• Status: Draft POC/MVP requirements
• Device: Universal industrial data-acquisition and predictive-maintenance gateway.
• Purpose: Connect to diverse retrofit sensors, collect and locally retain equipment data, upload it to an online platform, and support baseline, anomaly-detection, and prediction workflows.
• Initial validation case: HVAC/glycol chiller, using a dedicated harness configuration rather than constraining the carrier architecture.
• Product strategy: One reusable carrier board plus swappable, labelled sensor harnesses and configurable interface modules for different equipment classes.

• Prototype and small-batch proof-of-concept deployments on HVAC/chillers, boilers, generators, transformers, gasifiers, pumps, motors, printers, and other industrial equipment.
• Non-invasive or minimally invasive retrofit monitoring where practical.
• Installed by appropriately qualified electrical, mechanical, refrigeration, or gas personnel according to the equipment involved.
• The MVP is a data and diagnostics platform; it must not replace, bypass, or duplicate machinery safety controls.

1. Prove that one common carrier can acquire useful data from multiple industrial sensor types.
2. Prove reliable Wi-Fi/cellular transmission and store-and-forward operation.
3. Prove field configuration, channel calibration, sensor-health detection, and modular harness replacement.
4. Establish a reusable data schema for different equipment classes.
5. Validate the first full harness on an HVAC/glycol chiller.
6. Generate enough healthy and labelled maintenance data to begin anomaly-detection development.

• Direct 110/240 VAC input on the custom PCB.
• Replacing certified protection, interlock, flame-safeguard, low-water, relief, or machine-control systems.
• Universal support for every sensor without configuration or interface adaptation.
• Guaranteed remaining-useful-life predictions before sufficient labelled field data exists.
• Medium/high-voltage direct measurement or energized-panel installation by unqualified personnel.

• Acquire configurable analog, temperature, current, vibration, pulse/frequency, contact, and industrial digital-bus signals.
• Timestamp and convert measurements into engineering units with calibration and quality flags.
• Buffer data locally for at least 7 days of normal summary telemetry.
• Use Wi-Fi as primary backhaul and cellular as automatic fallback.
• Continue acquisition through network outages and forward queued records after reconnection.
• Report device, power, storage, network, and per-channel sensor health.
• Support secure remote configuration and firmware updates with recovery provisions.
• Support continuous summary telemetry plus scheduled or triggered high-rate waveform bursts.

Diagram

• ESP32-S3-class controller and Wi-Fi.
• Cellular modem interface or integrated cellular module.
• Protected 24 VDC input and internal power conversion.
• RTC, non-volatile configuration memory, and removable or soldered bulk storage.
• Watchdog, brownout supervision, status indicators, and service/programming interface.
• Internal expansion buses and standardized module connectors.
• Optional isolated RS-485 transceiver where cost and board space permit.

• 24 V protected field power where required.
• 5 V and 3.3 V logic power with defined current limits.
• SPI and I2C.
• UART.
• ADC/control lines where required.
• Interrupt/synchronization line.
• Module identification or configuration EEPROM.
• Ground, shield, and optional isolated-side boundaries clearly defined.

• 0–5 V.
• 0–10 V.
• 4–20 mA.
• Low-voltage ratiometric sensors where supported by the selected ADC and excitation scheme.

• Per-channel configuration through jumpers, switches, or software-controlled front ends; the POC may use clearly documented hardware configuration.
• External ADC with resolution and noise performance suitable for process trending.
• Input scaling, anti-alias filtering, overvoltage protection, ESD protection, and fault-current limiting.
• 4–20 mA burden resistor selected for ADC range and power rating.
• Sensor open-circuit, over-range, and under-range detection where the interface permits.
• Optional isolated analog module variant reserved for field loops with grounding or long-cable concerns.
• Protected 24 V sensor excitation output with per-module current limiting where required.

• PT100/PT1000 RTDs using dedicated converter circuitry.
• Thermocouples using cold-junction-compensated interfaces; type selected per equipment and atmosphere.
• 1-Wire digital probes for moderate-temperature, cost-sensitive channels.
• I2C/SPI temperature sensors for enclosure or local board measurements.

• Open/short detection.
• 2-, 3-, or 4-wire RTD support selected by module variant; 3-wire is the preferred general industrial compromise.
• Thermocouple module must support the required type and preserve polarity and cold-junction accuracy.
• Connector and firmware configuration must prevent silently treating one probe type as another.

