Industry Solutions / Telco

Power and Cooling for Base Stations, Exchanges and Network Sites Across Australia

Most critical power vendors sell into telco with a data centre catalogue and hope nobody notices. They quote a three-phase double-conversion UPS for a site that has never had an AC bus in its life. A macro base station, a transmission hut and a street cabinet do not run on AC with a UPS in front of it. They run on a rectifier plant that converts incoming mains to nominal minus 48 volts DC, feeds the radios and transmission gear directly, and floats a battery string permanently across that same DC bus. There is no transfer, no inverter, and no static switch, because there is nothing to transfer to.

That single architectural fact changes everything downstream. It changes what fails, what you maintain, how you measure runtime, how the site is earthed, what your spares holding looks like, and what a technician needs to be competent in when they arrive. It also changes the cooling problem, because in a shelter the equipment is usually not the thermal constraint. The battery is. Radios tolerate heat that would shut a server down. Lead acid does not, and the entire commercial case for shelter cooling in Australia turns on that point.

Indigi Power and Cooling is an Indigenous and Veteran owned critical power and cooling specialist, registered with Supply Nation and ICN Gateway. We work across both halves of a telco site, the DC plant and the thermal envelope that protects it, and we work at fleet scale rather than one unit at a time, because that is how carriers, tower companies, utilities and state agencies with private networks actually buy.

Telco site classes and what each one actually needs

Before specifying anything, establish which class of site you are dealing with. The answer to "AC UPS or DC plant" is decided here, not by preference, and the answer is genuinely different at each tier of the network.

Read that table across and the pattern is clear. From the tower to the exchange, the load is DC. Once you reach the parts of the network that look like IT, meaning the network operations centre, the office floor and the edge compute cabinet full of standard servers, the load is AC and a conventional UPS is exactly right. Vendors who only sell one of those two things will tell you their half is the answer everywhere. It is not.

Why telco runs on minus 48 volt DC and not on AC UPS

What a DC power plant is made of

A telco DC plant has four parts. Rectifiers convert incoming mains AC into regulated DC at roughly 54 volts float, which is what nominal minus 48 volts means in practice. They are hot-swappable modules in a shelf, and redundancy is achieved by fitting one more module than the load requires, which is why a rectifier failure at a well-configured site is a maintenance ticket rather than an outage. DC distribution takes that bus and splits it out through fused or circuit breaker protected feeds to each piece of equipment. The battery string sits permanently across the bus, floating, with no switching device between it and the load. A controller supervises the lot, managing float and boost charge, low voltage disconnect, temperature compensation and alarm reporting.

The consequence of that architecture is the thing generic UPS vendors miss. There is no transfer time and no transfer event, because the battery is already connected. An AC UPS has to keep an inverter running continuously so it can carry the load when mains fails. A DC plant simply stops charging and starts discharging. The failure modes are different, the efficiency profile is different, and the maintenance regime is different.

Why the polarity is negative

The positive side of a telco DC plant is bonded to earth, which makes the live conductor negative with respect to earth. That convention comes from the copper network era and it survives for a good electrochemical reason: a negatively biased conductor corrodes far more slowly than a positively biased one where moisture and dissimilar metals meet. On a network built from tens of thousands of buried and exposed connections, that mattered enormously, and it still does at the exposed cable entries and earthing points of a modern site.

Where an AC UPS genuinely is the right answer

Being right about DC plant does not mean being dogmatic. AC UPS belongs in several places on a telco network and specifying DC there would be an error. Network operations centres run standard servers, workstations, video walls and switching, all AC, and want three-phase online double conversion with generator backing. Edge compute cabinets holding commodity servers are AC loads even when they sit at the base of a tower. Office, retail and call centre sites are AC. Some transmission, microwave and test equipment is AC only. And at many exchanges the building services that keep the room alive, meaning the cooling plant, the fire system and the lighting, are AC and deserve UPS backing every bit as much as the switching equipment does, because a hot exchange fails just as surely as an unpowered one.

