic inductance detectors (MKIDs) or KIDs as the baseline sensor technology (Vavagiakis et al., 2022). Across the full instrument, Prime-Cam is designed to field over $10^5$ detectors, and multiple papers describe it as the largest submillimeter focal plane under development or built to date (Gazda et al., 14 Oct 2025). Its science program spans Big Bang cosmology, Sunyaev–Zel’dovich measurements of galaxy clusters, Galactic polarization, star formation, dusty galaxy evolution, and line-intensity mapping during the Epoch of Reionization (Vavagiakis et al., 2022).

1. Instrument role within FYST and CCAT

FYST is a 6 m crossed-Dragone submillimeter telescope on Cerro Chajnantor in Chile’s Atacama Desert at an elevation of 5600 m, optimized for a wide field of view and high optical throughput (Vavagiakis et al., 2022). Prime-Cam is the main first-generation science camera for this telescope and is explicitly intended to “take advantage of the high-efficiency telescope and high-elevation site to enable wide-field and deep mapping between 100 and 900 GHz” (Vavagiakis et al., 2022). In the more instrument-specific descriptions, Prime-Cam is a modular, direct-detection receiver in a 1.8 m diameter cryostat that can host seven instrument modules, each intercepting a patch of the telescope focal plane and carrying its own optics, filters, detectors, and readout interfaces (Freundt et al., 2024).

The module concept is central to Prime-Cam’s architecture. The cryostat can house up to seven $\sim 41$ cm diameter modules, each with up to $1.3^\circ$ field of view, filling a total of $4.9^\circ$ of FYST’s $8^\circ$ diameter field of view at 3 mm (Vavagiakis et al., 2022). When fully populated, Prime-Cam is planned to field up to five broadband polarization-sensitive modules for observations between 220 and 850 GHz and at least two imaging spectrometer modules for line intensity mapping from 210 to 420 GHz (Vavagiakis et al., 2022). A later optics status update describes the initial funded configuration as three broadband imaging modules near 280, 350, and 850 GHz plus one EoR-Spec spectrometer module, with the remaining module positions reserved for later staged population (Huber et al., 2024).

Earlier Prime-Cam design studies described a detector mix including multichroic TES bolometers between 190 and 450 GHz and KIDs at 860 GHz (Vavagiakis et al., 2018). Later work converged on KIDs and MKIDs across the currently planned modules, explicitly stating that all currently planned Prime-Cam modules will use microwave kinetic inductance detectors to meet desired sensitivity and detector-density requirements (Vavagiakis et al., 2022). This evolution reflects the importance of scalable microwave multiplexing for large submillimeter focal planes. A plausible implication is that the instrument concept became more strongly unified around a common KID/MKID readout ecosystem as hardware matured.

Prime-Cam’s scientific context is unusually broad. The collaboration consistently frames it as a survey instrument addressing “science goals ranging from Big Bang cosmology through reionization and the formation of the first galaxies to energetic transients, galaxy cluster evolution via the Sunyaev-Zel’dovich (SZ) effects, galactic polarization and star formation within the Milky Way” (Vavagiakis et al., 2022). Sensitivity forecasts for the full instrument describe 60,000 polarimetric KIDs at combinations of 220/280/350/410 GHz, 31,000 KIDs at 250/360 GHz coupled with Fabry–Perot interferometers for EoR science, and 21,000 polarimetric KIDs at 850 GHz (1908.10451). More recent descriptions instead emphasize 280, 350, and 850 GHz broadband modules and 210–420 GHz spectrometers (Freundt et al., 2024). This suggests an instrument program that preserves the same broad scientific reach while refining specific module deployment priorities.

2. Cryostat, module geometry, and optical architecture

Prime-Cam is repeatedly described as a 1.8 m diameter cryogenic receiver (Vavagiakis et al., 2022). A detailed optics update describes it as a large vacuum vessel with nominal 80 K, 40 K, 4 K, 1 K, and 100 mK temperature stages, cooled by one Cryomech PT-90, two Cryomech PT-420s, and a Bluefors dilution refrigerator (Huber et al., 2024). Internal G10 support tabs provide mechanical support and thermal isolation, while the 4 K stage is the principal mechanical mount for the optics tubes or instrument modules (Huber et al., 2024). The modules themselves are self-contained cold reimaging systems, each with its own silicon lenses, filters, Lyot stop, and focal plane.

