The short answer to "which units can we transfer": far fewer than the inventory suggests — and almost none of them are sensors.
The AROYA roster shows 133 device rows at facility 2721, and 31 of them are bound to no room at all, which looks like a ready-made transfer pool. It isn't. When you stop trusting the roster and check what each device has actually emitted, 23 of those 31 unassigned devices have produced no telemetry and no zone data for the entire 14-day window. They are not spares sitting in a box waiting to be redeployed — they are dead rows in the database. Only 6 of the 31 (all mesh nodes) are alive.
Add the three orphan repeaters — one each stranded in Mother Room, Dry Room and Sj Room 2, rooms that hold a radio and nothing else — and the genuinely free, alive, movable hardware comes to 9 devices: 9 radios and 0 sensors. There is not one live, unattached probe in the fleet. Every working TEROS-12 and ATMOS-14 at this site is currently feeding a room that is actively being irrigated.
Separately, 18 devices sit in the three SJ rooms (SJ Room 1, SJ Nursery, SJ Room X). They are alive, they are the new generation (TEROS One / Climate One / EE894), and they have delivered data every single day of the window — but their substrate reads 0.4–9.0% VWC with pore EC at 0.0 mS and zero irrigation shots in 14 days. That is hardware watching an empty room. Whether that is a site being stood up or a site being wound down is the one thing the telemetry cannot tell us, and it changes the answer completely — this is the question to put to Derek on the call.
Finally, the generation split: 69 of 133 devices are old-gen (TEROS-12, ATMOS-14, H11x repeaters) and 41 of them are load-bearing in live grow rooms right now. Only 12 sensors at the whole site are current generation, and all of them are in the SJ rooms. If Sipkoi consolidates sites, the choice is not "move the old kit" vs "leave it" — it is "move it" vs "trade it in." That decision is laid out in §6; it is not made for you here.
Respuesta corta: se pueden transferir muchos menos equipos de los que sugiere el inventario, y casi ninguno es un sensor. De los 31 dispositivos sin asignar, 23 están muertos (sin telemetría ni datos de zona en 14 días). Solo 6 están vivos. Sumando 3 repetidores huérfanos, el hardware realmente libre y vivo son 9 dispositivos: 9 radios y 0 sensores. No hay ni una sonda viva sin asignar. Otros 18 dispositivos están en las salas SJ: vivos, nueva generación, pero con 0.4–9.0% VWC, CE de poro 0.0 mS y cero riegos en 14 días — hardware observando una sala vacía. 69 de 133 son de generación antigua.
Every device in exactly one bucket. Alive means it emitted measured telemetry in the last 48 h, or its zone delivered data in the last 48 h — not that AROYA's roster says online: true. Radios and sensors counted separately throughout.Cada dispositivo en un solo grupo. Vivo = emitió telemetría medida en 48 h. Radios y sensores contados por separado.
| Bucket | Radios | Sensors | Total | What it means | Action |
|---|---|---|---|---|---|
| IN SERVICE | 34 | 31 | 65 | Alive and delivering into one of the 7 Moreno Valley grow rooms that show an active crop signal (real shots, real dryback, real EC).Vivo y entregando datos en una de las 7 salas activas. | Do not move.No mover. |
| TRANSFERABLE — free now | 9 | 0 | 9 | Alive, bound to nothing or stranded in a stripped room. This is the only hardware that is unambiguously free.Vivo, sin vínculo o en sala vacía. Único hardware libre sin ambigüedad. | Move — but note: all radios, no probes.Mover — todos radios, sin sondas. |
| TRANSFERABLE — SJ, idle | 8 | 10 | 18 | Alive, new-gen, 100% delivery — but the rooms show no crop: 0.4–9.0% VWC, 0.0 mS pore EC, 0 shots in 14 d.Vivo, nueva gen, pero sin cultivo: 0.4–9.0% VWC, 0.0 mS, 0 riegos. | Derek's call — see §5.Decisión de Derek — §5. |
| DEAD / RETIRE | 15 | 24 | 39 | No device telemetry in 48 h and no zone data across the full 14 d. Nothing to transfer — there is nothing there.Sin telemetría en 48 h y sin datos de zona en 14 d. | Retire / trade-in / RMA.Retirar / canje / RMA. |
| UNVERIFIABLE | 2 | 0 | 2 | The 2 network sinks. They emit no per-device series and are bound to no room, so neither liveness nor death can be proven from the API.Los 2 sinks. Sin serie por dispositivo ni sala — no se puede probar nada. | Check physically on site.Verificar en sitio. |
| TOTALTOTAL | 68 | 65 | 133 | Radios = mesh (node · repeater · gateway · sink). Sensors = probes (TEROS-12 · TEROS One · ATMOS-14 · Climate One · EE894). No serial in this fleet carries a -2 suffix, so there is no second-probe-port double-count.Radios = malla. Sensores = sondas. Sin sufijos -2 — no hay doble conteo. | |
31 devices (23% of the fleet) are bound to no room. On paper this is the transfer pool. In reality 25 of the 31 are dead — no per-device telemetry in 48 h and no zone data at any point in the 14-day window. Only 6 are alive, and every one of them is a mesh node. Serial numbers shown for radios (H-prefix); sensor-head serials masked.31 dispositivos sin sala. En papel, el grupo de transferencia. En realidad 25 de los 31 están muertos. Solo 6 están vivos, todos nodos de malla.
