Point east-south-east, but do not expect a fast link at 10–15 m
The prior survey was right about the bearing and the Deloraine tower. It was too optimistic about the height, because it used a bare-earth terrain model and a 25 m tower. With the real 39 m tower, the ACMA licence data and the 1.19 m canopy model, the link is better than the prior model in one direction and worse in another: the tower is higher and stronger, but the ridge at 5.8 km is covered in 18–38 m forest, and a 15 m mast sits inside that canopy.
What to do
- Aim 124.6° true (about 111° magnetic; verify local declination) at the Deloraine tower 7304001, 27.4 km away. Use a compass app that shows true north, or sight on a known landmark and correct for declination.
- Put the mast at the SE edge of the container clearing, 30–40 m from the container, on a healthy 25–30 m eucalypt. A 15 m mount is the minimum; 20–25 m is materially better. Keep the cable run short and the PoE surge at the container.
- Test Optus n8 900 MHz first (25 MHz, ~75 dBm EIRP), then Telstra B28 700 MHz (20 MHz, ~66 dBm EIRP). The ATL has one physical SIM + eSIM, so test one carrier at a time. Do not commit to a plan until you have measured both at the tree.
- If 15 m is the hard limit, expect a marginal link: roughly 8–20 Mbps in good conditions, less in rain, and a real chance of dropouts. If you can go to 25 m, the model roughly doubles; if you can move 225 m SE to the big clearing, 40–80 Mbps is available on the Sassafras path — but that is a different site, and you said you do not want it.
The three numbers that matter
EIRP figures are the ACMA licensed maxima for the sector facing the property, minus a 2 dB operating backoff in the link budget. They are not a promise of what the tower runs at every hour.
What changed from the 22 October report
The good news: the Deloraine tower is 39 m high, not 25 m; its 700 MHz sector is licensed at 3,926 W EIRP (65.9 dBm), not 55 dBm; and its 270° sector points at the property. The prior report's 10–13 Mbps at 15 m was conservative on the radio but optimistic on the canopy. The bad news: the 5.8 km ridge is forested to 18–38 m, and the 15 m ray is 13–25 m inside that canopy. The canopy loss is the largest single term in the link budget — bigger than the terrain diffraction.
The container is in a forest pocket. The first kilometre falls away; the ridge at 5.8 km does not.
The site is not a bare paddock. The container sits in a small clearing roughly 25–30 m across, surrounded by 20–25 m eucalypt forest. To the ESE the ground drops away for the first kilometre — which is why a 15 m mast is not immediately blocked by the nearby trees — but at 5.7–6.0 km the path crosses a forested ridge. That ridge, not the trees beside the container, is the limiting obstruction.
Canopy by distance from the container
| Ring | Median | 90th percentile | Max |
|---|---|---|---|
| 0–10 m | 0 m | 0 m | 0 m |
| 10–20 m | 0 m | 4 m | 13 m |
| 20–30 m | 0 m | 16 m | 22 m |
| 30–50 m | 14 m | 22 m | 27 m |
| 50–100 m | 22 m | 26 m | 32 m |
| 100–200 m | 22 m | 28 m | 37 m |
| 200–300 m | 22 m | 28 m | 43 m |
| Ridge, 5.9 km | — | — | 29.8 m |
The clearing is real, but it is small. A 15 m mast at the SE edge is just above the local regrowth (13–19 m) and still below the 20–25 m forest around it. At the 5.9 km ridge the canopy is ~30 m; the 15 m ray passes ~24 m below the canopy top.
The Deloraine path at 15 m and 25 m
The profile below is the real DEM plus the canopy model. Tan is the ground; green is the canopy top; the blue band is the first Fresnel zone; the red fill is the part of the path where the ray is inside the canopy. At 15 m the ray clears the bare-earth ridge by only ~5 m, but the canopy top is ~24 m above the ray. At 25 m the ray is still ~16 m inside the canopy, but the obstruction length and foliage loss fall.
The 5.9 km ridge is the critical point. The first kilometre is clear because the ground falls away; the 225 m SE clearing is not needed for the Deloraine path — it is needed for the Sassafras path at 63°, which is blocked by the near ridge at 600 m.
Why the close towers fail
The Sassafras tower is only 12.5 km away and has a 40 m antenna, but the near ridge at 600 m is 75 m above the ray at 15 m. The Sheffield tower is 4.2 km west and has 110 m of terrain in the way. Neither gets better at 20–25 m: the obstruction is at 600–900 m, close to the container, so raising the mast by 10 m only improves the clearance by ~10 m against a 75–110 m block. The only way to use those towers is the 225 m SE clearing, which puts the mast on the other side of the near ridge.
The ridge is forested, not bare
At 5.9 km the bare-earth ground is 260.6 m and the canopy top is 290.4 m. The 15 m ray is at 266.5 m — 5.9 m above the ground but 23.9 m below the canopy top. The signal has to pass through ~300 m of forest at a grazing angle. Using ITU-R P.833, that is ~20 dB of foliage loss at 700 MHz and ~24 dB at 900 MHz. This is the single largest term in the link budget, and the largest uncertainty in this report.
45 sites within 30 km. Only two site groups are even worth testing from the container.