• Split-core current-transformer compatibility for retrofit monitoring.
• Permanently fitted burden and secondary protection where required; a CT secondary must never be left open while its primary is energized.
• Biasing and anti-alias filtering for bipolar waveform capture.
• Simultaneous or sufficiently synchronized sampling for three-phase/load comparison where feasible.
• Support RMS, peak, crest factor, load trend, and scheduled high-rate waveform capture.
• CT ranges are selected in the harness/module configuration, not assumed universal.
• Inline Hall-effect or shunt-based measurement is outside the default retrofit path but may be supported by future modules.

• SPI-connected industrial vibration sensor preferred for predictable bandwidth and lower analog-front-end risk.
• Rigid sensor mounting and shielded/robust cabling appropriate to the installation.
• Configurable sampling and burst duration.
• Local calculation of RMS, peak, crest factor, and selected spectral features to reduce cellular data use.
• Preserve selected raw waveform bursts for server-side algorithm development.
• Future module options may include IEPE accelerometers, analog MEMS sensors, and ultrasonic microphones.

• Dry-contact or open-collector state inputs.
• Pulse/frequency counting for RPM, flow, or energy-meter pulses.
• Debounced alarm/run-state inputs.
• 1-Wire sensors where not placed on the temperature module.

• Input protection suitable for external cabling.
• Configurable pull-up/excitation and thresholding.
• Galvanically isolated variant for 12/24 V industrial status signals is preferred for field deployment.
• Outputs for control are not part of the initial MVP; the unit is monitoring-only.

• Isolated RS-485 supporting Modbus RTU is the first-priority industrial bus.
• CAN/CAN-FD or J1939 interface should be available through a future module or carrier option.
• Ethernet and BACnet/IP are future options, not required for the first hardware spin unless site requirements justify them.
• Existing controller data should be read where authorized rather than duplicating sensors unnecessarily.

• Equipment class and harness revision.
• Connector and cable pinout.
• Sensor type, measurement range, excitation, output format, and calibration.
• Module type and per-channel configuration.
• Cable type, shield termination, maximum recommended length, and environmental rating.
• Safe installation method and required installer qualification.
• Firmware channel name, units, sample rate, filtering, alarm limits, and upload policy.

• 3 temperature channels.
• 2 refrigerant pressure channels.
• 3 split-core CT current channels.
• 1 vibration channel.

• Glycol supply temperature.
• Glycol return temperature.
• Compressor discharge, suction-line, condenser, or ambient temperature.
• Refrigerant suction pressure.
• Refrigerant discharge pressure.
• Compressor current.
• Condenser/cooling-tower fan current.
• Glycol/circulation-pump current.
• Compressor or priority rotating-equipment vibration.

• ESP32-S3 module or equivalent, selected for adequate interfaces, security features, memory, Wi-Fi support, and ecosystem maturity.
• Cellular modem must be compatible with target-region carriers; LTE Cat-M1/NB-IoT is preferred where coverage is verified, with LTE Cat-1 bis considered where it provides better regional availability.
• Independent hardware watchdog preferred.
• Hardware should expose commissioning and recovery interfaces without opening permanent security weaknesses.

• At least 7 days of normal summary telemetry at the configured channel count.
• Storage policy must distinguish continuous summaries from high-rate waveform bursts.
• RTC with backup retention and network time synchronization.
• Each record includes device, site, equipment, harness revision, channel, timestamp, value, units, quality flags, and configuration revision.
• Use power-loss-resistant writes and storage-health monitoring.

• Primary: 2.4 GHz Wi-Fi.
• Fallback: Region-compatible cellular.
• Automatic failover and return without stopping acquisition.
• Queue-and-forward with acknowledgement and duplicate protection.
• MQTT over TLS or HTTPS with authenticated devices.
• Local commissioning method for network and sensor configuration.
• Antenna connections must allow external antennas when the unit is installed inside a metal cabinet.

• Device input: protected nominal 24 VDC.
• Expected input operating range: to be finalized after transient/protection design; the design should accommodate normal industrial 24 V supply tolerance.
• Existing-panel installations: connect to a verified 24 VDC supply with adequate spare capacity.
• Mains-only installations: use an external certified isolated DIN-rail power supply.
• External supply input: universal 85–264 VAC, 50/60 Hz.
• External supply output: regulated 24 VDC.
• Initial supply class: 25 W minimum planning target; final rating follows the complete datasheet-based power budget.
• No 110/240 VAC is routed onto the custom PCB.
• Battery backup is optional and outside the first hardware baseline; storage protects against communication outages, not extended site-power loss.