A recurring design lineage connects Prime-Cam to the Simons Observatory Large Aperture Telescope Receiver optics tubes. The 280 GHz module, Mod-Cam receiver, EoR-Spec module, and the general optics updates all state that Prime-Cam modules inherit or adapt Simons Observatory optics-tube design principles (Vavagiakis et al., 2022). This includes AR-coated silicon lenses, cryogenic filter stacks, Lyot stops, and module-level thermal staging at 4 K, 1 K, and 100 mK (Vavagiakis et al., 2022). In the single-module Mod-Cam precursor, the 280 GHz optics tube has a 4 K stage carrying the front optics and first silicon lens, a 1 K stage carrying later lenses and the Lyot stop, and a 100 mK stage carrying feedhorn and detector-array packages (Vavagiakis et al., 2022). Prime-Cam is explicitly designed to mount up to seven of the same module format on its 4 K plate (Vavagiakis et al., 2022).

The optical chain in the broadband modules is based on reimaging through several cryogenic lenses and a pupil stop. In the 280 GHz module, light passes through an AR-coated UHMWPE vacuum window, warm and cryogenic IR-blocking filters, low-pass edge filters, and a sequence of three metamaterial-AR-coated silicon lenses, with the Lyot stop at 1 K and a final low-pass filter near the feedhorn arrays (Vavagiakis et al., 2022). Silicon is preferred for “high resistivity, extremely low loss, high thermal conductivity… and high index of refraction,” and CNC-machined metamaterial AR coatings are described as achieving “<1% reflection across an octave of bandwidth” (Vavagiakis et al., 2022). These design choices recur across the module suite, though the number of lenses and exact stop placement vary by module.

The optics status paper frames Prime-Cam as a set of specialized but mechanically compatible optical systems. The 280 GHz broadband module reuses the Simons Observatory 220/280 GHz optics design; the 350 GHz module adopts the same mechanical layout while accepting a trade-off between Strehl and beam ellipticity; the 850 GHz module uses a more demanding four-lens design for a smaller, higher-quality central field; and EoR-Spec uses a four-lens optical train optimized for a highly collimated Lyot stop hosting a Fabry–Perot interferometer (Huber et al., 2024). In this sense, Prime-Cam is not a single camera in the conventional monolithic sense but an instrument platform for multiple co-located, cryogenically integrated cameras and spectrometers.

The cryogenic distribution system also reflects this platform architecture. A later optics update describes 1 K and 100 mK copper “BUS” structures mounted at the back of the cryostat and fed from the dilution refrigerator, with each module connected via copper straps to the common sub-Kelvin buses (Huber et al., 2024). Split BUS concepts are being explored to preserve serviceability of lower modules without removing the full thermal manifold (Huber et al., 2024). This design directly supports staged deployment and module replacement, which is a defining operational feature of Prime-Cam.

3. Frequency coverage, detector populations, and mapping-speed concept

Prime-Cam’s overall frequency coverage is described in several overlapping ways. The broadest statement is “between 100 and 900 GHz” for wide-field and deep mapping (Vavagiakis et al., 2022). For the MKID-based instrument as framed in the Mod-Cam and detector papers, broadband and spectroscopic measurements span 280–850 GHz, while the division between module classes is broadband polarization-sensitive modules between 220 and 850 GHz and imaging spectrometer modules for line intensity mapping from 210 to 420 GHz (Vavagiakis et al., 2022). The 2025 detector optimization paper narrows the currently active KID development bands to 280–850 GHz and identifies broadband modules at 280, 350, and 850 GHz, plus imaging spectrometers including EoR-Spec covering roughly 210–420 GHz (Gazda et al., 14 Oct 2025).