We screened every Telstra and Optus site in the 2026 ACCC Mobile Infrastructure Report within 30 km, then pulled the ACMA licence records for the sites that matter: antenna height, sector azimuth, tilt, EIRP, beamwidth and channel width. The table below ranks the result at a 15 m mast, with the canopy model included. “Blocked” means the near ridge or the northern/western rim gives more than 40 dB of diffraction — no realistic speed, regardless of band.
| Site · band | Carrier | km | Bearing | Tower | EIRP dBm | RSRP 15 m | Mbps 15 m | Verdict |
|---|---|---|---|---|---|---|---|---|
| 7304001 LTE B28 700 (20 MHz); 5G n5 850 (15 MHz) |
Telstra | 27.4 | 124.6° | 39.0 m ACMA |
65.9 | -95.0 | 8–12 | Viable — primary (B28) |
| 7304001 5G n8 900 (25 MHz); LTE/n28 700 (10 MHz) |
Optus | 27.4 | 124.6° | 30.5 m ACMA |
74.9 | -91.7 | 15–25 | Viable — fastest (n8) |
| 7304004 LTE B28 700 (20 MHz); LTE850; NR850 |
Telstra | 22.8 | 144.9° | 30.0 m assumed |
65.9 | -104.0 | 3–8 | Marginal fallback |
| 7304007 LTE B28 700 (10 MHz) |
Optus | 22.9 | 144.9° | 30.0 m assumed |
68.0 | -105.0 | 2–5 | Marginal fallback |
| 7304002 LTE B28/850/1800/2100; NR850 |
Telstra | 15.2 | 110.8° | 28.0 m assumed |
65.9 | -112.0 | 0–2 | Blocked by ridge |
| 7307001 LTE B28 700 (20 MHz) |
Telstra | 12.5 | 64.5° | 40.0 m ACMA |
57.7 | -118.0 | 0 | Blocked by near ridge |
| 7307011 LTE/n28 700 (10 MHz) |
Optus | 12.7 | 63.7° | 41.0 m ACMA |
69.1 | -116.0 | 0 | Blocked by near ridge |
| 7306008 LTE B28/850/1800/2100/2600; NR850/NR2600 |
Telstra | 4.2 | 269.8° | 29.7 m ACMA |
66.3 | -150.0 | 0 | Buried behind western ridge |
| 7306008 LTE B28/1800/2600 |
Optus | 4.2 | 269.8° | 35.0 m ACMA |
67.0 | -150.0 | 0 | Buried behind western ridge |
| 7305003 LTE B28 700 |
Optus | 4.4 | 55.6° | 25.0 m assumed |
65.0 | -140.0 | 0 | Buried behind near ridge |
| 7305001 LTE B28 700 |
Telstra | 5.6 | 34.2° | 25.0 m assumed |
65.0 | -140.0 | 0 | Buried behind near ridge |
| 7307008 LTE B28/1800; NR850/NR3600 |
Telstra | 15.6 | 13.0° | 25.0 m assumed |
65.0 | -140.0 | 0 | Blocked by northern rim |
| 7310001 LTE B28/1800/2100; NR850/NR3600 |
Telstra | 18.5 | 346.6° | 25.0 m assumed |
65.0 | -150.0 | 0 | Blocked by northern rim |
| 7307005 LTE/5G all bands |
Telstra | 25.1 | 29.8° | 25.0 m assumed |
65.0 | -150.0 | 0 | Blocked by northern rim |
| 7315001 LTE/5G 850/3500 |
Telstra | 29.4 | 324.5° | 25.0 m assumed |
65.0 | -150.0 | 0 | Blocked by western rim |
EIRP and tower height are ACMA licence values for the sector closest to the property. Rows marked “assumed” have no matched ACMA record; they use a 25–30 m tower and a 65 dBm EIRP and should be treated as screening only. “Mbps 15 m” is the central model output, not a speed test. The full band-by-band table is in §04.
Deloraine 7304001
27.4 km, 124.6°. Telstra 39 m; Optus 30.5 m. Both have low-band 5G, but the tower heights and sector patterns differ. This is the only site group with a workable path to the container.
Mole Creek Rd 7304004/7304007
22.8 km, 144.9°. A marginal fallback: a lower tower, a nearer ridge and more canopy. Useful only if Deloraine is congested or fails in a storm; not a primary path.
Everything else
Sassafras (12.5 km), Sheffield (4.2 km), Claude Rd (4.4–5.6 km), the northern 5G sites and the western sites are all behind the near ridge or the rim. Distance is not the deciding factor here; terrain is.
The surprise: Optus's 900 MHz 5G is the fastest pipe from the container
The prior report recommended Telstra because its Deloraine site has more bands and a taller tower. That is still true for robustness. But the ACMA data shows something the prior model missed: Optus has a 25 MHz channel at 900 MHz (5G n8) with a licensed EIRP of 74.9 dBm, while Telstra's low-band 5G n5 is 15 MHz at 850 MHz (63.2 dBm) and its 700 MHz B28 is 20 MHz at 65.9 dBm. At a 15 m mast, Optus's n8 has roughly a 3–4 dB advantage over Telstra's B28, and a larger advantage over n5, even after the higher 900 MHz foliage loss and Optus's lower 30.5 m tower.
Telstra — the safe, high-tower choice
- 7304001 Deloraine, 27.4 km, 39 m antenna, 270° sector; the property sits at its northern edge (~35° off-axis).
- LTE B28 700 MHz, 20 MHz, EIRP 65.9 dBm — the workhorse. Modelled 8–12 Mbps at 15 m, 25–35 at 25 m.
- 5G n5 850 MHz, 15 MHz, EIRP 63.2 dBm — usable but narrower. 4–6 Mbps at 15 m, 12–20 at 25 m. A 10 MHz LTE B5 carrier at 65.5 dBm is also on site.
