Bloomfield Innovations · Mesh, Gateways & Sensor ConnectivityBloomfield Innovations · Malla, Gateways y Conectividad de Sensores

HeadlineResumen

Fleet Inventory — Radios vs SensorsInventario — Radios vs Sensores

Connectivity Timeline — the New Internet Service GatewayCronología de Conectividad

• 2026-04-18 21:49 UTC
  Repeater H3400002795 goes silent. Last packet. It is the only device assigned to room 10 section 2z. Still silent 86 days later.Último paquete.
• 2026-04-24 16:42–16:44 UTC
  Four repeaters drop together. H3400002802 stops at 16:42; H3400002791, H3400002799 and H3400002800 stop at 16:44. All three of the latter were routing through H3400002802. One parent repeater died and took its three children with it. Still silent 80 days later.Un repetidor padre murió y arrastró a sus tres hijos.
• 2026-06-29 → 2026-07-13
  Eight rooms hold 100% data coverage. 600 (Room 1), 600 (Room 2), Greenhouse #1, Greenhouse #2, Greenhouse #3, Jims Veggies (#1), Jims Veggies (#2) and Jims Room #3 return an unbroken 14-day substrate series (~2,240 raw points per room). Whatever happened to the internet service, these rooms never stopped reporting.Ocho salas mantienen 100% de cobertura durante 14 días.
• 2026-07-11 13:00 UTC
  1Z and Jims Room #2 start reporting. Both rooms return zero data for 2026-06-29 → 2026-07-11 and then a continuous series from 13:00 UTC on 2026-07-11 through now. Verified by direct re-query of three separate earlier windows — this is a real onset, not a pull artifact. This is the strongest candidate for “the moment the new gateway came online”, but it could equally be a sensor/zone re-assignment made that day. Question for Nipat: what date did the new internet service gateway go in?Ambas salas empiezan a reportar. Pregunta para Nipat: ¿qué día se instaló el nuevo gateway?
• 2026-07-13 05:47 UTC
  Repeaters H3400002789 and H3400002790 drop. Both were routing through gateway H2200001018. They went quiet ~7.5 hours before this report and have not returned. Battery reads 100%% and last link quality was 100, so this is not a battery or signal failure — it looks like a mesh/route event. These are the two most recent, most recoverable losses. Worth a physical look while onsite.Dos repetidores caen. Batería 100%, no es fallo de batería.
• 2026-07-13 13:14–13:17 UTC
  Current state. All 3 gateways, all 24 node radios and all 48 substrate sensors reported within the last 5 minutes.Los 3 gateways, 24 radios y 48 sensores reportaron en los últimos 5 minutos.

Gateways & SinksGateways y Sinks

RepeatersRepetidores

Node Radios & Substrate Sensors by RoomRadios y Sensores por Sala

The 7 Dark Rooms — Configuration, Not NetworkLas 7 Salas Sin Datos

OpenSprinkler — Why It Cannot ReconnectOpenSprinkler

The subnet is the thing to check firstLo primero: la subred

OpenSprinkler is a LAN device. AROYA reaches it over the local network, which means the OpenSprinkler and the AROYA gateway have to be on the same subnet. Bloomfield's three AROYA gateways are currently on 192.168.1.5, 192.168.1.27 and 192.168.1.233 — so the new internet service gateway is handing out 192.168.1.x.OpenSprinkler es un dispositivo LAN. AROYA lo alcanza por la red local, por lo que deben estar en la misma subred. Los tres gateways AROYA están en 192.168.1.x.

If the OpenSprinkler was given a static IP on the old router's subnet (a 192.168.0.x, 10.0.0.x or similar), it is now sitting on a network that no longer exists and nothing can route to it — which is exactly the symptom Nipat described. The fastest checks, in order:Si el OpenSprinkler tiene IP estática en la subred del router antiguo, ya no es alcanzable.

1. Read the IP off the OpenSprinkler's own LCD / app. Does the third octet read 1 (i.e. 192.168.1.x)?Leer la IP en la pantalla del OpenSprinkler.
2. If not, either set it to DHCP and let the new gateway assign it, or give it a static address inside 192.168.1.x that is outside the new router's DHCP pool.Ponerlo en DHCP o darle una IP estática dentro de 192.168.1.x.
3. Confirm the OpenSprinkler is on the same physical network segment / VLAN as the AROYA gateways (a new ISP gateway sometimes brings a guest network or a second SSID with client isolation).Confirmar que está en el mismo segmento/VLAN.
4. Once it answers on 192.168.1.x, re-add it in AROYA and confirm irrigationSupportStatus populates and drip/drain series start arriving.Volver a añadirlo en AROYA.

Firmware — WP4 → WP5Firmware — WP4 → WP5

All three gateways already report ggVersion = 2, i.e. WP5. The H340 repeaters and H344 nodes are new-generation models that move to WP5 automatically with the gateway push — the update is sent over the air by the gateway to every device that is online at the scheduled time, with a go/no-go of online + battery ≥ 30% + signal ≥ 50%.Los tres gateways ya reportan WP5. La actualización se envía por aire desde el gateway.

Every node radio at Bloomfield is online at 100% battery and link quality 92–100, so all 24 clear the go/no-go comfortably. The six offline repeaters will simply not receive WP5 — they stay on WP4, unharmed. Nothing is lost and nothing is damaged by missing a push; the constraint is WP4/WP5 compatibility, not device health. If they are brought back online later, a gateway can be reverted to WP4 and the push re-run — there is no time limit on that. The two repeaters that dropped this morning are the realistic ones to recover before any scheduled push; the five that have been silent since April are better handled as a retire / trade-in conversation.Los seis repetidores fuera de línea simplemente no recibirán WP5 — se quedan en WP4, sin daño.

Data Quality & What I Could Not VerifyCalidad de Datos

• No 30-day per-device gap chart. AROYA's public_api /devices/<id>/chart/ caps at a 2-day window and only retains roughly the last 2 days of radio telemetry. A control test on a demonstrably healthy node returned zero points for 2026-06-20 → 06-22. The timeline above is therefore built from lastCommunication plus room-level substrate coverage, both of which are reliable.No hay gráfico de 30 días por dispositivo: la telemetría solo retiene ~2 días.
• Causation of the 2026-07-11 onset is unproven. 1Z and Jims Room #2 genuinely produced no data before 2026-07-11 13:00 UTC (verified across three independent earlier windows). Whether that is the new gateway coming online or a sensor/zone re-assignment made the same day cannot be distinguished from the API. Confirm the swap date with Nipat.La causa del inicio del 2026-07-11 no está probada.
• 1Z reports data but has zero devices assigned to it in the device registry, and its reporting zones (KeeKee, Blue Dream, KeeKee 2) are not the zones any device is currently mapped to. Something in the zone/device mapping for that room is inconsistent. Flagging rather than guessing.1Z reporta datos pero no tiene dispositivos asignados.
• Wirepas policy is not exposed by the API. The portal shows mesh state only, never policy. Any statement about routing policy here is inferred from observed next-hop behaviour, not read from the network manager.La política Wirepas no está expuesta en la API.
• Retired devices could not be enumerated. The portal_api retired-device endpoint needs a portal token, not the public API key; it was not available in this run. All 59 devices visible via the SPA are retired = false.No se pudieron enumerar los dispositivos retirados.
• Cold start. There is no Bloomfield Innovations wiki, no prior AROYA audit and no recorded commitments. Nothing in this report is graded against a previous promise, because there is no previous promise on file. Contact detail beyond Nipat Bussarakonsirivit was not invented.Inicio en frío. No hay wiki ni compromisos previos registrados.