| Serial | Model | Class | Gen | Measured status | Bucket |
|---|---|---|---|---|---|
| H3110010927 | node | radio | mesh | dead | DEAD / RETIRE |
| H3110010947 | node | radio | mesh | dead | DEAD / RETIRE |
| H3110010973 | node | radio | mesh | dead | DEAD / RETIRE |
| H3210010834 | node | radio | mesh | alive | TRANSFERABLE |
| H3210010835 | node | radio | mesh | alive | TRANSFERABLE |
| H3210010839 | node | radio | mesh | dead | DEAD / RETIRE |
| H3210010841 | node | radio | mesh | alive | TRANSFERABLE |
| H3210010851 | node | radio | mesh | alive | TRANSFERABLE |
| H3210010857 | node | radio | mesh | dead | DEAD / RETIRE |
| H3210010860 | node | radio | mesh | alive | TRANSFERABLE |
| H3210010861 | node | radio | mesh | dead | DEAD / RETIRE |
| H3210010866 | node | radio | mesh | dead | DEAD / RETIRE |
| H3210010909 | node | radio | mesh | alive | TRANSFERABLE |
| H1100002872 | repeater | radio | mesh | dead | DEAD / RETIRE |
| H1110006109 | repeater | radio | mesh | dead | DEAD / RETIRE |
| H1110006137 | repeater | radio | mesh | dead | DEAD / RETIRE |
| SINKH2100002197 | sink | radio | mesh | unverifiable | UNVERIFIABLE |
| SINKH2110001130 | sink | radio | mesh | unverifiable | UNVERIFIABLE |
| —— masked —— | atm14 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | atm14 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
| —— masked —— | ter12 | sensor | old-gen | dead | DEAD / RETIRE |
The 11 unassigned TEROS-12 probes and 2 unassigned ATMOS-14 climate sensors have emitted nothing — not a battery reading, not a signal reading, not a data point — for the entire window. Whatever their physical state (in a drawer, unpowered, or genuinely failed), they cannot be counted as deployable inventory today. The 7 dead nodes and 3 dead repeaters are the same story. Do not build a transfer plan on top of these 23 rows. The 6 live nodes are real and can move. The 2 sinks need a physical eyes-on check — the API cannot see them.Las 11 sondas TEROS-12 y 2 ATMOS-14 sin asignar no han emitido nada en toda la ventana. No cuentan como inventario desplegable. No construya un plan de transferencia sobre estas 23 filas. Los 6 nodos vivos sí pueden moverse. Los 2 sinks requieren verificación física.
Crop-active = the room's substrate shows real irrigation shots and real dryback over the 14-day window. Those 7 rooms are off-limits for transfer. The 4 dark Moreno Valley rooms (Pump, Mother, Nursery, Dry) return zero series of any kind — they hold a stray radio or a dead climate sensor and nothing else.Cultivo activo = riegos y dryback reales en 14 días. Esas 7 salas no se tocan. Las 4 salas oscuras no devuelven ninguna serie.