- LTE B3/B7 and 5G n78 are licensed but canopy-limited at 27 km; n78 is not design-grade.
- Why Telstra: the 39 m tower gives the best terrain margin; B28 is a wide, robust 20 MHz channel; Telstra's regional coverage and backhaul are generally stronger.
Optus — the faster low-band pipe
- 7304001 Deloraine, 27.4 km, 30.5 m antenna, 330° sector — 25.5° off-axis at the property.
- 5G n8 900 MHz, 25 MHz, EIRP 74.9 dBm — the widest low-band channel on the tower. Modelled 15–25 Mbps at 15 m, 40–60 at 25 m.
- LTE/n28 700 MHz, 10 MHz, EIRP 68.1 dBm — narrower than Telstra's 700; 5–8 Mbps at 15 m.
- 5G n1 2100 MHz (5 MHz) — too narrow and too high in frequency for this path.
- 5G n78 3500 MHz — only on the 200° sector, which points away from the property (104° off-axis). Not available.
- Why Optus: 25 MHz at 900 MHz is a big low-band pipe, and 900 MHz penetrates canopy better than 1800/2100. The lower tower is the trade-off.
Band-by-band at the container
| Carrier | Band | EIRP dBm | RSRP 15 m | Mbps 15 m | RSRP 25 m | Mbps 25 m | Note |
|---|---|---|---|---|---|---|---|
| Optus | 5G n8 900 MHz · 25 MHz |
74.9 | -91.7 | 15–25 | -85.1 | 40–60 | Best low-band pipe |
| Telstra | 4G B28 700 MHz · 20 MHz |
65.9 | -95.0 | 8–12 | -88.4 | 25–35 | Most robust fallback |
| Telstra | 5G n5 850 MHz · 15 MHz |
63.2 | -99.6 | 4–6 | -92.7 | 12–20 | 10 MHz LTE B5 at 65.5 also on site |
| Optus | 4G/n28 700 MHz · 10 MHz |
68.1 | -96.4 | 5–8 | -90.1 | 15–20 | Narrow channel |
| Telstra | 4G B3 1800 MHz · 20 MHz |
71.8 | -101.3 | 2–4 | -94.0 | 10–15 | Canopy-limited |
| Telstra | 4G B7 2600 MHz · 20 MHz |
71.4 | -102.1 | <2 | -95.0 | 5–8 | Canopy-limited |
| Telstra | 5G n78 3500 MHz · 80 MHz |
80.3 | -92.1 | not design-grade | — | — | Cell edge at 27 km |
| Optus | 5G n1 2100 MHz · 5 MHz |
66.1 | -106.7 | <1 | — | 2–3 | Narrow, weak |
| Optus | 5G n78 3500 MHz · 35 MHz |
78.2 | — | 0 | — | 0 | Sector points 200°, away |
Central model values with a 3 dB tree-crown loss at 15 m and no crown loss at 25 m; 2 dB EIRP operating backoff; 5 dB clutter/misc; 1 dB cable. Bandwidths are the assumed carrier bandwidth; the ACMA licensed emission can be wider (e.g., 35 MHz at 1800, 40 MHz at 2600), so this is a per-carrier estimate, not the full licensed emission. The 25 m column assumes a pole above the local crown, not a trunk mount inside it. Ranges are in the height section.
Do not read “5G” as “fast” here
The useful 5G at this site is low-band: Optus n8 900 MHz and Telstra n5 850 MHz. The 3.5 GHz n78 layer is licensed at high EIRP, but at 27 km it is far beyond the normal cell edge, and the uplink/timing/beam management will not close reliably through ~300 m of canopy. The ATL may see an n78 cell; do not design around it. LTE B28 700 MHz remains the most robust bearer.
Coverage and plans
Neither carrier's “5G Home Internet” product is designed for a BYO CPE. Telstra's 5G Home Internet ($70/$80/$95 a month for 1000 GB, with a supplied Smart Modem) and Optus's 5G Home Internet use carrier-supplied hardware and an address check; the ATL 5G is not on either supported-device list. Ask for a data-only mobile broadband SIM (month-to-month if possible), confirm the APN and that the plan permits a third-party router, and test before committing. NBN wireless is not an option because the ATL is 4G/5G only. Starlink is the benchmark alternative if the cellular link fails the field test.
15 m is the minimum. 25 m roughly doubles the speed. 30 m is better still.
The height curve is not linear because the limiting obstruction is the canopy on the 5.9 km ridge. At 10–15 m the ray is deep inside that canopy; at 20–25 m it is near the top; at 30 m it clears most of it. The tree-crown loss also falls as you move the antenna above the local crown. Move the slider to see the modelled central value and range.
Telstra B28 700 MHz
Optus n8 900 MHz
Central values are the model midpoint; the shaded band is a ±6 dB uncertainty range, which covers the canopy-density, EIRP-backoff and CPE-gain assumptions. The 15 m values include a 3 dB tree-crown loss; 20 m+ assumes a pole above the local crown.
10 m
Deep in the crown and deep in the ridge canopy. 2–8 Mbps modelled. Only worth it if the tree cannot take a higher mount; expect rain fade and dropouts.
15 m
The minimum for a usable link. 8–20 Mbps modelled, with Optus n8 at the top of that range. Still inside the ridge canopy; a wet canopy or high wind will push it down.
25 m
The sweet spot if the tree and budget allow. 25–50 Mbps modelled; the ray is near the top of the ridge canopy and above the local crown. This is the height to aim for.