Our practical position on a mixed site is simple. Put the network load on DC plant where the equipment is DC, put the IT and ancillary load on a properly sized AC UPS, and make sure someone has actually drawn the boundary between them rather than assuming it. We cover the AC side through three phase UPS installation and commissioning and single phase UPS maintenance, and the outdoor and high ambient cases through our high temperature UPS systems for hot and outdoor environments.

Batteries at base stations, and the Australian heat problem

Why the shelter setpoint is set by the battery, not the radios

This is the single most useful thing to understand about a telco shelter in Australia. The active equipment is not usually what dictates the temperature you need to hold. Radio, transmission and power equipment is commonly specified for continuous operation at ambient temperatures far above what IT hardware tolerates, frequently to 45 degrees Celsius or higher. Valve regulated lead acid batteries are not. They want 20 to 25 degrees Celsius, and their published design life assumes it.

The widely used engineering rule of thumb is that VRLA float life roughly halves for every 10 degrees Celsius of sustained temperature above 20 degrees Celsius. Apply that to a real Australian site. A battery with a nominal 10 year design life at 20 degrees Celsius is looking at roughly 5 years at a sustained 30 degrees, and roughly 2.5 years at a sustained 40 degrees. A shelter in Mount Isa, Katherine or the Pilbara that is allowed to sit at 40 degrees through summer will consume battery strings at four times the rate the business case assumed. Multiply that across a fleet of a thousand sites and the cooling system stops being an HVAC line item and becomes the dominant driver of battery capital expenditure.

VRLA against lithium in a shelter

VRLA remains the incumbent at most Australian base stations, and for good reasons: low purchase price, universal familiarity, a mature recycling path and no battery management system to integrate. Its weaknesses are the ones above, plus weight, floor loading and a habit of failing without much warning between scheduled tests.

Lithium iron phosphate changes the arithmetic. It tolerates high ambient temperatures far better, which is precisely the Australian problem, delivers a much higher usable depth of discharge so you buy less nameplate capacity for the same runtime, occupies roughly a third to a half of the footprint and weight for equivalent energy, and reports its own state of health continuously through the battery management system instead of waiting for a technician. Against that, it costs more up front, it needs the DC plant controller and the battery management system to actually talk to each other, and it introduces fire and gas considerations that a lead acid string does not, which matters where batteries sit in an enclosed shelter or an exchange battery room. See gas detection for battery rooms and BESS for the detection side of that.

Our honest split: for a hot, remote, hard-to-reach site where a truck roll is expensive and the ambient is punishing, lithium usually wins on total cost even at a higher purchase price. For a temperate metropolitan site with good cooling, a short drive from a depot and an established VRLA spares pool, lead acid is often still the rational choice. Anyone telling you one answer fits a whole fleet has not costed the fleet.

Testing and replacement planning

A float voltage reading tells you almost nothing about whether a string will hold up for its rated runtime. Impedance or conductance testing at scheduled intervals, temperature logging inside the enclosure rather than at the wall, and periodic discharge testing on a representative sample of the fleet are what actually tell you where you stand. We run scheduled battery maintenance and replacement programmes and can model the cost of a staged fleet replacement using the UPS battery replacement cost calculator as a starting point for budget conversations.

Shelter and cabinet cooling, and why Free Cooling is a fleet decision

How Free Cooling works in a shelter

Free Cooling uses filtered outside air to hold the internal temperature instead of running a refrigerant compressor. The compressor is the single most energy-hungry component in a conventional air conditioner, so removing it from duty for most of the year is where the saving comes from. There are two implementations. Turbulent free cooling pushes and circulates outside air throughout the whole shelter and suits larger buildings with a high and evenly distributed heat load. Displacement free cooling introduces air at low velocity at floor level and lets the warm air rise and be extracted at the ceiling, forming a cushion of cool air around equipment height. Displacement uses less fan energy but does not suit large volume shelters, and the equipment layout has to be designed around the principle.

A well-designed free cooling unit also holds the shelter at slight positive pressure, so air leaves through any gap rather than entering it. That keeps dust, red sand, salt spray and insects out of the enclosure, which in inland and coastal Australia is worth as much as the energy saving. Unit manufacturers quote compressor energy savings in the range of 60 to 90 per cent depending on climate. Treat those as vendor figures to be validated against your own site data, because the achievable saving in Darwin is not the achievable saving in Ballarat.

