{"id":"a0455beb-10fa-4273-8ced-d1a944830b60","arxiv_id":"2505.07986","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"τHK is a Eurocard-based housekeeping system that reads resistive and diode thermometers, drives cryogenic heaters, and switches payload power, using an STM32 microcontroller instead of an FPGA.","lead":"τHK is a modular, low-power electronics crate for reading cryogenic thermometers and controlling heaters in scientific balloons and cryostats. This paper describes the hardware, daughter cards, and claimed performance; the value is a flexible, low-cost housekeeping alternative for experimental groups.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 256-channel/7.5 W scalability and the 256-lock-in claim are not backed by full-load measurements, and the paper's own channel counts are inconsistent.","rationale":"The reader's weakest assumption correctly identifies the MCU real-time full-load operation as unpreserved, and the paper's channel-count inconsistency is also noted in the reader's rationale. I agree that the missing full-load validation is the core concern. My emphasis is slightly different: the concrete contradiction between 256 lock-in amplifiers and the 8-channel RTD card specification makes the scalability claim ambiguous, so even a successful benchmark must first define which configuration is being claimed. This does not invalidate the paper as a design description, but it strengthens the case for a conditional verdict requiring either corrected text or a full-load measurement. The noise claim 'limited only by device sensitivity' is also under-supported, as Fig. 4's 'measured performance matches or exceeds' is not accompanied by measured data or error bars, but that is a validation gap rather than an internal inconsistency. The most decisive single check is the full-chassis acquisition test with power measurement, which would settle whether the architecture can sustain the claimed channel count and power envelope.","tokens_in":8047,"tokens_out":4165,"duration_ms":44549,"concrete_test":"Populate a 16-slot chassis in the configuration that matches the intended claim (e.g., 16 RTD cards for lock-in operation, and separately 16 diode cards for maximum channel count) and run continuous acquisition for at least 24 hours at the specified sample rates. Instrument the firmware to count missed SPI frames, interrupt overruns, and clock jitter, and measure the total input power from the 12-72 V supply. If any sample is missed or total power exceeds 7.5 W, the claim fails. Also verify the lock-in channel count directly from the RTD card specification: 16 cards x 8 channels = 128, not 256.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is that a fully populated 16-card system supports 256 independent channels under 7.5 W and that the RTD readout is limited only by device sensitivity. The load-bearing condition is that a single STM32H723 microcontroller, with low-latency tasks in ISRs and hardware timer clocks, never misses samples while servicing 16 daughter cards. No throughput, latency, jitter, CPU-load, or sample-loss measurements are presented for the full configuration, so the central scalability claim is unsupported. This is compounded by an internal inconsistency in the paper's own specifications: Table I lists 8 lock-in channels per RTD card, so 16 RTD cards provide 128 lock-in amplifiers, while Section II states that a fully populated 16-card system can run 256 simultaneous lock-in amplifiers. The only way to reach 256 independent channels is with 16-channel diode or heater cards, which are not lock-in readouts. Thus the headline '256 channels' and the lock-in '256 amplifiers' are different claims, and neither is demonstrated with a measured full-chassis test. The missing support is not a matter of consensus; it is a gap in the evidence for the paper's primary performance assertions.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the design and specifications of τHK, a modular housekeeping system for cryogenic and balloon-borne instruments. The system uses an IEEE Eurocard subrack with up to 16 interchangeable daughter cards, an STM32H723 microcontroller as the host, Ethernet/Protobuf communication to an agent computer, and three existing card types: an 8-channel RTD lock-in readout, a 16-channel diode/thermistor bias-and-readout card, and load/heater and power-distribution cards. The headline claims are that a fully populated 16-card system supports up to 256 independent channels at under 7.5 W and that the RTD readout noise is limited only by device sensitivity. The paper also reports a measured noise spectral density (Fig. 5) and an empirical noise model (Fig. 4).","tokens_in":8265,"tokens_out":4370,"duration_ms":41523,"significance":"If the headline performance claims are substantiated, τHK would be a useful low-cost, modular alternative to BLASTbus and commercial cryogenic readout systems for experiments requiring hundreds of thermometers and heaters. The design choices—shared excitation with logDAC attenuation, hardware-timer clocks, standardized SPI/I2C interfaces, and