The detector count is one of Prime-Cam’s defining quantitative features. The Mod-Cam precursor paper states that when populated with seven instrument modules, each deploying three MKID arrays, Prime-Cam will field a total of $>100{,}000$ detectors, “larger than any deployment of broadband KIDs yet” (Vavagiakis et al., 2022). The readout software paper translates this to infrastructure: each Xilinx ZCU111 RFSoC board can drive up to 4000 KIDs in the current configuration, and Prime-Cam is implementing approximately 25 boards, giving practical support for order-$10^5$ detectors (Burgoyne et al., 2024). The 2025 Mod-Cam readout papers describe Prime-Cam as fielding approximately 100,000 KIDs across seven instrument modules, with readout harnesses and stripline technology inherited from the Mod-Cam pathfinder (Keller et al., 29 Sep 2025). The 2025 detector characterization paper states that over 100,000 sensors are under development and that these arrays will soon form the largest submillimeter focal plane ever built (Gazda et al., 14 Oct 2025).

A concise module-by-module summary from the supplied papers is given below.

Module class	Frequency coverage	Detector/module notes
Broadband imaging	220–850 GHz	Up to five polarization-sensitive modules (Vavagiakis et al., 2022)
EoR-Spec spectrometer	210–420 GHz	$R\sim100$, FPI-based line-intensity mapping (Huber et al., 2022)
280 GHz broadband module	280 GHz	$>10{,}000$ feedhorn-coupled polarization-sensitive MKIDs in three arrays (Vavagiakis et al., 2022)
850 GHz broadband module	850 GHz	$\sim40{,}000$ KIDs, 1.1° diameter field (Chapman et al., 2022)

Prime-Cam’s mapping-speed claim is stated in headline form rather than derived from an explicit formula in the main design papers. The Mod-Cam paper says that Prime-Cam “will deliver over ten times greater mapping speed than current and near-term facilities” (Vavagiakis et al., 2022). The optics status paper repeats the same claim for 210–850 GHz submillimeter experiments (Huber et al., 2024). The sources do not provide a single analytical mapping-speed equation for Prime-Cam, but they repeatedly point to the combination of FYST’s large throughput, a high-elevation site, large field of view, and very high detector counts as the basis for this gain (Vavagiakis et al., 2022). This suggests that the instrument’s performance advantage is primarily architectural: large étendue filled with densely multiplexed KIDs rather than incremental sensitivity improvements at fixed focal-plane size.

Sensitivity forecasts for earlier Prime-Cam configurations provide more detailed, though historically earlier, quantitative framing. The 2019 sensitivity paper models wide 15,000 deg² surveys, 410 deg² star-formation surveys, and 8 deg² [C II] targeted spectroscopy, using detector counts corresponding to a fully populated seven-module instrument (1908.10451). Beam sizes are tabulated from 57″ at 220 GHz to 14″ at 850 GHz, with sensitivity forecasts expressed in NEI, NET, map depth, and angular noise spectra (1908.10451). Because later papers revised module band allocations and emphasized KIDs across the instrument, these numbers are best interpreted as forecasts for a closely related Prime-Cam configuration rather than immutable final specifications.

4. Detector technology: KIDs, MKIDs, LEKIDs, and focal-plane implementations

Prime-Cam’s current module suite is based on kinetic inductance detector technology. The Mod-Cam precursor paper states that “all of the currently planned instrument modules for Prime-Cam will use microwave kinetic inductance detectors (MKIDs)” (Vavagiakis et al., 2022). The readout software paper specifically refers to lumped-element kinetic inductance detector arrays, LEKIDs, fabricated by NIST for the relevant configuration (Burgoyne et al., 2024). The detector principle is described consistently: incident photons are coupled to a superconducting inductive element of an LC resonator; when Cooper pairs are broken, the kinetic inductance changes, shifting the resonant frequency and allowing the optical signal to be inferred from microwave transmission changes (Vavagiakis et al., 2022).