If 15 m is a hard limit
At 15 m the link is usable but marginal. The model's central values are 10 Mbps (Telstra B28) and 19 Mbps (Optus n8), with a range of 3–28 and 6–48 Mbps. The lower end is a wet canopy, a low tower power setting or a congested cell. Plan for 8–15 Mbps typical, not 30. If that is not enough, the options are: (a) a 20–25 m mount, (b) a carefully cleared 30–50 m window toward 124.6° (check Tasmanian native-vegetation rules first), or (c) the 225 m SE clearing.
Within 75 m of the container, placement barely changes the link. The 5.9 km ridge does.
We scored a 10 m grid within 75 m of the container at a 15 m mast: local canopy, ground height, distance and the canopy obstruction along the 124.6° path. The differences are under 1 dB. The best local position is the SE edge of the clearing (30–40 m from the container), because the cable run is short, the local regrowth is 13–19 m, and the path enters the forest slightly sooner. Do not expect a big gain from moving the tree 20 m; the ridge is the dominant term.
The map uses Esri World Imagery and OpenStreetMap. The canopy overlay (toggle in the layer control) is the 600 m local canopy image. Tower popups show the modelled 15 m speed and the ACMA tower height. Deloraine is 27 km ESE; Mole Creek Rd is 23 km SSE.
Candidate mast positions
| Position | Distance from container | Local canopy | Path at 124.6° | Comment |
|---|---|---|---|---|
| Container −41.372605, 146.378650 | 0 m | 0 m | Clear first 1 km; 300 m inside ridge canopy | Shortest cable; mast in the clearing. |
| C1 — SE clearing edge −41.37270, 146.37900 | 31 m | 13.5 m | Clear first 1 km; 300 m inside ridge canopy | Best compromise: short cable, local regrowth, slight head start. |
| C2 — further SE −41.37280, 146.37900 | 36 m | 19.2 m | Clear first 1 km; 300 m inside ridge canopy | Taller tree, but the crown is 19 m; a 15 m mount is inside it. |
| C3 — forest edge −41.37290, 146.37910 | 50 m | 20.9 m | Clear first 1 km; 300 m inside ridge canopy | More cable, no modelled gain; only if the tree is exceptional. |
| 225 m SE clearing −41.37431, 146.38011 | 225 m | 23.6 m (surrounding) | Opens the 63° Sassafras path, not the Deloraine path | The engineering alternative; 40–80 Mbps to Sassafras. You do not want it, so treat it as a benchmark. |
Which tree?
Choose a healthy, wind-firm eucalypt at the SE edge of the clearing, 25–30 m tall, with a straight trunk and no dead limbs or lean toward the container. C1 (31 m from the container) is the best compromise. A 15 m mount is inside the tree's crown if the tree is 25–30 m; a 20–25 m mount is above the local regrowth and near the top of the ridge canopy. Have an arborist assess the tree before committing; do not climb it yourself.
5G is useful here only in its low-band form
The ATL 5G is a 4x4 MIMO Cat 20 LTE/5G CPE. It supports n5 (850), n8 (900), n28 (700) and n78 (3500). At 27 km, the low bands are the only ones that close; n78 is a capacity layer for much shorter links, not a 27 km rural bearer.
n8 900 MHz · Optus
25 MHz, 74.9 dBm EIRP. The best low-band pipe in the model. 15–25 Mbps at 15 m, 40–60 at 25 m. The catch: it relies on Optus's lower 30.5 m tower and a 25.5° off-axis sector; test it before trusting it.
n5 850 MHz · Telstra
15 MHz, 63.2 dBm EIRP. Usable but narrower. 4–6 Mbps at 15 m, 12–20 at 25 m. A 10 MHz LTE B5 carrier at 65.5 dBm is also on site. Telstra's 39 m tower gives it the better terrain margin, but the channel is half the width of Optus's n8.
n78 3500 MHz · both
80 MHz on Telstra, 35 MHz on Optus. High EIRP, but 27 km is far beyond the normal n78 cell edge; the uplink and beam management will not close reliably. Telstra's n78 sector is at 270°; Optus's is at 200° and points away. Do not design around it.
The 4G workhorse
Telstra's LTE B28 700 MHz, 20 MHz is the most robust bearer at this site: the widest low-band LTE channel, a 39 m tower and a sector pointed at the property. It is not 5G, but it is the link that will still work in rain, wind and congestion. Use it as the anchor and let the ATL prefer n8/n5 when the signal is good.
What you bought, what it can do, and the four things to fix before you build
The core equipment is sensible for an off-grid rural link. The corrections are about compatibility, not capability: the ATL is a single-SIM (plus eSIM) device with no external antenna port; the Reolink WiFi camera cannot use the 6 W solar panel; the PoE switch must output active 48 V, not passive 12 V; and the MPPT should be a 150/85, not a 100/30.
MikroTik ATL 5G R16
Outdoor 5G CPE, IP66, −40 to +70°C, 311 × 311 × 179 mm, max 10 W, 12–56 V DC or 802.3af/at PoE. 4×4 MIMO, 16 dBi peak antenna gain (the 16 dBi figure is the mid-band peak; the low-band elements are physically smaller, so we model ~9 dBi at 700–900 MHz). Bands: 5G n1/3/5/7/8/20/28/38/40/41/75/76/77/78; LTE B1/3/5/7/8/20/28/32/38/40/41/42/43. LTE Cat 20, 5G SA/NSA.
- One micro SIM + eSIM — not two physical SIMs. You can run one carrier at a time (or an eSIM profile plus the physical SIM, but only one modem). The prior report's “two SIMs” is wrong; plan to swap/test.