software-configurable daughter cards—are sensible and reduce both cost and development time. The paper's strengths are its concrete system architecture, explicit specification table, the reported cost estimate under USD 2000, and at least one measured noise spectrum. However, the central quantitative assertions, namely 256 channels under 7.5 W, uninterrupted sample timing at full load, and RTD noise limited only by device sensitivity, rest on design assertions and an empirical model without full-chassis measurements or error bars. These are correctable deficiencies rather than fundamental flaws, but they must be addressed before the claims can be accepted.","major_comments":[{"comment":"The abstract and §II claim 256 independent channels and “256 simultaneous lock-in amplifiers” for a 16-card system, but Table I states that the RTD card has 8 lock-in channels, so 16 RTD cards provide 128 lock-in amplifiers, not 256. The 256 total can only be reached by counting 16-channel diode or heater cards, which are not lock-in readouts. The manuscript should state precisely how the 256-channel count is composed, specifying the number of each card type, and should avoid conflating total I/O channels with lock-in amplifiers.","section":"Abstract, §II, Table I"},{"comment":"The statement that “low latency tasks are run inside interrupt service routines to never miss samples” is load-bearing for the system's scalability claim, but no measurement or analysis is presented for a fully populated 16-card chassis. No CPU load, SPI throughput, interrupt latency, jitter, or sample-loss data are reported. Provide a worst-case timing budget or a full-load test demonstrating that the STM32H723 can sustain 256 channels without dropped samples at the specified update rates.","section":"§II"},{"comment":"The claim that the RTD readout is “limited only by device sensitivity” is not established by the data shown. Fig. 4 is an empirical noise model without an experimental overlay or error bars, and the text's assertion “Measured performance matches or exceeds the presented data” lacks a reference to any measurement. Fig. 5 shows frequency-dependent 1/f noise from the logDAC switches, so the abstract's unconditional wording should be qualified. Please show a measured-versus-modeled noise comparison with error bars, and state the operating conditions, such as bias power, temperature, and sensor type, under which the device-sensitivity limit is actually reached.","section":"§III A, Fig. 4"},{"comment":"The conclusion states that τHK has been used over multiple cryogenic runs and has undergone vacuum-chamber and beamline radiation testing, but no details, test conditions, or results are provided. Since suitability for scientific ballooning is an explicit design goal, include at least a summary of the radiation test, giving dose, beam type, and pass/fail criteria, and a reference to any flight or run report.","section":"§IV"}],"minor_comments":[{"comment":"The phrase “the the harsh environment” in the abstract and the introduction contains a duplicated article.","section":"Abstract and §I"},{"comment":"“Discreet solution” should read “discrete solution.”","section":"§III A"},{"comment":"The part number appears as “TMUX723414” in the text but as “TMUX7234” in reference 14; please make the two consistent.","section":"§III A and reference 14"},{"comment":"The “I/O Expander” and “TCA9548” blocks are not explained in the caption; a one-sentence description would help a reader unfamiliar with these components.","section":"Fig. 2"},{"comment":"The entry “RTD resistance range 0–5 MΩ” combined with “Bias range 500 pA–10 µA” is not sufficient to derive the maximum bias power limits; consider adding a sentence or a reference to Fig. 4 explaining how the bias limits are set.","section":"Table I"},{"comment":"The truth table columns labeled “Latch MCU Output” and “Remote Latch Set/Reset” are not entirely self-explanatory; adding a short note defining each column would improve readability.","section":"§III D, Table II"}],"recommendation":"major_revision","confidential_remarks":"This is a legitimate instrument-development paper, and the main obstacle is evidentiary rather than conceptual. I recommend major revision and would be willing to review a revised version that clarifies the channel-count claims and adds full-load or worst-case timing validation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my read of the τHK paper. It describes a real, working modular housekeeping system built around a subrack of 16 Eurocard daughter cards and a stock STM32 part. That combination — generic MCU instead of FPGA, low-cost card set, onboard calibration and ID bits — is the genuinely useful bit. The logDAC-based lock-in readout for RTDs is the most interesting circuit: it's a clever way to get wide dynamic range in a small power budget. The paper is honest about what the cards do and where they're limited, e.g., the heater card's latching over-current protection and its unsuitability for low-impedance loads. That's the kind of practical detail that makes instrument papers valuable.