Diagram

• The earlier 15–25 W supply class may be insufficient for the highest-loaded universal configuration.
• Use a 30 W or larger external supply as the current planning assumption, then size from the verified peak budget with margin.
• Use switching conversion from 24 V; do not use a linear regulator for the main 24 V-to-5 V or 24 V-to-3.3 V conversion.
• Cellular transmit peaks must not cause brownout or corrupt storage.
• Field excitation outputs require defined current limits so one faulty sensor cannot collapse the carrier rails.

• Reverse-polarity protection and input overcurrent protection.
• Surge/transient suppression appropriate to industrial 24 V panels.
• EMI filtering coordinated with converter stability and conducted-emissions needs.
• ESD, overvoltage, and fault-current protection on all external channels.
• Defined shield/chassis/logic-ground strategy.
• Isolation applied by module/interface need rather than claiming one non-isolated circuit is universal.
• Pluggable, keyed connectors and clear terminal labelling.

• Per-channel configurable sample rate, filtering, scaling, units, and upload interval.
• Typical process-summary interval: 1–10 seconds.
• High-rate current and vibration capture occurs in configurable bursts.
• Preserve raw ADC/sensor values alongside calibrated engineering units where practical.
• Detect disconnected, shorted, saturated, stale, and implausible signals.
• Provide synchronized timestamps across channels.
• Configuration must be versioned and included with uploaded data.

• Common ingestion format independent of equipment class.
• Time-series storage linked to device, site, equipment, harness, operating state, and maintenance labels.
• Dashboards for telemetry, device health, network status, storage, and alerts.
• Begin with thresholds, trend analysis, correlations, and unsupervised/semi-supervised anomaly detection.
• Supervised failure classification and remaining-useful-life prediction require sufficient labelled fault history and are not immediate MVP guarantees.
• Operators must be able to record alarms, shutdowns, inspections, replaced parts, and confirmed root causes.

• Board and harness/module identification.
• Per-channel configuration and calibration storage.
• Wi-Fi/cellular failover state machine.
• Store-and-forward queue with acknowledgement and retry control.
• Sensor-health diagnostics and acquisition scheduling.
• Local feature extraction for vibration/current waveforms.
• Secure boot and signed update support where available.
• Watchdog, brownout recovery, storage recovery, and update rollback.
• Commissioning mode showing live raw and scaled channel readings.

• Carrier and modules suitable for a serviceable prototype enclosure.
• DIN-rail or secure panel mounting.
• IP54 minimum inside a protected environment; IP65 preferred for exposed installations.
• External antennas when installed in a metal enclosure or cabinet.
• Maintain air space around internal PCB antennas and keep copper, wiring, screws, bosses, and conductive material out of the antenna region.
• Separate high-current/noisy field wiring from low-level analog, RTD, thermocouple, and vibration channels.
• Provide strain relief and field-replaceable connectors.

• The custom electronics remain on the 24 VDC SELV/PELV side.
• Mains conversion uses an external certified isolated supply installed under local electrical rules.
• Monitoring circuitry must not bypass or interfere with equipment safety systems.
• CT, transformer, generator, boiler, gasifier, pressure-vessel, refrigerant, or gas-train installations require appropriately qualified personnel.
• Sensor taps and pressure connections must use approved fittings and procedures.
• De-energized installation is preferred wherever possible.

• POC quantity: one to a small batch.
• Prioritize modularity, accessible test points, replaceable modules, and debugging over minimum size and BOM cost.
• Use readily sourced modules for cellular and specialized acquisition functions in revision 1 where that reduces integration risk.
• Carrier should be assembled professionally; field harnesses can be built and labelled separately.
• Lock module and harness connectors/pinouts before ordering a larger batch.

• Equipment class, manufacturer, model, serial number, age, duty cycle, and operating modes.
• Existing controller, BMS/PLC, alarm history, maintenance records, and known recurring failure modes.
• Measurements most valuable to operators and the action expected from each alert.
• Existing Modbus, BACnet, CAN/J1939, pulse, relay, or analog data that can be read safely.

• Available 24 VDC range, spare current, grounding, noise, transients, and panel location.
• If using mains: voltage, frequency, phase, protective earth, breaker, DIN-rail space, and applicable installation rules.
• High-current equipment nameplate ratings, conductor sizes, phase arrangement, VFDs, contactors, and CT installation clearance.
• Lockout/tagout procedures and qualified installer responsibility.