The 280 GHz module provides the clearest concrete example of a Prime-Cam broadband focal plane. It contains more than 10,000 feedhorn-coupled, polarization-sensitive MKIDs divided between three tiled array packages with two unique designs (Vavagiakis et al., 2022). Array 1, using TiN and aluminum feedhorns, has 3456 detectors, of which 3450 are optically coupled. Arrays 2 and 3, using Al and silicon feedhorns, each have 3448 detectors, of which 3418 are optically coupled (Vavagiakis et al., 2022). Each pixel is polarization-sensitive, with two MKIDs per spatial pixel (Vavagiakis et al., 2022). The mixed TiN/Al design was introduced after dark testing showed reduced low-frequency spectral noise in Al devices, while the TiN array preserves continuity with the earlier design lineage (Vavagiakis et al., 2022).

The 2025 optimization paper deepens this comparison by describing witness-pixel measurements from the 280 GHz TiN and Al arrays (Gazda et al., 14 Oct 2025). TiN KIDs have $\sim 41$0 K and Al KIDs have $\sim 41$1 K (Gazda et al., 14 Oct 2025). The inductor is a cross or plus-shaped absorber located at the mouth of a feedhorn-coupled waveguide, while the capacitor is an interdigitated rectangular structure; together they set the resonant frequency in the few-GHz readout band (Gazda et al., 14 Oct 2025). TiN devices show steeper, nearly linear responsivity versus optical power but more pronounced $\sim 41$2-like noise, whereas Al devices show lower and more nonlinear responsivity with different noise characteristics (Gazda et al., 14 Oct 2025). The paper explicitly concludes that these trade-offs inform material choice and design details for the deployed arrays (Gazda et al., 14 Oct 2025).

The 850 GHz module pushes the detector-density concept much further. It is described as deploying $\sim 41$3 KIDs over a 1.1° diameter field, coupled through silicon platelet feedhorns (Chapman et al., 2022). A more detailed design paper gives a refined value of 41,400 KIDs, with a 1.42 mm pixel pitch and $\sim 41$4, distributed over three silicon feedhorn arrays (Huber et al., 2022). The 850 GHz detector design uses TiN with $\sim 41$5 K and explicitly introduces “junk inductance” to maintain capacitor-finger widths around 5 µm while fitting the much denser pixel pitch required at 350 µm (Chapman et al., 2022). This design is a direct consequence of Prime-Cam’s requirement to sustain high multiplexing density at the shortest wavelength.

EoR-Spec represents the spectroscopic branch of the detector program. The 2024 status update states that EoR-Spec will feature more than 6500 MKIDs on three non-polarization-sensitive arrays: two low-frequency arrays covering 210–315 GHz and one high-frequency array covering 315–420 GHz (Freundt et al., 2024). The low-frequency arrays each have 1728 pixels centered near 265 GHz, while the high-frequency array is planned with 3072 pixels (Freundt et al., 2024). These arrays are front-illuminated through monolithic aluminum feedhorn arrays and are read out through multiple RF networks with LNAs at 4 K, using the same RFSoC-based infrastructure that serves the broadband modules (Freundt et al., 2024).

A common misconception is that Prime-Cam is a single detector technology demonstration centered only on one broadband band. The detector papers show the opposite: the 280 GHz module is a pathfinder, the 850 GHz module is a high-density specialized imager, and EoR-Spec is a spectroscopic implementation with different optical and array-level constraints. What unifies them is not a single focal-plane design but the use of superconducting resonator arrays and highly multiplexed microwave readout.

5. Readout, software, calibration, and operational optimization

Prime-Cam’s readout architecture is one of its most technically distinctive features. The 2022 RFSoC readout paper presents a baseline firmware design that can read out four independent RF networks, each with 500 MHz of bandwidth and 1000 detectors, for about 30 W on a Xilinx RFSoC (Sinclair et al., 2022). The later readout software paper states the same architecture operationally: up to 4 RF networks per ZCU111 board, each with 500–512 MHz usable bandwidth and up to 1000 resonators, giving 4000 KIDs per board (Burgoyne et al., 2024). Prime-Cam is planning to use approximately 25 boards in this configuration (Burgoyne et al., 2024). The signal chain uses digital waveform playback from RAM to DAC, complex downconversion around a local oscillator, channelization to each tone, extraction of per-detector $\sim 41$6 and $\sim 41$7, and UDP packetization of timestreams to the data acquisition computer (Burgoyne et al., 2024).