- No external antenna connectors. The unit is sealed. The integrated directional antenna is what you get; aim it carefully. The 16 dBi peak is at 3.5 GHz; do not expect 16 dBi at 700 MHz.
- RouterOS band/cell lock — use it to force B28, n5, n8 or n28 during the field test and to record RSRP/RSRQ/SINR, PCI and cell ID.
WAVLINK AX3000 outdoor AP
WN573HX3-class, AX3000 (573 Mbps 2.4 GHz + 2402 Mbps 5 GHz), 4 × 8 dBi omni antennas, IP67, 1 GbE port, 802.3af/at PoE. The included passive PoE converter is not waterproof; use the active PoE output from the switch. Typical draw is ~12–18 W; allow 30 W at the switch for 802.3at class 4.
- Mount it 1 m below the ATL on the same pole, antennas vertical. It provides local WiFi for the camera and the container; it is not the backhaul.
- Use 2.4 GHz for the Reolink camera (better penetration through the container/trees) and 5 GHz for local devices.
- Configure it as a plain AP, not a repeater, with the ATL as the router.
Reolink TrackMix WiFi + 6 W panel
The TrackMix WiFi is a 12 V 1 A, <12 W camera with 2.4/5 GHz WiFi. It has no battery and no solar input. The 6 W panel belongs to the battery/LTE version of the camera (TrackMix LTE), not the WiFi model. The panel cannot run this camera.
- Power the camera from the 12 V bus at the container (fused) or from a PoE splitter if the switch supplies 802.3af. Use a 12 V 1 A feed and a drip loop.
- Mount it under the container eave or on a short pole, aimed at the approach; connect it to the WAVLINK AP's 2.4 GHz SSID.
- If you want a solar-only camera, use the TrackMix LTE with its battery and 6 W panel; that is a different product, and it needs its own SIM.
12 V industrial PoE switch
The switch must output active 802.3af/at (48 V) to power the ATL and WAVLINK. Many “12 V PoE” switches are passive 12 V on the spare pairs and will not power these devices. Check the model's datasheet for: 802.3af/at compliance, 48 V output, total PoE budget (≥60 W), per-port 30 W, 12 V input range, DIN rail, and −20 to +60°C rating.
- If the switch is passive 12 V only, add a 12 V → 48 V active PoE injector (or a Victron Orion 12/48 DC-DC) for the ATL and AP, or replace the switch with an active model.
- Use a PoE surge protector at the container end of the tree cable; bond it to the ground rod.
- Keep the total PoE budget ≥60 W: ATL 10 W + AP 30 W + margin.
Compatibility checklist
| Item | Requirement | Status / action |
|---|---|---|
| ATL 5G | 802.3af/at 48 V PoE, 10 W; 12–56 V DC alternative | OK if the switch is active 48 V; otherwise use a 48 V injector. |
| WAVLINK AP | 802.3af/at 48 V, 30 W class | Check the switch's PoE standard and budget; the included passive injector is indoor-only. |
| Reolink TrackMix WiFi | 12 V 1 A, <12 W; no solar input | Fix power from 12 V/PoE splitter; the 6 W panel is not usable. |
| Renogy 200Ah | 12.8 V, 2560 Wh, low-temp charge cutoff | OK; insulate/heat the battery box for winter. |
| 2 × Powerbay 440 W | 2 series: 69.8 V Voc, 59.3 V Vmp, 15.94 A Isc | OK for a 150 V Victron MPPT; 150/85 recommended. |
| Victron MPPT | ≥70 A output at 12 V; ≥1000 W PV | Choose 150/85 (1200 W, 85 A) for headroom; 150/70 is the minimum; 100/30 is unsuitable. |
| PoE cable | UV-rated Cat6, 23 AWG, ≤50 m | OK; use a surge protector and drip loops. |
The 6 W panel is not a camera power supply
This is the same correction the prior report made: the TrackMix WiFi is a 12 V mains-powered camera. The 6 W panel is an accessory for the battery/LTE TrackMix. Do not plan the camera around it. Power the WiFi camera from the container's 12 V bus or a PoE splitter, and keep the 6 W panel for a future LTE camera or a small shed light.
880 W of panels and a 200 Ah battery are adequate — but winter is tight
The load is bigger than the prior report assumed: the PoE switch has to convert 12 V to 48 V, the AP can draw 30 W at 802.3at, and the camera is a 12 W device. A realistic design load is 40–45 W typical, 0.9–1.0 kWh/day. The 880 W array produces 1.05 kWh/day in June and 3.59 kWh/day in January. The battery gives ~2.4 days of autonomy at the design load. The system works, but a cloudy June week will test it.
Load schedule
| Load | Typical | Peak | Notes |
|---|---|---|---|
| ATL 5G CPE | 8 W | 10 W | 802.3af/at; 10 W max |
| WAVLINK AX3000 | 15 W | 30 W | 802.3at class 4; typical 12–18 W |
| PoE conversion + switch | 7 W | 10 W | 90% conversion + electronics |
| Reolink TrackMix WiFi | 8 W | 12 W | 12 V 1 A; IR on at night |
| MPPT self-consumption | 0.6 W | 0.6 W | Victron 150/85 |
| Misc. (router, phone) | 5 W | 10 W | Intermittent |
| Total | 43.6 W | 72.6 W | 0.94 kWh/day at 90% duty; peak is worst-case simultaneous |
The PoE switch input includes the ATL and AP power plus conversion loss; do not add the ATL/AP again. If your switch is 85% efficient and the AP draws 18 W, add ~3 W.