When mechanical cooling is still required

Free cooling is not a universal answer and we will not pretend otherwise. It fails in two situations. First, where ambient temperature periodically exceeds the internal temperature you need to hold, which in northern Australia is most summer afternoons. Second, where the enclosure is simply too small to house a free cooling unit, which describes a great many compact 5G cabinets. In both cases the practical answer is a hybrid: free cooling doing the work for the large majority of hours, with a mechanical DX stage that engages when ambient conditions defeat it, and each capable of acting as partial redundancy for the other.

The shelter cooling range we work with covers exactly these cases. Wall mounted and split units for shelters where a compressor is required, telecom-specific units built for the dust and vibration of a radio site, free cooling units for the sites where ambient conditions allow it, and compact cabinet units for street furniture and small enclosures. For larger exchanges and network operations rooms the equipment becomes conventional precision cooling and the design rules of a server room or switchroom apply instead.

A commercial point on refrigerant licensing

Any work involving refrigerant requires ARCtick licensing under the Ozone Protection and Synthetic Greenhouse Gas Management Regulations, and there is a finite pool of licensed technicians, particularly outside the capital cities. A free cooling unit contains no refrigerant, so filter changes and routine servicing do not require an ARCtick licensed technician. Across a fleet of a thousand rural sites that is not a footnote, it is a material change to who can service your network and how quickly they can get there. Where mechanical cooling is fitted, every refrigerant task we perform is done by ARCtick licensed technicians, and air handling hygiene in occupied exchange spaces follows AS/NZS 3666.

Fleet economics: a telco buys a rollout, not a unit

A data centre operator buys one CRAC and one UPS and cares intensely about that specific machine. A carrier, tower company or utility network owner buys the same decision two thousand times and then lives with it for fifteen years. Everything about the evaluation changes.

Standardisation beats optimisation. The best unit for a given site is rarely the best unit for the fleet. A single standardised shelter cooling model and a single DC plant configuration mean one spares holding, one training package, one set of drawings, one commissioning checklist and one alarm profile in the network management system. A fleet with eleven different cooling units on it will spend more on the resulting complexity than it ever saved by picking each one on a site-specific basis.

Swap-out beats repair in the field. At fleet scale, diagnosing and repairing a unit on site is the expensive option. The efficient model is a like-for-like swap with a pre-tested unit and a bench repair back at the depot, which turns a four-hour roadside diagnosis into a forty-minute exchange and keeps the network running while the fault is investigated somewhere with light, tools and parts.

Truck rolls dominate the cost model. For a site four hours from the nearest depot, the cost of getting a two-person crew there and back, with vehicle, permits, site access and traffic management, routinely exceeds the value of the component being replaced. That is why remote monitoring is not a nice-to-have on a telco fleet. Rectifier alarms, battery state of health, shelter temperature, door position and cooling unit status reported back over SNMP or Modbus into your existing network management platform let you distinguish a genuine fault from a nuisance alarm, batch several jobs into one visit, and carry the right part the first time.

Maintenance windows are network windows. Work on a live site is planned around traffic, not around business hours. We schedule to the network's change window, work to a documented method statement, and treat "no unplanned service impact" as the acceptance criterion rather than "the job got done".

Regional, remote and coastal sites

Australia has network sites in conditions that most equipment catalogues were not written for, and pretending otherwise produces early failures.

Cyclonic regions. Sites across northern Queensland, the Top End and the Pilbara sit in wind regions C and D under AS/NZS 1170.2, and the shelter, its mounting, the external cooling unit and its brackets all have to be rated and fixed accordingly. A split condenser bolted to a shelter wall with generic fixings is a projectile in a category three system, and the resulting damage is rarely limited to the condenser.

Coastal salt exposure. Sites within a few kilometres of the coast fall into the highest corrosivity categories under ISO 9223, and standard aluminium fin coils fail there in a fraction of their design life. Coated coils, marine-grade fixings, sealed cable entries and a genuine washdown regime in the maintenance schedule are the difference between an eight year asset and a three year one.