\n\nWhat's missing? Three things, roughly in order of size. First, the 256-lock-in claim in Section II is wrong for the cards they describe. Sixteen RTD cards at 8 channels each gives 128 lock-ins. You only get 256 channels by mixing diode, heater, and power-distribution cards, which are not lock-in readouts. The abstract's \"256 independent channels\" is defensible for a mixed chassis, but the lock-in sentence has to be corrected. This is a text error, not a design flaw.\n\nSecond, the full-load claim that a single STM32 can 'never miss samples' while servicing 16 cards is asserted, not demonstrated. There's no CPU-load or sample-loss measurement. For housekeeping rates — tens of samples per second per channel — the workload is tiny for an H723, so I'd expect it to be fine, but a one-line benchmark would settle it.\n\nThird, the noise validation is thin. The paper says \"measured performance matches or exceeds the presented data\" but doesn't show the measurement overlaid on the model or give error bars. One measured spectrum is a start, but the claim is stronger than the evidence. Again, minor.\n\nThe power budget of under 7.5 W for a full chassis is plausible from the card-level specs, but I'd like a measured number.\n\nOverall, this is a solid instrument paper for a design journal like NIM A or JINST. The system is real, it's been field-used on TIM, and the modular architecture is worth being in the literature. It deserves a serious referee. I'd send it out, with a request that the authors fix the lock-in count and add a few benchmark numbers.","headline":"A genuinely useful modular housekeeping system with real field use, but the 256-lock-in claim is internally inconsistent and the full-load and noise validation need concrete benchmarks.","tokens_in":8833,"tokens_out":2972,"would_cite":true,"duration_ms":29229,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper presents τHK, a modular Eurocard housekeeping system for cryogenic and balloon experiments that claims to support up to 256 channels under 7.5 watts.","keywords":["cryogenic temperature readout","housekeeping electronics","lock-in amplifier","RTD resistance measurement","balloon payload instrumentation","modular data acquisition","Eurocard backplane","STM32 microcontroller"],"falsifier":"Populate a full 16-card chassis with every RTD and diode channel sampling at its stated rate, log Ethernet packets and sample counters, and measure input power and lock-in phase jitter. Finding dropped samples, clock jitter that degrades the 20 Hz lock-in reference, or input power above 7.5 W under this load would refute the central capacity and timing claims.","tokens_in":7857,"feed_emoji":"❄️","tokens_out":7313,"duration_ms":71607,"temperature":0.7,"pith_summary":"τHK is a modular housekeeping system for cryogenic and balloon-borne experiments. The paper claims that one Eurocard subrack with up to 16 interchangeable daughter cards can support 256 independent channels while drawing under 7.5 W, using only an STM32 microcontroller and no FPGA. Three card types are described: an RTD lock-in readout said to be limited only by the thermometer's own sensitivity, a general-purpose diode/thermistor bias-and-readout card with chopping, and heater/power-output cards. The authors argue the design fills the gap between expensive commercial cryogenic readouts and bespoke balloon systems, and they report a year of lab use plus vacuum and radiation testing.","feed_headline":"Modular housekeeping crate runs 256 channels on 7.5 W","feed_subtitle":"RTD readout noise is claimed to hit the sensor's own floor across all temperature ranges.","key_machinery":"The mechanism that carries the design is a microcontroller-centric modular backplane: daughter cards plug into a shared Eurocard bus and communicate over SPI at 8 Mbit/s for time-critical samples and I2C for configuration, each card identified by 6 hard-wired bits. On the RTD card, a shared sine-wave bias is attenuated by a 16-step logarithmic resistor ladder (logDAC), so a single waveform drives all channels while attenuating the DAC's own noise, and the sense line is read by a fully differential lock-in amplifier using a common 24-bit delta-sigma ADC. On the controller, hardware timer peripherals generate the jitter-sensitive clocks and interrupt service routines perform low-latency tasks, which is the design choice that lets a stock STM32H723 replace an FPGA.","core_discovery":"The central claim is that large cryogenic housekeeping does not require an FPGA or a specialized commercial system. A fully populated 16-card chassis is claimed to run up to 256 simultaneous lock-in amplifiers and stay under 7.5 W, with jitter-sensitive clocks generated by hardware timer peripherals and low-latency tasks handled in interrupt service routines so samples are never missed. The RTD card is a fully differential lock-in amplifier whose input-referred noise is claimed to be limited only by device sensitivity at every temperature range; the empirical white-noise model is reported to match or be exceeded by measurements. The