The software framework primecam_readout is a distributed system built around a centralized control computer and multiple RFSoC boards (Burgoyne et al., 2024). A control process, queen.py, communicates via Redis pub/sub with up to four drone.py instances per board, each controlling one RF network (Burgoyne et al., 2024). Board-side commands include setting the local oscillator, writing VNA and target tone combs, sweeping the band, identifying resonator peaks, and choosing calibration tones (Burgoyne et al., 2024). Because commands can be sent to one board, one network, or broadcast across the full system, the framework provides parallel asynchronous control that scales with the number of boards rather than the total detector count (Burgoyne et al., 2024).

The standard operational flow is sweep-based. The system first generates a VNA comb of 1000 equally spaced tones across the 512 MHz baseband, then performs a VNA sweep by stepping the LO to map $\sim 41$8, then identifies resonances and writes a targeted comb centered near each one (Burgoyne et al., 2024). A finer “target sweep” refines the resonant frequencies, after which the final science comb is placed not exactly at the resonant minimum but on the leading edge of the resonance curve, where responsivity is high and dynamic range for frequency shifts is preserved (Burgoyne et al., 2024). This logic is closely mirrored in the later RFSoC demonstration paper, which uses VNA-style sweeps, candidate-resonance identification from $\sim 41$9 minima with positive phase slope, higher-resolution target sweeps, and final fixed-tone timestream acquisition for the full $1.3^\circ$0-detector 280 GHz module (Patel et al., 7 Oct 2025).

The conversion from raw $1.3^\circ$1 to detector observables is also standardized. The readout software paper adopts an IQ-angle method derived from Gao, in which a target sweep defines an approximately circular IQ loop, the loop center is estimated, and the observed angle $1.3^\circ$2 is mapped to frequency shift $1.3^\circ$3 through an interpolation $1.3^\circ$4 (Burgoyne et al., 2024). The 2025 detector optimization paper instead describes fitting a cubic polynomial to the phase-frequency relation $1.3^\circ$5 from narrowband sweeps and inverting it to map $1.3^\circ$6 to $1.3^\circ$7 (Gazda et al., 14 Oct 2025). Both are low-cost nonlinear calibration approaches designed to support large arrays and non-negligible resonator motion under changing optical load.

Several papers emphasize that Prime-Cam performance depends not only on detector design but on active readout optimization. The 2025 optical responsivity and noise study shows that moving the readout power 10 dB away from the optimal setting can degrade NEP by up to $1.3^\circ$8 at a given optical loading, and that detuning the probe tone by roughly half a linewidth due to loading drift can worsen the effective NEP by a factor of $1.3^\circ$9 for an optical power change of $4.9^\circ$0 pW (Gazda et al., 14 Oct 2025). The recommended operating point is about 1–2 dB below bifurcation for each resonator, with periodic tone retuning to keep detuning significantly below about half a linewidth (Gazda et al., 14 Oct 2025). This is operationally important because Prime-Cam is intended for long-duration surveys under changing atmospheric loading rather than static laboratory conditions.

Scalability has been validated experimentally. Mod-Cam’s 280 GHz module has 18 RF networks, and the 2025 RFSoC paper reports simultaneous readout of all 18 networks using five RFSoCs (Patel et al., 7 Oct 2025). The stripline performance paper describes six 46 cm flexible striplines carrying 18 RF chains for the current Mod-Cam module, while Prime-Cam’s first harness is designed around 18 stripline assemblies to read out three instrument modules and provide 108 RF chains, with some held in reserve as spares (Keller et al., 29 Sep 2025). Electrical testing found that the striplines themselves meet project crosstalk requirements, with inter-channel crosstalk more than 50 dB below transmission, while transition PCBs were identified as the dominant crosstalk source and redesigned accordingly (Keller et al., 29 Sep 2025). This matters because Prime-Cam’s scale makes RF cleanliness a system-level requirement rather than a component-level luxury.