| Room | Crop / data status | Radios | Sensors | Alive | Dead | Generation | Models |
|---|---|---|---|---|---|---|---|
| Grow #1 | ACTIVE GROW | 4 | 5 | 8 | 1 | old-gen | 1× atm14 · 3× node · 1× repeater · 4× ter12 |
| Grow #2 | ACTIVE GROW | 5 | 5 | 10 | 0 | old-gen | 1× atm14 · 4× node · 1× repeater · 4× ter12 |
| Grow #3 | ACTIVE GROW | 6 | 5 | 11 | 0 | old-gen | 2× atm14 · 3× node · 3× repeater · 3× ter12 |
| Grow #4 | ACTIVE GROW | 6 | 7 | 12 | 1 | old-gen | 1× atm14 · 5× node · 1× repeater · 6× ter12 |
| Grow #5 | ACTIVE GROW | 6 | 6 | 11 | 1 | old-gen | 1× atm14 · 4× node · 2× repeater · 5× ter12 |
| Grow #6 | ACTIVE GROW | 4 | 4 | 8 | 0 | old-gen | 1× atm14 · 3× node · 1× repeater · 3× ter12 |
| Room X | ACTIVE GROW | 3 | 2 | 5 | 0 | old-gen | 1× atm14 · 1× gateway · 1× node · 1× repeater · 1× ter12 |
| Pump Room | DARK | 0 | 0 | 0 | 0 | mesh only | — |
| Mother Room | DARK | 1 | 0 | 1 | 0 | mesh only | 1× repeater |
| Nursery | DARK | 0 | 1 | 0 | 1 | old-gen | 1× atm14 |
| Dry Room | DARK | 1 | 0 | 1 | 0 | mesh only | 1× repeater |
| SJ Room 1 | SENSED, NO CROP SIGNAL | 4 | 5 | 9 | 0 | new-gen | 1× ee894 · 1× gateway · 1× h421 · 3× node · 3× ter15 |
| SJ Nursery | SENSED, NO CROP SIGNAL | 3 | 4 | 7 | 0 | new-gen | 1× ee894 · 1× h421 · 2× node · 1× repeater · 2× ter15 |
| SJ Room X | SENSED, NO CROP SIGNAL | 1 | 3 | 2 | 2 | new-gen | 1× ee894 · 1× h421 · 1× node · 1× ter15 |
| Sj Room 2 | DARK | 6 | 5 | 1 | 10 | old-gen | 1× atm14 · 4× node · 2× repeater · 4× ter12 |
| Unassigned | NO ROOM BINDING | 18 | 13 | 6 | 23 | old-gen + mesh | 2× atm14 · 13× node · 3× repeater · 2× sink · 11× ter12 |
AROYA's per-device telemetry endpoint only retains 48 hours. To see further back we read the zone series instead: every device row carries a room and a zone, so a hole in a zone's series is that device failing to deliver. This is the same technique used for the GTI Danville longitudinal.El endpoint por dispositivo solo retiene 48 horas. Para ver más atrás usamos las series de zona: un hueco en la zona es el dispositivo sin entregar.
Not a single device went dark during the window. All 28 delivering zones ran at 14/14 days and full ~9-minute cadence — roughly 160 points per day, every day, with no dropouts. And the 39 dead devices delivered nothing on day 1 and nothing on day 14. There is no decay curve here, no gradual mesh degradation, no batteries slowly fading. The fleet is cleanly split into hardware that works perfectly and hardware that was already gone before this window opened.
What that means for the transfer decision: the dead units did not die recently, so there is no "bring them back online" play — they have been off long enough that they pre-date any measurement we can reach. And the live units are rock-solid, so there is no reliability argument for replacing them. The split is stable and you can plan against it.
Ningún dispositivo se apagó durante la ventana. Las 28 zonas activas entregaron 14/14 días a cadencia completa (~160 puntos/día). Y los 39 muertos no entregaron nada ni el día 1 ni el día 14. No hay curva de degradación: la flota está limpiamente dividida entre hardware perfecto y hardware que ya estaba ausente.