MPPT sizing
| Model | 12 V PV limit | Output | Verdict |
|---|---|---|---|
| 100/30 | ~440 W | 30 A | Unsuitable clips the 880 W array to half |
| 150/70 | 1000 W | 70 A | Minimum 880 W STC is close to the limit; little headroom for bifacial gain |
| 150/85 | 1200 W | 85 A | Recommended headroom for bifacial gain and cold Voc |
| 150/100 | 1450 W | 100 A | OK if you plan a third panel |
Two 440 W panels in series: Voc 69.8 V, Vmp 59.3 V, Isc 15.94 A, Imp 14.84 A. The 150 V input limit is fine. At 12 V, 880 W is ~65 A; the 150/70 is at ~93% of its 70 A rating, so the 150/85 is the better choice. A 30 A controller would clip the array to about 440 W.
Monthly energy balance
| Month | Generation kWh/day | Load kWh/day | Balance | Comment |
|---|---|---|---|---|
| Jan | 3.59 | 0.94 | +2.65 | Strong surplus |
| Feb | 3.20 | 0.94 | +2.26 | Strong surplus |
| Mar | 2.44 | 0.94 | +1.50 | Good |
| Apr | 1.63 | 0.94 | +0.69 | Marginal |
| May | 1.24 | 0.94 | +0.30 | Tight |
| Jun | 1.05 | 0.94 | +0.11 | Very tight |
| Jul | 1.17 | 0.94 | +0.23 | Tight |
| Aug | 1.60 | 0.94 | +0.66 | Marginal |
| Sep | 2.10 | 0.94 | +1.16 | Good |
| Oct | 2.73 | 0.94 | +1.79 | Good |
| Nov | 3.19 | 0.94 | +2.25 | Strong |
| Dec | 3.55 | 0.94 | +2.61 | Strong |
Generation: NASA POWER climatology for the site, 880 W array at 45° tilt, 0.75–0.82 performance ratio including temperature, soiling and MPPT losses. No bifacial gain is included; a 5% rear gain would add ~0.05–0.18 kWh/day. Load: 43.6 W at 90% duty. Autonomy: 2,304 Wh usable (90% DoD) ÷ 940 Wh/day = 2.4 days.
The winter problem
June generation is only ~1.05 kWh/day against a 0.94 kWh/day load. That is a 0.11 kWh/day surplus — less than the uncertainty in the model. A cloudy week, a cold battery that stops charging below 0°C, or an AP that draws 25 W will put the system into deficit. Fixes, in order: (1) shed load at night (camera IR off, AP schedule, CPE low-power), (2) add a third 440 W panel (1320 W total; June ~1.58 kWh/day), (3) add a second 200 Ah battery for autonomy, (4) insulate and heat the battery box so charging starts earlier on frosty mornings.
Battery, wiring and protection
Renogy 12 V 200 Ah LiFePO4
- 2,560 Wh nominal, 200 A BMS, 90% usable = 2,304 Wh.
- Low-temperature charge cutoff — the BMS will not charge below 0°C. Discharge is allowed to about −20°C. Insulate the battery box and fit a small heating pad (5–10 W) on a thermostat, or keep the battery in the insulated container.
- Charge at 0.2–0.5C (40–100 A); the 880 W array at 12 V is ~65 A, within range.
- Use a Victron SmartShunt for SOC, and a Victron MPPT with VE.Direct/Bluetooth for logging.
Wiring and fusing
- Main battery fuse: 100 A Class T or MRBF at the battery positive.
- MPPT fuse: 80 A between the MPPT and battery; 6 mm² (AWG 10) minimum, 10 mm² for a longer run.
- PV isolator: DC-rated 2-pole isolator on the roof/array; 4 mm² PV cable.
- PoE run: 48 V, 23 AWG Cat6, ≤50 m. At 30 W, 50 m of 23 AWG has ~4 V drop (8%); use 22 AWG or a shorter run if the switch output is 48 V rather than 50–57 V.
- Surge: PoE surge protector at the container, bonded to a ground rod; Ethernet surge at the switch.
Bill of materials (approximate, excluding owned equipment)
| Item | Spec | Qty | Est. AUD |
|---|---|---|---|
| Victron SmartSolar MPPT | 150/85, MC4 or Tr | 1 | 700 |
| Victron SmartShunt | 500 A | 1 | 250 |
| Active 48 V PoE switch / injector | 12 V input, 802.3af/at, ≥60 W | 1 | 300 |
| Cat6 UV direct-burial cable | 23 AWG, 50 m | 1 | 120 |
| PoE surge protector | 48 V, outdoor-rated | 1 | 150 |
| Mast/pole/brackets | 50 mm galv, stainless bands, standoffs | 1 | 400 |
| Lightning/ground kit | Rod, clamp, bonding, surge | 1 | 300 |
| Battery box/insulation/heater | Insulated, thermostat, 10 W pad | 1 | 300 |
| Fuses, breakers, cable, lugs | 100 A main, 80 A MPPT, 6–10 mm² | 1 | 300 |
| Mounting/hardware | Stainless, cable ties, conduit | 1 | 200 |
| Total | ~3,020 |
A short, well-grounded build: tree mast, one cable, one switch, one battery
The build is deliberately simple: the ATL and AP on a pole up a tree, one PoE cable back to the container, an active switch, a 12 V battery bank and two panels. The hard parts are the tree, the lightning protection and the winter power budget — not the networking.
