# PedestalCheck: Continuous Shore Power Quality Monitoring for RVs **By Jeffrey Thomas, P.E.** --- ## Project Summary Most RV owners rely on surge protectors or EPO devices to guard against shore power problems. These devices work by watching for voltage or current to cross a threshold — too high, too low, too fast — and disconnecting when that happens. What they don't tell you is everything that happens *below* the threshold. PedestalCheck is a continuous shore power monitoring system that measures both legs of a 50-amp RV shore power supply every 13 seconds, computes a suite of power quality metrics, and scores each park stay on a 0–100 scale. Over five months and 26 stays across eight states, it captured more than 700,000 individual readings — and found that the quiet, sub-threshold problems are often more consequential than the dramatic ones. The full dataset and methodology are published on Zenodo: **DOI: 10.5281/zenodo.18805474** --- ## The Problem A 50-amp RV shore power supply is a split-phase system: two 120-volt legs sharing a common neutral conductor. When the neutral connection is healthy, both legs track each other closely. When it's not — when the neutral has resistance from a worn receptacle, a loose lug, or years of oxidation — the two legs start to drift apart. A 2-volt imbalance. A 3-volt imbalance. These numbers won't trip your surge protector. But they tell a story about what's happening inside your air conditioner compressors. A compressor running at 112 volts instead of 120 volts works harder, draws more current, and runs hotter. The damage accumulates silently over days and weeks. There's no event log entry. There's just a compressor that fails two years earlier than it should have, and no way to trace it back. **The Neutral Drift Index (NDI)** is the core metric this project introduces: a continuous measure of load-dependent voltage differential that quantifies neutral path resistance without requiring a fault event to occur. --- ## Hardware - **Shelly Pro 3EM** — three-phase energy monitor measuring both 120V legs and neutral current at the shore power inlet - **Raspberry Pi** running Node-RED for data ingestion and processing - **PostgreSQL** for time-series storage and analytics - Shore power inlet tap — passive measurement, no interference with park infrastructure The system runs entirely aboard the coach, publishing data via MQTT to a local broker. No cloud dependency. --- ## Scoring Model Each stay produces seven penalty components that are subtracted from a score of 100: | Component | Metric | Threshold | Max Penalty | |-----------|--------|-----------|-------------| | pen_dvpa | Load-dependent dV/A | > 0.12 Ω | 40 pts | | pen_dv_p90 | Voltage delta p90 | > 3.0 V | 30 pts | | pen_peakdv | Peak voltage delta | > 8.0 V | 20 pts | | pen_rn | Neutral resistance p90 | > 0.35 Ω | 25 pts | | pen_vnoise | Voltage sum stddev | > 1.6 V | 10 pts | | pen_corr | Correlation(dV, I) | < 0.45 | 5 pts | | pen_lowv | Low leg p10 | < 114 V | 15 pts | **Grading:** GREEN ≥ 85 | YELLOW ≥ 70 | RED < 70 The dominant failure mode in RED-graded parks is a co-firing pattern across pen_dvpa, pen_peakdv, and pen_rn — diagnostic of elevated neutral impedance. The voltage differential increases proportionally with load current, consistent with Ohm's Law applied to a resistive neutral path. --- ## Key Findings ### Same Park, Completely Different Power The single most striking result: two sites at the same campground, visited five weeks apart, produced scores of **30.3 (RED)** and **91.4 (GREEN)**. | Metric | Site E32 | Site J12 | |--------|----------|----------| | Score | 30.3 | 91.4 | | Grade | RED | GREEN | | dv_p90 (V) | 8.43 | 3.20 | | rn_est_p90 (Ω) | 0.777 | 0.394 | | Samples | 116,505 | 31,756 | Field observation confirmed that Site J12 was served by recently installed infrastructure — pole-mounted transformers, multiple parallel weatherhead conductors, short pedestal runs. Site E32 was on older overhead distribution with no confirmed recent upgrades. The neutral resistance difference of 4.1× is consistent with this infrastructure disparity. **Practical implication:** If a park gives you a choice of sites, the electrical quality may vary dramatically. An owner with monitoring capability can assess their pedestal within minutes of plugging in and request a different site if what they see is marginal. ### The Fault That Built in Plain Sight On May 19, the monitoring system captured a complete neutral failure event — every 60-second reading in the 40 minutes leading up to it, through the trip, and through the recovery. The data shows: - A measurable leg imbalance **before any AC load was applied** — the problem was in the connection before any current flowed - Gradual degradation over 40 minutes as AC loads cycled - A single 60-second interval where leg 1 dropped from 116.5V to 103V and current spiked from 23A to 87A - Post-reset restoration of balanced voltages — followed by return of the same 2.5V imbalance by the next morning The failure didn't fix the problem. It just interrupted it briefly. ### Geographic Pattern All RED and YELLOW observations occur in Texas and New Mexico. Arizona and California parks score uniformly GREEN. The sample size is insufficient for strong regional claims, but the pattern is consistent across the full dataset. | Region | Stays | RED | YELLOW | GREEN | |--------|-------|-----|--------|-------| | Texas | 8 | 2 | 0 | 6 | | New Mexico | 2 | 0 | 1 | 1 | | Arizona | 4 | 0 | 0 | 4 | | California | 1 | 0 | 0 | 1 | | Missouri | 1 | 0 | 0 | 1 | | Indiana | 1 | 0 | 0 | 1 | | Michigan | 1 | 1 | 0 | 0 | ### Undervoltage Is Not the Primary Risk The pen_lowv penalty (triggered below 114V) fired in zero observations across all 26 stays. The primary power quality risk in this dataset is neutral impedance, not supply voltage deficiency. --- ## What's Next - **PowerMonitor V2** — adding a third ADS1115 for active conductor impedance measurement (xMI) across L1, L2, neutral, and ground. Signal injection via ESP32-S3 DAC at programmable frequency. - **Dataset expansion** — ongoing data collection as travel continues. Summer 2026 Sandusky, MI home base stay. Return visits to previously scored parks for seasonal comparison. - **Zenodo v5 release** — consolidated dataset update following completion of the current travel segment. --- ## Links - **Zenodo publication:** https://doi.org/10.5281/zenodo.18805474 - **PowerMonitor hardware project:** *(coming soon)* --- ## Components - Shelly Pro 3EM energy monitor - Raspberry Pi 4 (Node-RED host) - PostgreSQL 16 - Mosquitto MQTT broker - Node-RED - Custom PostgreSQL views and scoring model (schema published on Zenodo) --- *Jeffrey Thomas is a licensed Professional Engineer (Electrical) registered in four states, with a B.S.E.E. from Texas A&M University and an M.S. in Power Systems from the University of Kansas. He lives full-time with his wife Sharon in a 2022 Thor Tuscany 45' diesel pusher motorcoach.*