6. Module-specific implementations and scientific use cases

The 280 GHz module is the first deployed broadband Prime-Cam module and functions as the template for the broader module architecture (Vavagiakis et al., 2022). It is polarization-sensitive, feedhorn-coupled, and read out through 18 RF networks across three arrays (Vavagiakis et al., 2022). In scientific terms, it is repeatedly associated with Galactic dust, high-frequency SZ measurements, continuum emission from dusty galaxies, and general broadband survey capability (Vavagiakis et al., 2022). The calibration and readout optimization work on TiN and Al witness pixels is explicitly tied to the 280 GHz module’s ability to meet on-sky mapping-speed and calibration goals (Gazda et al., 14 Oct 2025).

The 350 GHz broadband module is mechanically similar to the 280 GHz design but occupies a higher-frequency regime where telescope aberrations and beam ellipticity trade-offs become more significant (Huber et al., 2024). The optics team evaluated biconic lenses to improve Strehl, but ultimately retained the 280 GHz lens design for the 350 GHz module because the gain in Strehl and FWHM came at the cost of increased PSF ellipticity and more complex AR-coating requirements (Huber et al., 2024). This choice illustrates a recurring Prime-Cam design philosophy: modest optical gains are not automatically accepted if they complicate beam systematics or fabrication.

The 850 GHz module is the most specialized broadband imager in the system. It is the highest-frequency module, occupies the central position in Prime-Cam, and is described as deploying around 40,000 or 41,400 KIDs over a 1.1° diameter field (Chapman et al., 2022). The optical design paper explicitly explores the trade space between $4.9^\circ$1 spacing, beam size, pixel sensitivity, and detector count, settling on a four-lens design with $4.9^\circ$2, 1.42 mm pixel pitch, and 41,400 KIDs to increase mapping speed by about 80% relative to the previous baseline (Huber et al., 2022). The module is motivated by surveys of dusty star-forming galaxies, dust polarization for CMB foreground characterization, protostellar monitoring, and Milky Way magnetic fields (Chapman et al., 2022).

EoR-Spec is the dedicated spectroscopic module. Its 2022 optical design paper describes a four-lens cold optical train with a Fabry–Perot interferometer at the Lyot stop, stepping through 210–420 GHz to measure redshifted [C II] emission from $4.9^\circ$3 to 8 at $4.9^\circ$4 (Huber et al., 2022). A key optical requirement is a highly collimated beam at the stop; the final design targets $4.9^\circ$5 there and uses four lenses rather than the three used in other Prime-Cam modules (Huber et al., 2022). Biconic and then all-aspheric variants were explored, with later status updates settling on an all-aspheric four-lens design and placing the FPI and Lyot stop at 4 K rather than 1 K to ease thermal loading and improve serviceability (Freundt et al., 2024). EoR-Spec is planned to carry out the CCAT Deep Spectroscopic Survey over the E-COSMOS and E-CDFS fields, with about 4000 hours over five years (Freundt et al., 2024).

From a scientific-program perspective, Prime-Cam’s module complement is deliberately complementary. Broadband imaging modules provide high-throughput continuum and polarization surveys; EoR-Spec provides tomographic line-intensity mapping. The 2024 EoR-Spec update notes that this enables cross-correlation between [C II] tomography, broadband continuum maps, and external surveys, while the 850 GHz module strengthens dust SED constraints and foreground characterization (Freundt et al., 2024). This modular complementarity is a more accurate description of Prime-Cam than the simplistic image of a multiband camera with uniform detectors and optics.