Facility 2721 is named R&D Moreno Valley, but four of its fifteen rooms carry an SJ prefix. Three of them (SJ Room 1, SJ Nursery, SJ Room X) hold all 12 of the site's current-generation sensors — TEROS One, Climate One and EE894 — and every one of them has delivered data on all 14 days. The hardware is healthy. The rooms are empty.Cuatro de las quince salas llevan prefijo SJ. Tres contienen todos los 12 sensores de generación actual. El hardware está sano. Las salas están vacías.
| Room | Devices | VWC range | Pore EC | Shots/day (14 d) | Air temp °F | VPD kPa | Crop signal | |
|---|---|---|---|---|---|---|---|---|
| SJ Room 1 | 9 | 2.3–2.4% | 0.0 mS | 0.0 | 80.54 | 2.15 | no crop | |
| SJ Nursery | 7 | 8.3–9.0% | 0.08 mS | 0.0 | 79.9 | 1.94 | no crop | |
| SJ Room X | 4 | 0.4–1.1% | 0.0 mS | 0.0 | no climate series | — | no crop | |
| Sj Room 2 | 11 | No series of any kind. 10 of its 11 devices are dead — and they are old-gen (4× node, 4× TEROS-12, 1× ATMOS-14, 1× repeater), not the new kit the other SJ rooms have.Ninguna serie. 10 de 11 dispositivos muertos, y son de generación antigua. | dark | |||||
Says: substrate VWC of 0.4–9.0% with pore EC pinned at 0.0–0.08 mS, no P1 ramp-up, no P2 maintenance pulses and a flat P3 overnight line (dryback ≤ 0.45 absolute pp) across 14 days is not an under-watered crop — it is a probe sitting in dry, unfertigated media or open air. Lights-on RH is 40–44% with VPD 1.9–2.2 kPa: a dry, uncropped room. SJ Room X has additionally lost both its climate sensors (its Climate One and its EE894 return no air-temp and no CO₂ series).
Can't say: whether SJ is a new site coming online — probes commissioned and reporting ahead of the first crop — or an old site being emptied. The delivery history is identically flat across all 14 days, so there is no ramp-up or wind-down signature to read. Both stories fit the data equally well, and they point to opposite decisions.
The ask on the call: if SJ is being stood up, these 18 devices are already exactly where they belong and nothing should move. If SJ is being wound down, this is the single largest block of good, current-generation hardware at the site and it is the first thing that should move — worth more than the entire unassigned pool combined.
Dice: VWC de 0.4–9.0% con CE de poro en 0.0–0.08 mS y cero riegos en 14 días no es un cultivo con poca agua — es una sonda en medio seco o al aire. HR 40–44%, VPD 1.9–2.2 kPa: sala seca y sin cultivo.
No puede decir: si SJ se está montando o desmantelando. Ambas historias encajan igual de bien y llevan a decisiones opuestas.
La pregunta: si SJ se monta, nada se mueve. Si SJ se desmantela, estos 18 dispositivos son el mayor bloque de hardware de generación actual del sitio.
The fleet is overwhelmingly the previous hardware generation. Surfacing the choice — not making it for you.La flota es mayoritariamente de la generación anterior. Se presenta la opción — no se decide por usted.
| Generation | Models | Count | In service | Dead | Free to move | Note |
|---|---|---|---|---|---|---|
| OLD-GEN | TEROS-12 (41) · ATMOS-14 (12) · H11x repeaters (16) | 69 | 41 | 26 | 2 | Pre-TEROS One. Trade-in eligible.Pre-TEROS One. Elegible para canje. |
| NEW-GEN | TEROS One (6) · Climate One (3) · EE894 (3) | 12 | 0 | 2 | 10 | All of it sits in the SJ rooms. None of it is in a live grow.Todo está en las salas SJ. Nada en cultivo activo. |
| MESH | node (46) · repeater H34x (2) · gateway (2) · sink (2) | 68 | 34 | 15 | 17 | Radios. H34x repeaters are battery; H11x are PoE.Radios. H34x a batería; H11x por PoE. |
69 of the 133 device rows are old-gen, and 41 of them are load-bearing right now — they are the probes and repeaters keeping the seven live Moreno Valley grow rooms visible. Moving them means going dark in a room that is currently mid-crop, so they are not really "available" in any near-term sense.
Of the old-gen units that are not load-bearing, 26 are dead. A dead TEROS-12 has no relocation value at all — but it may still have trade-in value against current-generation hardware. That is the trade worth pricing: rather than crating 26 dead old-gen units to a new site, put them toward TEROS One / Climate One credit.
The point to sit with: "transfer the equipment" and "the equipment is worth transferring" are different claims. The hardware that is genuinely free to move is 9 radios. The hardware with real value is either in service (and shouldn't move mid-crop) or in the SJ rooms (and depends on the SJ answer). A trade-in conversation may be the more valuable one.
69 de 133 son de generación antigua, y 41 están en uso ahora mismo — mantienen visibles las siete salas activas. Moverlas = dejar a ciegas una sala en pleno cultivo.