Access and mobilisation. Remote sites carry constraints that have nothing to do with engineering: wet season road closures, permits and cultural heritage clearances, four wheel drive or helicopter access, and long lead times on anything that has to be freighted. As an Indigenous owned business we take the cultural heritage and community engagement side of remote site access seriously rather than treating it as an obstacle, and we plan mobilisation so that a single trip covers power, cooling, battery and inspection scope together.

Solar and hybrid sites. A remote transmission hut running on solar with an oversized battery bank is a different design problem again, because there the battery is a daily cycling asset rather than a standby one, and cycle life, temperature and depth of discharge drive the whole economics.

Runtime, emergency calls and what the rules actually say

Base station outages affect people's ability to reach Triple Zero, and that is the reason runtime at a radio site is treated more seriously than runtime at a typical commercial building. It is worth being accurate about what is and is not required, because this area attracts a lot of loose claims.

There is no blanket Australian rule setting a minimum number of hours of battery backup at every mobile base station. The approach taken here has been targeted and largely voluntary or grant-funded rather than universally mandated. Under the Federal Government's Mobile Network Hardening Program, funding was provided for battery backup upgrades at 466 base stations to lift backup operation at those sites to at least 12 hours, alongside a second stage funding further resilience upgrades. Proposals to require a minimum of 24 hours at all base stations in high bushfire risk areas have been costed and debated but are not a general obligation on carriers.

Two practical points follow. First, if a site is in scope for a resilience programme, or supports a hospital, an emergency services facility or a community with no alternative coverage, its runtime target is a design input and not an afterthought, and it usually drives battery sizing and shelter cooling together, because you cannot hold 12 hours of runtime in a shelter that has cooked its battery string. Second, Australian mobile handsets can place a Triple Zero call over any available carrier's network, so a single site outage does not automatically remove emergency access where another carrier still has coverage. That is genuinely useful nuance, and it is also why single-carrier sites in areas with no overlapping coverage deserve the longest runtime in the fleet.

What Indigi does for telco network operators

Indicative pricing gives you a starting point before we quote. Cooling design and installation runs from $8,000 to $15,000 for small rooms, $25,000 to $45,000 for medium installations and $45,000 to $60,000 plus for large projects, with shelter and cabinet cooling quoted per site and per fleet once the standard configuration is fixed. On the AC side, UPS maintenance is $720 ex GST for a one-off visit or $1,940 per year ex GST for a standard annual plan. Single-phase UPS installation starts from $850 up to 3 kVA, $1,250 for 3 to 10 kVA and $1,800 for 10 to 20 kVA. Three-phase installation starts from $3,200 for small systems and $6,500 for medium. DC plant, battery and remote site work is quoted individually because mobilisation and access dominate the cost at fleet scale.

Frequently asked questions

Why do telecom equipment use DC?

Telecom equipment runs on DC because it removes conversion stages and lets the battery sit permanently across the load. In an AC system the mains has to be converted to DC to charge a battery, then back to AC by an inverter, then back to DC again inside every piece of equipment. A minus 48 volt DC plant does one conversion at the rectifier and feeds the equipment directly, with the battery floating on the same bus so there is no transfer event when mains fails. Fewer conversions means fewer things to fail, better efficiency and instantaneous backup, which is why the architecture has survived from the copper era into 5G.

What is a telecom shelter?

A telecom shelter is a purpose-built enclosure at a network site that houses radio, transmission and power equipment along with the DC plant and battery strings. It is typically a prefabricated concrete, fibreglass or steel structure at the base of a tower, sealed against dust and water, and fitted with its own cooling, fire detection, security and alarm monitoring. In Australia shelters in the north also have to be rated for cyclonic wind regions, and coastal shelters need corrosion protection on every external component. Smaller versions of the same idea appear as outdoor cabinets at street level and at compact 5G sites.

How does free cooling work?

