The Station

Ikara is two machines that only make sense together: one in a paddock north of Canberra, one in a data centre in Sydney. Everything else — the radio, the battery, the sensors, these pages — hangs off that pair.

The site is not a hilltop, which is what most people picture. Ikara sits in a shallow valley north of Canberra — a high water table underneath it, and a very low noise floor above it. The second of those is the entire reason it exists.

What follows is what those two machines are looking after.

The unit

The Ikara cabinet is a custom-made enclosure, roughly 1.2 m by 1.2 m and about 500 mm high. It is built from marine plywood, and the choice of wood was deliberate: steel would have turned the inside into an oven in summer, and in cold weather it would have run with condensation on the inner surfaces. Wood behaves better on both counts.

This is the first version of the cabinet. It may not stay wooden forever, but for now that is the plan.

The roof is a separate stood-off section of corrugated roofing. It is not there to keep the rain out so much as to keep the sun off the top of the unit — a shade layer with air moving between it and the cabinet itself.

How well any of that works is not a matter of opinion: the environment page shows the current temperature inside the unit, and it is public.

Power

At the centre of it is a 300 Ah lithium iron phosphate (LiFePO₄) battery, charged through a Victron 100/30 MPPT controller. Live voltage, charge state and daily yield are on the power page.

The array is being expanded to three 250 W panels, bought secondhand for A$40 the lot. Testing so far shows them yielding around 200–210 W, which is a respectable result from panels that were very nearly free.

Protection is layered, and both layers fail safe:

• Under-voltage. A Victron BatteryProtect drops the load if battery voltage falls too far. “Load” here means everything inside the Ikara cabinet except the solar charging circuit — so the site can always keep charging itself back up, even after it has shut everything else down.
• Over-temperature. A simple normally-closed thermal cut-out sits at a high point inside the cabinet, set to 50 °C. If it gets too hot the switch opens, which pulls the disable line on the BatteryProtect and shuts down everything that produces heat. No logic, no firmware, nothing to crash — it either conducts or it does not.

Both of those are likely to change. A second battery is a possibility, and the battery may yet move out of the Ikara cabinet altogether into its own enclosure outside. For now, this is what is there.

Radio

Ikara exists first and foremost so that I can operate a FLEX-8400 from home. I run it remotely from my QTH in Canberra, either through SmartSDR — Marcus’ client, DL8MRE — or from a Maestro, which is a physical console with real knobs.

But I am not on the air anywhere near as often as the station is switched on. A transmitter sitting idle in a quiet valley is a waste of a quiet valley, so the receive side has been opened up to the amateur community at large.

There are two KiwiSDRs in the cabinet, both running the latest firmware, which gives eight simultaneous users each with their own frequency and waterfall. They live at sdr.ikara.vk1hf.com — no software needed, they run in a browser.

The plan is to give each receiver its own antenna rather than share one: perhaps a north–south and an east–west, so the two are hearing genuinely different things instead of the same thing twice.

One honest caveat for anyone listening: the site is quiet by suburban standards, but it is not clean yet. There is still some QRM being generated at Ikara itself. Tracking down what is making it, and removing it one source at a time, is an ongoing job.

Connectivity

The site is on Starlink. That solves the hard problem — there is no cable to a paddock — and immediately creates another one.

Starlink puts its customers behind carrier-grade NAT. You share a public address with a great many other subscribers, which is fine for browsing and hopeless for a remote station: outbound connections work, and nothing can ever connect inwards. No port forwarding, no fixed address, nothing to aim a radio client at.

The way around it is to stop trying to reach the site at all, and have the site reach out instead — which is exactly why the station is split in two.

When the cabinet powers up and Starlink comes good, IKARAPI dials out to IKARA-EXIT-NODE and holds a tunnel open. The exit node forwards traffic back down it. The effect is that everything inside the Ikara cabinet behaves as though it were sitting in a rack in Sydney, on a public address — and carrier-grade NAT stops mattering.

Every page on this site, and both KiwiSDRs, are reached that way.

Environment

As of August 2026 the open question is heat. A sealed wooden box in a paddock, full of equipment that makes its own warmth, through an Australian summer — nobody knows yet how that goes. So it is measured, from several directions at once, and all of it is on the public environment page.

• Raspberry Pi CPU. The Pi reports its own core temperature, which is the first thing to complain if the cabinet is cooking.
• The radio. The FlexRadio reports its own internal temperature too.
• BME680. Temperature, humidity and gas, wired to the Pi. Its readings run a little optimistic and I do not entirely trust them — but the gas sensing is worth having, so it stays.
• Ikara 1, 2 and 3. Three Zigbee temperature and humidity sensors placed deliberately rather than conveniently: left to right, and low to high, so the readings describe how heat stratifies inside the cabinet instead of just what one corner is doing.
• Water. A Zigbee leak sensor in the bottom of the cabinet. If there is ever real water in there, I want to know before the equipment does.

The Zigbee devices talk to a SONOFF USB transceiver on IKARAPI, and anything that matters is pushed to my phone through Pushover. If you are not using Pushover yet, you should be — go and have a look.

One detail that only makes sense once you know to look for it: the cabinet is mounted very slightly out of level, on purpose. Any water that gets in runs to one particular corner — which is exactly where the leak sensor is sitting.

Security

Ikara sits alone in a paddock, a long way from anyone who could keep an eye on it. So it keeps an eye on itself.

• Fixed camera. A high-resolution camera covering the cabinet and its immediate surroundings.
• 360° camera. A second high-resolution camera covering everything else, so there is no approach that is not seen.
• Microphone. There is audio in the unit as well as vision.
• Human detection. Not simple motion detection — kangaroos and moving grass would trigger that constantly. The system distinguishes people, and alerts me if someone walks near the site.

None of this is subtle, and it is not meant to be. A remote site is only worth building if you can trust it to be there when you come back.