IoT 3D Printer Farm Monitor with Job Queue Dashboard

College makerspaces and small print farms often run three to ten 3D printers at once, and nobody can tell at a glance which machine is free, which one failed at 2 AM, or which is about to run out of filament. The usual answer is someone walking the room, checking screens and guessing — so failed prints waste hours and queued jobs sit idle while a printer stands free. This project gives the farm a nervous system: each printer gets a small ESP32 monitor node that reads the printer's DC supply current through an ACS712 Hall-effect sensor, watches a filament-runout switch, and snaps progress photos with an ESP32-CAM. The nodes publish over MQTT to a central dashboard that renders the whole farm as a live grid — printing, idle or error — alongside a shared job queue, a filament alert feed and job history. Because status is inferred from real power draw, no printer firmware changes are needed, so any 12/24 V DC machine can join the farm.

1. The ACS712 sensor is wired in series with the printer's 12/24 V DC supply line (low-voltage side only — mains is never touched), and the monitor node runs on USB 5 V.
2. At installation, firmware captures each printer's idle standby baseline, and the idle/printing/error current thresholds adapt to that baseline.
3. The ESP32 samples current at approximately 1 kHz, computes RMS over a 1 s window and classifies the printer state; filament-switch changes fire an immediate MQTT alert message.
4. Every 5 s the node publishes a JSON packet (current, state, filament, uptime) to farm//telemetry; retained messages keep the dashboard truthful on reload.
5. An ESP32-CAM grabs a snapshot every 10 min and on every alert (stills only — no video stream, keeping farm bandwidth low) for the printer's dashboard card.
6. The single-file dashboard app connects to the broker over MQTT WebSockets and renders the farm grid, job queue and alert feed live in any browser on the local network.
7. Operators queue jobs with estimated durations; progress bars advance from elapsed time, and jobs complete or flag failed when the node reports a state change.

Tech stack:

• ESP32 dev board · ACS712 5 A current sensor
• Optical filament runout switch
• ESP32-CAM snapshot module
• MQTT · Mosquitto broker
• Dashboard: HTML/CSS/JS · MQTT WebSockets
• Arduino IDE · C/C++ firmware
• 3D-printer 12/24 V DC interfacing

• [Current-signature state detection] ACS712 5 A Hall-effect sensor on the printer's DC input; the ESP32 samples current and a firmware state machine classifies each printer as idle, printing or error from its power signature — no printer firmware changes needed.
• [Filament-runout alerts] Optical runout switch per printer wired to the node; an MQTT alert fires within about 2 seconds (design target) of filament depletion, flagging the printer as ERROR on the dashboard.
• [Live farm grid dashboard] Every printer renders as a card with status, live current draw, filament state, elapsed time and progress bar — the whole farm readable in one glance.
• [Shared job queue] Operators add jobs (G-code filename, estimated duration, priority) and assign them to printers; the queue tracks pending, running, completed and failed states with a job history log.
• [Progress camera snapshots] ESP32-CAM captures a still every 10 minutes (design target) and on every alert; the latest snapshot appears on each printer card for quick visual checks.
• [MQTT telemetry backbone] Each node publishes JSON telemetry to a local Mosquitto broker every 5 seconds on its own topic; the dashboard subscribes over WebSockets so status updates land live.
• [Alert feed] Runout, error-state and node-offline events append to a timestamped alert feed — the farm's single incident log instead of word of mouth.

• Working monitor-node prototypes — 3 complete nodes (ESP32, ACS712, runout switch, ESP32-CAM), expandable to 8
• Complete node firmware source (current sampling, state machine, MQTT, alerts)
• Farm dashboard web app (live grid, job queue, alert feed, job history)
• Wiring diagrams and DC-side installation guide with safety notes
• MQTT topic map and Mosquitto broker setup guide
• Calibration procedure for idle-baseline thresholds
• Project report PDF (background, power-signature method, firmware design, MQTT architecture)
• PPT presentation for final review
• Viva Q&A preparation document (current sensing, state machines, MQTT, embedded power budgets)

• Status is inferred from current draw, not the printer's firmware — a print paused from the screen reads as idle, and progress percentage is an elapsed-time estimate, not true layer tracking.
• The ±1.5% accuracy figure is the ACS712 datasheet typical, not a measured calibration result; the report documents the buyer-run calibration procedure instead.
• Monitoring is snapshot-based: stills every 10 min plus on alerts — there is no live video stream, by design, to keep farm bandwidth low.
• The dashboard is local-network only in this build; remote access needs a VPN or reverse proxy, listed as future scope.
• Mains voltage is never touched — the sensor sits on the low-voltage DC side only, so this node cannot detect mains-side failures.
• Stable Wi-Fi is required at the farm; nodes buffer only the last state locally and do not log long offline histories.