| Location | Equipment | Notes |
|---|---|---|
| Tree, 15 m | ATL 5G R16 (top), WAVLINK AP (1 m below) | Aim ATL at 124.6° true. AP antennas vertical, 2.4 GHz for camera. One Cat6 cable, PoE. |
| Container | Active PoE switch, PoE surge, battery, MPPT, SmartShunt, fuses | Keep the battery and electronics out of the weather; insulate the battery box. Camera on the eave, 12 V feed. |
| Ground/roof | 2 × 440 W panels, 45° tilt, north | Series array; no shade; bifacial rear gain; DC isolator; 4 mm² PV cable. |
| Ground | Earth rod, bonding, surge | Bond the mast, switch, battery negative and surge protector; one earth point. |
Tree mounting
Choosing and attaching
- Tree: a healthy, wind-firm eucalypt at C1 (31 m from the container), 25–30 m tall, straight trunk, no lean toward the container, no dead limbs or fungal brackets. Have an arborist assess it.
- Pole: galvanised steel, 60 mm OD, 4 mm wall, 2.5–3 m long, mounted with two stainless band clamps and 25 mm standoffs. A 50 mm/3.5 mm pole is marginal at 15 m in a 40 m/s gust (about 2.3 kNm at the base); use 60 mm/4 mm or guy wires. Do not screw into the tree; allow for growth. Inspect the bands annually.
- Height: 15 m is the minimum; 20–25 m is materially better. If the tree is 25–30 m, a 20–25 m mount is above the local regrowth and near the top of the ridge canopy. Do not exceed the tree's safe working height or the pole's wind rating.
- Wind: the ATL is 0.096 m² frontal area; the AP plus antennas ~0.06 m². At a 40 m/s gust the load is ~150 N; the bending moment at 15 m is ~2.3 kNm. A 60 mm/4 mm pole is the minimum for an un-guyed 15 m mount; use guy wires if the tree is exposed or the pole is longer than 3 m.
Lightning and cable
- Lightning: a tree-mounted mast is exposed. Fit a separate earth rod and bond the mast, the PoE surge protector and the switch to it. Keep the antenna below the tree top if possible; the tree may take the strike, but side flashes are still a risk. Follow AS/NZS 1768 and AS/NZS 3000; have the installation checked.
- Cable: UV-rated Cat6, 23 AWG, in conduit on the ground, with a drip loop at the tree and at the container. Stainless cable ties, strain relief at both ends, 50 m maximum. Use a PoE surge protector at the container end.
- Service loop: leave 1 m of slack at the mast for tree sway and maintenance; do not pull the cable tight.
- Labelling: label both ends “ATL PoE — do not disconnect live” and record the port.
AP, camera and local WiFi
The WAVLINK AP is the local WiFi, not the backhaul. Mount it 1 m below the ATL with the four omni antennas vertical. Use 2.4 GHz for the Reolink camera (better penetration through the container and trees) and 5 GHz for phones/laptops. The camera is 12 V 1 A; power it from a fused 12 V feed at the container or a PoE splitter. Do not use the 6 W panel. Set the AP to a plain AP mode with the ATL as the router; use WPA3/WPA2 and a strong password, and disable WPS.
Solar and battery
Array
- Two 440 W panels in series: 69.8 V Voc, 59.3 V Vmp, 15.94 A Isc. The 150 V MPPT input is fine.
- 45° tilt, true north. No shade from the container, trees or the mast. Bifacial rear gain is a bonus; do not design around more than 5–10%.
- DC isolator at the array, 4 mm² PV cable, MC4 connectors, cable glands.
- Keep the panels clear of the tree canopy; the container clearing is small, so consider a ground mount at the southern edge of the clearing rather than the container roof.
Battery
- Insulated battery box, out of the weather, with a 10 W thermostat-controlled heater pad for winter. The BMS will not charge below 0°C.
- 100 A main fuse at the battery positive; 80 A MPPT fuse; SmartShunt; DC isolator.
- Vent the box (LiFePO4 does not vent hydrogen in normal use, but the enclosure should not trap heat).
- Keep the battery above 5°C for charging; a container corner that gets morning sun is better than a shaded corner.
Do not climb the tree yourself
Working at 15 m with a pole, antennas and a live PoE cable is a two-person job with fall arrest, a throw line, and a spotter. Use a certified arborist or tower climber; check insurance and ask for a method statement. If the tree is near powerlines, call the network operator before any work.
Test at 10 m, 15 m and 20 m before you commit to a plan
This report is a desk study. The only way to know which carrier and height work is to measure at the tree. Use a temporary mast or a ladder-safe test pole at C1, with a 12 V battery and a PoE injector, and record RSRP, RSRQ, SINR, band, PCI and speed for both carriers at each height. Do it on a dry day and a wet day; canopy loss changes with rain.
Test procedure
- Aim: 124.6° true (~111° magnetic; verify declination). Use a compass app with true north, or sight on a distant landmark and correct.
- Height: test 10 m, 15 m and 20 m (and 25 m if the tree allows). At each height, record the ATL's RSRP/RSRQ/SINR and the serving band/PCI.
- Band lock: in RouterOS, lock B28, n5, n8 and n28 one at a time; record the metrics and a speed test for each. Let the modem also run in auto mode and record what it picks.
- Carriers: test Telstra and Optus separately (one physical SIM + eSIM; swap or use the eSIM profile). Record the plan/APN and the time of day.
- Load: run a speed test with the AP and camera connected, not just the CPE, to see the real system throughput.
- Weather: repeat on a wet/windy day. If the link drops below 5 Mbps or loses the cell, note it; that is your design margin.