De las antiguas que no están en uso, 26 están muertas. Una TEROS-12 muerta no tiene valor de reubicación — pero sí puede tener valor de canje.
El punto: "transferir el equipo" y "vale la pena transferir el equipo" son afirmaciones distintas. La conversación de canje puede ser la más valiosa.
48 hours is the full retention of AROYA's per-device telemetry endpoint — measured, not assumed: a request spanning more than 2 days returns HTTP 400, and windows older than 48 h return zero points. Battery, signal, link quality and travel time therefore cannot be trended back 14 days for any device at any facility. Everything below is the complete available history.48 h es la retención completa del endpoint de telemetría por dispositivo — medido: >2 días devuelve HTTP 400, y ventanas anteriores devuelven cero puntos.
| Serial | Room | Battery (48 h mean) | Flag |
|---|---|---|---|
| No radio is below 3.5 V. Battery health across the live mesh is clean.Ningún radio por debajo de 3.5 V. | |||
| Serial | Room | Signal (48 h mean) | Flag |
|---|---|---|---|
| No device is weaker than −85 dBm. Mesh signal is healthy across every live room.Ningún dispositivo por debajo de −85 dBm. | |||
Nodes and TEROS probes are battery. The site's 16 H11x repeaters are PoE; its 2 H34x repeaters are battery. ATMOS-14 climate sensors, the 2 gateways and the 2 sinks are PoE. A PoE device reading dead is a cabling/switch question, not a battery question — worth checking before any of the 16 H11x repeaters gets written off.Nodos y sondas TEROS son de batería. Los 16 repetidores H11x son PoE; los 2 H34x son de batería. ATMOS-14, gateways y sinks son PoE. Un dispositivo PoE muerto es un problema de cableado, no de batería.
Relevant because it interacts with the transfer plan. The update is WP4 → WP5, pushed over the air by the gateway to every device that is online at the scheduled time. Go/no-go per device: online · battery ≥ 30% · signal ≥ 50%.La actualización es WP4 → WP5, enviada por aire desde el gateway a todo dispositivo en línea. Criterio: en línea · batería ≥ 30% · señal ≥ 50%.
Nothing bad. A device that is offline at the scheduled push simply does not receive WP5 — it stays on WP4, unharmed. It is not damaged and it is not lost. If one of these units is later brought back online, it can still be recovered by reverting a gateway to WP4 and re-running — there is no time limit on that. The constraint is WP4↔WP5 compatibility, not damage.
Practical read for Sipkoi: the 92 live devices meet the online criterion today and their battery and signal are comfortably inside the go/no-go bands (see the tables above — nothing below 3.5 V, nothing weaker than −85 dBm). The push is remote by default; a site visit is the exception, not the assumption. The 39 silent units simply sit the update out and stay on WP4. If any of them are worth reviving, bring them online before the scheduled push and they will take WP5 with everything else — otherwise they can be recovered later, or routed to retire / trade-in.
Nada malo. Un dispositivo fuera de línea en el momento del envío simplemente no recibe WP5 — se queda en WP4, sin daño. Es recuperable después revirtiendo un gateway a WP4, sin límite de tiempo. La restricción es compatibilidad WP4↔WP5, no daño.
Los 92 dispositivos vivos cumplen los criterios hoy. El envío es remoto por defecto; una visita al sitio es la excepción. Los 39 silenciosos se quedan en WP4 — si vale la pena revivirlos, póngalos en línea antes del envío; si no, van a retiro / canje.
online / battery / rssi / lastCommunication flags were unavailable. Every liveness call in this report is derived from measured telemetry instead — which is a stronger claim than a flag, but it means "alive" here means "emitted data", not "AROYA says online".La sesión SPA fue invalidada. Toda la vitalidad se deriva de telemetría medida, no de banderas./harvests/ returns 404. Nothing in this report infers a growth phase or a cultivar, and "crop-active" means only "the substrate shows shots and dryback" — not a named phase.public_api /harvests/ devuelve 404. Nada aquí infiere fase ni cultivar./devices/<id>/chart/ over >2 days → HTTP 400; day-2, -3, -5, -7, -10 and -14 windows all return zero points. Per-device battery/signal history cannot go back 14 days. The 14-day view here comes from zone series instead.Medido: >2 días → HTTP 400. El historial por dispositivo no llega a 14 días.