paper also claims that switching transients from PWM heater control are suppressed enough that RTD and diode time-streams are immune, and that the system has proven itself over a year of cryogenic runs and in representative space-like tests.","pith_inferences":["A natural consequence of the card mix is that the 256-channel headline counts independent mixed channels; an all-RTD chassis holds 128 lock-in channels, while diode and heater cards provide 16 each.","If the no-FPGA real-time claim generalizes, the same timer-and-ISR pattern could be reused for other dense synchronous measurements, such as many-channel impedance or photo-detector readout.","The noise model in Figure 4 makes a testable prediction for new sensors: the fractional noise should track bias power and resistance according to the presented curves, so a user could check device-limited performance before building a custom card."],"forward_implications":["A cryostat or balloon payload can instrument hundreds of thermometers, heaters, and switched loads from one 3U crate drawing under 7.5 W.","Experiments can add new capability by designing a daughter card against the standard SPI/I2C interface rather than a new readout stack.","If the RTD noise floor is truly device-limited, users can attribute measured temperature noise to the sensor and bias rather than to the electronics.","The use of commodity microcontrollers and common ADC parts keeps the per-channel cost low enough for mid-scale labs to assemble the system.","The reported vacuum and radiation testing supports deployment on stratospheric balloon missions such as Taurus."],"supporting_citations":[{"why":"Defines the Taurus balloon experiment requirements the system was designed to meet.","marker":"1"},{"why":"Describes the TIM experiment whose cryostat uses the pictured τHK system.","marker":"2"},{"why":"Documents the thermal challenges of a balloon-borne cryostat, the context motivating dense housekeeping.","marker":"4"},{"why":"Presents BLASTbus, the bespoke balloon readout system that τHK generalizes into a modular form.","marker":"5"},{"why":"The STM32H723 microcontroller whose hardware timers and interrupts are the load-bearing no-FPGA clocking mechanism.","marker":"11"},{"why":"The ADS131M08 ADC family shared by all daughter cards and the data path for synchronized samples.","marker":"13"},{"why":"The AD8421 instrumentation amplifier whose noise model sets the RTD readout floor.","marker":"15"},{"why":"Cryogenic RTD sensitivity data used to interpret fractional resistance noise as temperature noise.","marker":"16"}],"fun_headline_variants":["Modular housekeeping: 256 channels under 7.5 W","RTD noise floor: sensor-limited across all temperatures","Low-power portable housekeeping for balloon payloads","256 lock-in channels from a 7.5 W housekeeping crate","TauHK: modular, low-power, 256-channel housekeeping"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The scalability claim rests on the STM32H723's timer peripherals and interrupt routines keeping all clocks synchronized and never missing a sample when all 16 card slots are active, a guarantee the paper states but does not measure at full load.","fun_headline_variants_meta":{"raw":{"variants":["Modular housekeeping: 256 channels under 7.5 W","RTD noise floor: sensor-limited across all temperatures","Low-power portable housekeeping for balloon payloads","256 lock-in channels from a 7.5 W housekeeping crate","TauHK: modular, low-power, 256-channel housekeeping"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00131,"raw_usage":{"total_tokens":5345,"prompt_tokens":959,"completion_tokens":4386,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":575,"completion_tokens_details":{"reasoning_tokens":4313}},"tokens_in":575,"tokens_out":4386,"duration_ms":35882,"temperature":1.0,"reasoning_tokens":4313,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:06:15.211845+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Populate a full 16-card chassis with every RTD and diode channel sampling at its stated rate, log Ethernet packets and sample counters, and measure input power and lock-in phase jitter. Finding dropped samples, clock jitter that degrades the 20 Hz lock-in reference, or input power above 7.5 W under this load would refute the central capacity and timing claims.","supporting_citations":[{"cited_title":"Influxdb oss v2","cited_arxiv_id":null,"evidence_quote":"The STM32H723 microcontroller whose hardware timers and interrupts are the load-bearing no-FPGA clocking mechanism."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The ADS131M08 ADC family shared by all daughter cards and the data path for synchronized samples."},{"cited_title":"Purchased in 2024","cited_arxiv_id":null,"evidence_quote":"The AD8421 instrumentation amplifier whose noise model sets the RTD readout floor."},{"cited_title":"Ads131m08 8-channel, simultaneously-sampling, 24-bit, delta-sigma adc","cited_arxiv_id":null,"evidence_quote":"Cryogenic RTD sensitivity data used to interpret fractional resistance noise as temperature noise."}],"review_version":1}