Measurement sheet
| Height | Carrier | Band | RSRP | SINR | DL Mbps | UL Mbps |
|---|---|---|---|---|---|---|
| 10 m | Telstra | B28 | ||||
| 15 m | Telstra | B28 | ||||
| 20 m | Telstra | B28 | ||||
| 10 m | Optus | n8 | ||||
| 15 m | Optus | n8 | ||||
| 20 m | Optus | n8 |
Target: RSRP better than −95 dBm, SINR above 0 dB, and a stable cell for 10 minutes. If you see −100 dBm or worse at 15 m, the canopy is thicker than modelled; go higher or reconsider the site.
Decision rule
If 15 m gives ≥10 Mbps on one carrier and the cell is stable, build at 15 m with that carrier as primary and the other as backup. If 15 m gives 5–10 Mbps, build at 20–25 m if the tree allows. If 20–25 m still gives <5 Mbps, the container site is not viable for a reliable link; use the 225 m SE clearing (Sassafras path) or Starlink. Do not sign a 24-month plan before this test.
What was computed, what was assumed, and what this report cannot tell you
What was computed
- Tower inventory: 2026 ACCC Mobile Infrastructure Report (data at 31 January 2026), all Telstra and Optus sites within 30 km; 45 sites, 15 screened in detail.
- ACMA licence data: antenna height, sector azimuth, tilt, EIRP, beamwidth and channel width for the key sites, retrieved via GIS Geek/RFNSA mirrors on 23 October 2026. This is the source of the 39 m Deloraine height and the 65.9/74.9 dBm EIRPs.
- Terrain: AWS Terrain Tiles z14 (~7 m bare-earth DEM), sampled at 10–50 m along each path, with 4/3 Earth curvature and Deygout/P.526 diffraction.
- Canopy: Meta/ETH global canopy height v2 at 1.19 m, sampled along each path and around the container; cross-checked against Esri World Imagery. The 5.9 km ridge canopy (29.8 m) is the critical value.
- Foliage: ITU-R P.833 vegetation attenuation for the length of path inside the canopy, capped at 30 dB.
- Link budget: FSPL + diffraction (central = mean of single-edge+3 dB and Deygout; the multi-edge term can over-count a single ridge, so the two bracketing values are reported) + foliage + 5 dB clutter/misc + 1 dB cable; UE gain 9 dBi at 700–900 MHz, 11 at 1800, 13 at 2600, 16 at 3500; 2 dB EIRP operating backoff; 3 dB tree-crown loss at 15 m, 6 dB at 10 m, 0 above 20 m.
- Throughput: Shannon efficiency with 2×2 MIMO (4×4 for n78), 0.85 overhead and 0.75 load factor, capped by band. Ranges are ±6 dB.
- Power: NASA POWER monthly climatology for the site, 880 W at 45° tilt, 0.75–0.82 PR; load 43.6 W at 90% duty; 2,304 Wh usable battery.
What this cannot tell you
- No on-site measurements. The speeds are hypotheses, not promises. Tower power, cell load, weather and the actual canopy density can move the result by ±6 dB or more.
- The canopy model is not LiDAR. Meta's 1.19 m CHM is a modelled canopy height; the Tasmanian LIST LiDAR CHM (2 m) would be more accurate. Local tree heights and crown density should be measured.
- Tower EIRP is a licensed maximum. The tower may run lower, or the sector may be tilted/optimised differently. The 2 dB backoff is an assumption.
- Uplink is not fully modelled. The downlink budget is the focus; a 27 km uplink at 900 MHz is feasible but the cell edge and timing advance may limit it. Test upload speed.
- Planning and safety are not assessed. Tree clearing, native vegetation, bushfire, lightning and working at height need a local professional and the council/forest practices authority.
- Plan pricing changes. The carrier plan notes are indicative; confirm current prices, data allowances and BYO-router rules before ordering.
Sources
- ACCC Mobile Infrastructure Report — data release, 2026: data.gov.au.
- ACMA Register of Radiocommunications Licences / RFNSA, via GIS Geek — Telstra Deloraine and Optus Deloraine; RFNSA.
- MikroTik ATL 5G R16: product page, datasheet.
- WAVLINK AX3000 outdoor AP: product page.
- Reolink TrackMix WiFi: spec sheet (12 V 1 A, <12 W).
- Powerbay PB-DH96NE-440 panel: product page (440 W, Voc 34.92 V, Vmp 29.65 V, Isc 15.94 A).
- Renogy 12 V 200 Ah LiFePO4: representative product page; confirm your exact model's BMS and low-temperature behaviour.
- Victron SmartSolar MPPT 150/70–150/100: technical specifications.
- Meta CHM v2 canopy height: source.coop/tge-labs/meta-chm-v2.
- AWS Terrain Tiles: Registry of Open Data.
- Tasmanian LIST cadastre and Tree Canopy Height metadata: thelist.tas.gov.au.
- NASA POWER climatology: power.larc.nasa.gov.
- Telstra 5G Home Internet: telstra.com.au; Optus 5G Home Internet: optus.com.au.
- Esri World Imagery: esri.com; OpenStreetMap: openstreetmap.org.
Bottom line, one more time
Aim 124.6° true at Deloraine. Build at 15 m only if you accept a marginal link; aim for 20–25 m. Test Optus n8 900 MHz and Telstra B28 700 MHz at the tree before signing a plan. If 20–25 m still fails, the 225 m SE clearing is the engineering answer, even though it is not the answer you wanted.