Free cooling holds the temperature inside an enclosure using filtered ambient air rather than a refrigerant compressor. A controller compares outside and inside temperature and, whenever outside air is cool enough to do the job, brings it in through filters with fans and extracts the warm air. The compressor, which is the most energy-hungry part of a conventional air conditioner, only runs when ambient conditions defeat free cooling. Turbulent systems circulate air throughout the whole shelter, while displacement systems introduce air slowly at floor level and let warm air rise for extraction. Most units also hold the shelter at slight positive pressure so dust and salt cannot be drawn in.

How long can cell towers go without power?

It depends entirely on the battery capacity installed at that site and on the load the radios are drawing. Australian macro base stations commonly hold somewhere between 4 and 12 hours on battery, and sites upgraded under the Federal Government's Mobile Network Hardening Program were funded to reach at least 12 hours of backup operation. Remote transmission and solar-hybrid sites are often designed for 24 hours and beyond because nobody can reach them quickly. Sites with a permanent standby generator can run for days provided fuel is replenished. Real-world runtime is usually shorter than the design figure if the battery string is aged or has been running hot.

Do cell towers go out in a power outage?

Not immediately. Every carrier-grade base station has a battery string across its DC bus that picks the load up instantly with no transfer time, and larger sites add a standby generator behind it. A site goes down when that battery is exhausted, when the battery has degraded so far that it never had its rated runtime, or when the outage also takes out the backhaul link feeding the site. Long regional outages after storms, floods and bushfires are where site runtime is genuinely tested, which is why battery health testing and shelter cooling matter so much more in telco than the purchase price of the battery suggests.

What is the difference between a lithium-ion battery and a VRLA battery?

A VRLA battery is a sealed lead acid battery with a valve to vent excess pressure, cheap to buy, heavy, well understood and sensitive to heat, with float life roughly halving for every 10 degrees Celsius above 20 degrees Celsius. A lithium-ion battery, most commonly lithium iron phosphate in telecom use, stores far more energy per kilogram and per litre, tolerates high ambient temperatures much better, allows a deeper usable discharge, lasts longer in both float and cycling duty, and reports its own state of health through a battery management system. Lithium costs more up front and introduces fire, transport and integration considerations that lead acid does not. In hot, remote Australian sites lithium usually wins on total cost of ownership, while temperate metropolitan sites with good cooling can still justify VRLA.

What is passive cooled shelter?

A passive cooled shelter rejects its heat without any mechanical refrigeration, relying instead on insulation, thermal mass, shading, natural or fan-assisted ventilation and sometimes air to air heat exchangers. It works where the internal heat load is modest and the equipment tolerates temperatures that track ambient, which suits low-power transmission huts and repeater sites. It stops working when the heat load rises, when the site holds a lead acid battery string that needs to stay near 25 degrees Celsius, or when summer ambient regularly exceeds the internal limit. For most Australian macro base stations a passive shelter alone is not sufficient, and a free cooling unit with mechanical backup is the realistic configuration.

Where we work

Related services and industries

• High temperature UPS systems for hot and outdoor environments, for cabinets and shelters that run above standard ratings
• Battery maintenance and replacement including impedance testing and staged fleet renewal
• UPS maintenance plans for exchange, NOC and office AC loads
• Switchroom cooling for exchange and plant room spaces
• Server room cooling Australia for edge compute and network operations rooms
• Three phase UPS installation and commissioning
• Gas detection for battery rooms and BESS, relevant to enclosed shelters and exchange battery rooms
• UPS battery replacement cost calculator for fleet budget modelling
• Data centres and IT infrastructure, where the AC UPS and precision cooling rules apply
• Transportation, for roadside, rail and tunnel communications cabinets with the same remote site constraints
• Secure and critical facility UPS and CRAC cooling for government and emergency services networks

Planning a network power or shelter cooling programme? Send us your site count, DC plant load per site, current battery chemistry and age, shelter or cabinet type and target runtime. We will come back with a standardised configuration, a fleet maintenance model and an installed price per site. Indigi Power and Cooling is Indigenous and Veteran owned, and registered with Supply Nation and ICN Gateway. Contact Indigi Power and Cooling Battery Maintenance