HellzGate ESP-NOW Cluster Firmware · Open-Source BetaExplore the beta
Development update · October 11, 2026

A working foundation. More to prove.

  • Hardware validationNine-scanner bench and field logging completed on both transports. Power-input retesting and production-board qualification remain in progress. Completed logging tests do not establish production readiness.
  • Open-source beta updatedReleased October 2. The October 5 source update adds Wi-Fi-only / Wi-Fi + BLE controls and reliability improvements. Host checks passed; validation of the updated source on hardware remains pending.
  • Observation foundation testedWi-Fi/BLE observation, record integrity, counting, and deduplication tested on real hardware.
  • Local bus demonstratedAll nine FullGate scanners stayed online during a ten-minute wired soak, with zero I²C read errors.
Field logging · One master + nine scanners · September 24–29

Four runs. Two transports. Saved files checked.

423,592saved observations across four runs
2 transportswired I²C and ESP-NOW
4 CSVsstructure and timestamps checked

Wired I²C · September 24: 174,166 observations over 38m05s and 47,446 over 19m38s — 221,612 total.

ESP-NOW · September 29: 105,564 observations over 23m09s and 96,416 over 26m30s on different routes — 201,980 total.

Each file contains BLE plus 2.4 GHz and 5 GHz Wi-Fi observations, with 14 columns per record, populated timestamps and a complete final newline. The first ESP-NOW run's saved count exactly matches its stopped-session display.

GPS startup, bench recovery and what these results establish

Initial rows have zero coordinates: 2,069 and 339 in the I²C files; 1,043 and 503 in the ESP-NOW files. Later records have nonzero coordinates. Those startup placeholders are not valid locations. Raw logs, addresses and route coordinates remain private.

A separate September 29 ESP-NOW bench run saved 34,780 observations over 17m04s, matching its logged close. A controlled scanner reset showed nine-to-eight-to-nine recovery while the other scanners continued delivering records; the phone dashboard operated alongside logging.

Counts are observations, not unique devices. Durations span the first and last saved records, not battery endurance. Different routes and firmware builds prevent treating these totals as a transport benchmark. CSVs alone do not prove zero radio loss or uninterrupted node uptime; no independent Stop count was available for the second ESP-NOW field run.

Earlier bench results · One master + nine scanners · September 13

Two transports. Real SD files. Successful uploads.

9/9scanners online in both soaks
10 minseparate test for each transport
0reported SD write errors

ESP-NOW: 22,645 additional rows reported written, with zero reported lost frames, overflows or restarts. One duplicate frame was detected and discarded.

Wired I²C: 20,550 additional rows reported written across 105,867 additional polls, with zero read errors or wrong IDs. Initial overflow counts did not increase.

Both runs collected BLE and 2.4/5 GHz Wi-Fi observations and held a GNSS fix throughout. Both retrieved SD files had metadata and column headers, and every record had the expected 14 columns.

Saved files, upload results and remaining checks

The ESP-NOW file contained 47,273 records, all accepted by the tested upload service. The I²C file contained 29,102 records; an upload copy removed 63 placeholder timestamps, and all remaining 29,039 records were accepted. File totals include recording outside the timed soak windows. This verifies the tested upload path, not every service.

These September 13 results are retained as history. Field results are above; battery endurance and charging validation remain separate checks. Neither bench nor field runs alone establish production readiness.

Historical ESP-NOW bench test

20 scanners online for approximately 10 minutes.

20/20scanners stayed UP
23,580observation records
≈39/seccombined record average

This earlier test used a separate one-master, 20-scanner setup. Validation of 20 scanners on the current firmware is still pending. FullGate has nine physical scanner slots. The rate is a measured bench average, not unique devices, a per-scanner rate, or a maximum-throughput rating.

Test counters and limitations

The timed window recorded 13,782 frames, no new reported lost frames, overflow, dropouts, or restarts, one new duplicate indication, and zero bad-CRC, bad-field, or table-full errors.

These are timed-window deltas, not cumulative startup counters. All scanners had antennas, but antenna types and observation loads differed. This does not establish maximum RF capacity or complete field performance.

Next Steps

Field results in. Next checks ahead.

  1. Finish standalone firmware testingComplete targeted startup, buffered-write/overflow and SD-save recovery checks before firmware freeze. OLED operation and refresh have been observed; startup reliability remains under review.
  2. Extend full-array testingValidate 20 scanners enrolled to one ESP-NOW master on the current firmware, building on the completed nine-scanner bench and field runs.
  3. Qualify power and production hardwareFinish power-input retesting, measure battery endurance and qualify the production board. Field logging spans do not establish full-charge runtime.
  4. Complete the release preparationVerify enclosure fit and cooling, finalize the app interface, then complete device integration and release-readiness checks.

The app foundation exists; real-device integration and release preparation remain in development. FullGate is intended to remain useful without the app.