7. Mod-Cam as precursor, deployment path, and operational status

Mod-Cam is Prime-Cam’s direct pathfinder. It is described as a first-light and commissioning instrument for FYST, a single instrument module cryogenic receiver scaled down from the Prime-Cam design, and a dedicated testbed for Prime-Cam modules (Vavagiakis et al., 2022). It is 0.9 m in diameter, 1.8 m long, and reproduces the essential mechanical, thermal, optical, and readout interfaces of one Prime-Cam module (Vavagiakis et al., 2022). A defining feature is easier swapping of instrument modules due to its off-axis dilution refrigerator and faster turnaround relative to the larger cryostat (Vavagiakis et al., 2022).

This precursor role extends beyond optics. The cryogenic readout is explicitly partitioned in both Mod-Cam and Prime-Cam into a shared harness from 300 K to 4 K, individual instrument-module wiring from 4 K to 100 mK, and an isothermal 4 K transition between them (Vavagiakis et al., 2022). Mod-Cam modifies the harness details to prioritize modularity, but the module-side RF architecture is intentionally shared (Vavagiakis et al., 2022). The 2025 stripline paper makes the same point for flexible stripline interconnects: Mod-Cam’s six 46 cm striplines and transition PCBs were used to optimize attenuation, verify simultaneous 18-network readout, characterize thermal conductivity, and identify PCB-induced crosstalk that then informed the Prime-Cam redesign (Keller et al., 29 Sep 2025).

Cryogenic performance has also been validated in Mod-Cam under Prime-Cam-like conditions. In an optically open configuration, Mod-Cam achieved stable base temperatures of 1.5 K on the 1 K stage and 85 mK at the detector stage; in a cold-load configuration it achieved detector focal-plane RMS temperature stability of $4.9^\circ$6 K (Lin et al., 29 Sep 2025). Using measured thermal responsivities and optical responsivities from related detector work, the resulting equivalent power from focal-plane thermal fluctuations was shown to be only 0.0040% of a 5 pW incident photon power for aluminum detectors and 0.0023% for titanium-nitride detectors (Lin et al., 29 Sep 2025). The paper concludes that these fluctuations are negligible for CCAT science goals and thereby validates the cryogenic architecture that Prime-Cam inherits (Lin et al., 29 Sep 2025).

Deployment timelines have evolved across the literature. The 2022 Mod-Cam paper states that FYST would begin assembly for first light in 2024 and that Mod-Cam deployment with a 280 GHz instrument module was planned for early science observations in 2024 (Vavagiakis et al., 2022). Later papers shift the horizon: the 2025 detector optimization paper states that the laboratory work is intended to support FYST’s planned first light in 2026 (Gazda et al., 14 Oct 2025), and the 2024 EoR-Spec status paper states a goal of first light on FYST in 2026 for EoR-Spec (Freundt et al., 2024). These differences reflect schedule evolution rather than conceptual changes.

Prime-Cam itself is repeatedly described as under construction. The 2022 Mod-Cam paper states that the receiver was under construction by Redline Chambers for delivery to Cornell University for initial testing in 2022 (Vavagiakis et al., 2022). The 2024 optics status paper reports that major shells had been machined and fit-checked, internal G10 supports fabricated, and delivery to Cornell expected by the end of 2024, followed by fit checks, vacuum leak testing, and cryogenic testing (Huber et al., 2024). In parallel, optics for the 350 GHz and EoR-Spec modules had been submitted for fabrication, while the 280 GHz module was already undergoing cryogenic testing in Mod-Cam (Huber et al., 2024).

Prime-Cam therefore occupies an intermediate stage between concept and full observatory deployment: sufficiently mature that module-level optics, readout, and cryogenic designs are detailed and being validated in hardware, but still evolving in deployment sequence, exact module mix, and operational procedures. This does not indicate instability in the instrument concept. Rather, it reflects the reality of a modular observatory instrument whose modules, readout, and cryogenic subsystems are being commissioned on partly independent paths.

A final important point is terminological. “Prime-Cam” in the CCAT/FYST literature refers to the CCAT Observatory cryogenic receiver described above, not to unrelated prime-focus metrology cameras used in the Subaru Prime Focus Spectrograph project (Wang et al., 2014). The latter are distinct optical metrology systems at an entirely different observatory and instrumentation context.