• Required measurement, expected normal/fault range, accuracy, bandwidth, cable length, and response time.
• Sensor output: 0–5 V, 0–10 V, 4–20 mA, RTD, thermocouple, pulse, dry contact, RS-485, CAN, I2C, SPI, or 1-Wire.
• Sensor excitation, connector/fitting, media compatibility, ingress rating, and calibration requirements.
• Grounding/isolation needs and cable shield termination.
• Existing ports or mounting points and whether process penetration is permitted.

• Rotational speed range, bearing locations, mounting surface, background vibration, and accessible operating conditions.
• Required fault frequencies and minimum useful bandwidth.
• Current waveform load type, inrush, VFD presence, conductor access, and desired analysis depth.

• Wi-Fi coverage, security, onboarding, firewall, captive-portal, and internet-access restrictions.
• Cellular carrier, bands, technology availability, signal level, APN, SIM procurement, and data cost.
• Antenna location and cable route.
• Server endpoint, security, data ownership, retention, and remote-access policy.

• Mounting location, enclosure size, DIN rail, cable entries, service clearance, and antenna clearance.
• Temperature, humidity, condensation, washdown, dust, chemicals, sunlight, vibration, and corrosion.
• Required IP rating, cable glands, conduit/tray, strain relief, and maximum cable runs.

• Healthy baseline period and representative operating loads.
• Maintenance schedule and opportunities to capture before/after service data.
• Method for operators to label alarms, symptoms, shutdowns, replaced parts, and root causes.
• Acceptance-test window and responsible contacts.

• Chiller model, refrigerant, compressor type/staging, VFD use, and controller/BMS interfaces.
• Glycol supply/return pipe locations, material, diameter, insulation, and temperature range.
• Refrigerant suction/discharge pressure range, service-port fittings, and qualified technician availability.
• Compressor, fan, and pump currents, conductor sizes, phase arrangement, and CT clearance.
• Preferred rigid vibration-sensor mounting point and rotational-speed range.
• Condensation, frost, refrigerant/oil exposure, and cable-routing constraints.

• Carrier operates reliably from verified 24 VDC and through Wi-Fi/cellular transmit peaks.
• External certified 85–264 VAC-to-24 VDC supply provides mains compatibility without mains entering the custom PCB.
• At least one module of each MVP type is recognized and configurable: analog, temperature, CT/current, vibration, and digital/contact.
• Demonstrate 0–5 V, 0–10 V, and 4–20 mA acquisition using calibrated test sources.
• Demonstrate at least one RTD or thermocouple path and one digital temperature path.
• Demonstrate three CT channels and one vibration channel with summary metrics and waveform bursts.
• Demonstrate dry-contact/pulse acquisition and isolated RS-485/Modbus communication.
• Store at least 7 days of representative summary telemetry.
• Demonstrate automatic Wi-Fi/cellular failover, offline buffering, and later upload without unacceptable loss or duplication.
• Detect representative open, short, over-range, and disconnected-sensor faults.
• Complete the HVAC/glycol chiller nine-channel validation harness and receive all channels at the server.
• Installation does not modify or impair existing equipment safety functions.

• The MVP is a universal industrial DAQ platform, not a site-specific chiller board.
• The architecture uses one common carrier plus swappable interface modules and equipment-specific harnesses.
• HVAC/glycol chiller remains the first full validation use case.
• Wi-Fi is primary and cellular fallback is mandatory.
• The device accepts protected nominal 24 VDC only.
• Mains compatibility uses an external certified isolated 85–264 VAC-to-24 VDC DIN-rail supply.
• Monitoring-only behavior is the MVP default; control outputs are deferred.
• Local buffering and server-side analytics are required from the first prototype.

• Revision 1 may use specialist modules/daughterboards to reduce development risk.
• A universal connector standard can be defined without placing every possible field interface directly on the carrier.
• Final isolation levels, channel counts, sampling rates, sensor ranges, and connector selections will be resolved during schematic architecture and bench-test planning.
• Cellular technology is selected after confirming target-region coverage.
• A 30 W or larger supply is the provisional planning target until the datasheet-based budget is complete.

• Refocused the project from a chiller-specific pilot to a universal industrial data-acquisition POC/MVP.
• Defined common carrier, modular interface, and equipment-specific harness boundaries.
• Added configurable analog, temperature, CT/current, vibration, digital/contact, and industrial communication interfaces.
• Retained the HVAC/glycol chiller as the first full validation harness.
• Preserved the protected 24 VDC input and external certified mains-supply architecture.
• Expanded the power envelope, universal site-survey checklist, and acceptance criteria.