Telecom towers

Rectifier system

Every module you leave running at light load is heat you pay for twice.

The rectifier turns AC into the −48 V DC the radios run on and keeps the batteries charged. Modules are most efficient near half load, but most sites run all of them at a fraction of that. Their losses also turn into heat the air conditioner has to remove.

−48 V DC plant, 6 × 3 kW modules (N+1), float 54.0 V, boost 56.4 V

Built and run to ETSI EN 300 132-2 (−48 V DC interface)

Start with one cluster of sitesSee how we prove results
Sketch of a −48 V DC rectifier cabinet: AC input breakers, controller, a shelf of six 3 kW rectifier modules with fans, the DC bus bar, load and battery breakers with current shunts and the low-voltage disconnect contactor, feeding the battery string and the radios, with numbered sensor points.12345678from AC boardAC input−48 V busradiosbattery stringLVD
Peak efficiency
96.0% at 50% load
All 6 on at 2.4 kW
93.4%
Efficiency mode saves
57 W per site
Across 1,000 sites
501 MWh a year

Anatomy and nameplate

Sketch of a −48 V DC rectifier cabinet: AC input breakers, controller, a shelf of six 3 kW rectifier modules with fans, the DC bus bar, load and battery breakers with current shunts and the low-voltage disconnect contactor, feeding the battery string and the radios, with numbered sensor points.12345678from AC boardAC input−48 V busradiosbattery stringLVD
  1. 1DC bus voltage. Float, boost, and the dip when the mains fails.
  2. 2Load current. What the radios draw: 44 A at the reference site.
  3. 3Current per module. Each module should carry the same share.
  4. 4AC input voltage and current. AC in against DC out gives the real efficiency.
  5. 5Module temperature and fan status. A failed fan makes the module derate.
  6. 6Battery charge current. Recharge after an outage, and the float current after that.
  7. 7Low-voltage disconnect state. Drops the load at 43.2 V to save the battery.
  8. 8Controller alarms. Read over SNMP, with traps for module and mains faults.
Rectifier system nameplate
Plant
Output−48 V DC nominal
Modules6 × 3 kW, hot-swappable
Capacity18 kW (15 kW with one module spare)
Module current55.6 A at 54.0 V
Float / boost54.0 V / 56.4 V (2.25 / 2.35 V per cell)
Temperature compensation−3 mV per cell per °C
Low-voltage disconnect43.2 V (1.80 V per cell)
ControllerSNMP polling and traps
Efficiency
Peak module efficiency96.0% at 50% load
At full module load95.4%
At 2.4 kW, all 6 on93.4%, 170 W of losses
At 2.4 kW, efficiency mode2 on, 95.5%, 113 W of losses

Telecom site power chain

The grid or the generator feeds the site through the transfer switch. The rectifier makes −48 V DC for the radios and keeps the battery charged. The air conditioner keeps the battery cool.

Grid: Mains supply, 400 VGridMains, 400 VFuel tank: 500 L diesel tankFuel tank500 LGenerator: 20 kVA diesel generator setGenerator20 kVA dieselTransfer switch: Switches the site between mains and generatorTransferswitchMains or DGRectifier: Turns AC into −48 V DCRectifierAC to −48 V DCShelter cooling: Draws from the AC board and keeps the battery coolSheltercoolingAir conditionerRadios: Radio equipment on the −48 V DC busRadios−48 V DC loadBattery: Carries the radios in an outageBatteryBackup in outage
  • Air
  • AC power
  • −48 V DC
  • Diesel
  • Has its own twin page

The physics

The equations behind every curve on this page.

DC power

P_dc = V · I

2.4 kW at 54.0 V is 44 A.

AC input

P_ac = P_dc / η(load per module)

Efficiency depends on how hard each module works, and peaks near half load.

Losses as heat

Q = P_ac − P_dc

Every watt lost is a watt the shelter cooling has to remove.

Float with temperature

V_float = 24 · (2.25 − 0.003 · (T − 25))

A warm battery needs a lower float voltage.

Redundancy

N_needed = ceil(P_dc / P_module) + 1

2 modules carry 2.4 kW with one spare.

Efficiency mode

A module wastes a fixed few watts just being on. Spread a small load over six modules and each one sits low on its curve. Park the ones you do not need and the rest work where they are best.

Module efficiency against load per module

2.4 kW site load on a 6 × 3 kW plant

Module efficiency against load per moduleBest range868890929496980102030405060708090100All 6 on: 93.4%2 on: 96.0%Load per module (% of 3 kW)Efficiency (%)
  • Module efficiency
Losses, all on
170 W
Losses, efficiency mode
113 W
Saved
57 W

Six modules sharing 2.4 kW each run at 13% load, low on the curve. Switch 4 of them to standby and the rest run near their best.

η = P / (P + p0 + k1·P + k2·P²)

Sharing and float

The modules share the load, and the controller sets the float voltage for the battery. Both drift, and both cost battery life when they do.

Current per module over two days

Six modules on the same bus

Current per module over two days02468101214051015202530354045Module 5 failsHourCurrent per module (A)
  • Modules 1, 3, 4 and 6
  • Module 2, carrying more
  • Module 5, fails at hour 30
Module 2 share
+18% (limit ±10%)
Modules needed for N+1
2 of 5 working

Each module should carry the same share. One that carries more runs hotter and ages first. When a module fails, the others pick up its share and the twin checks the spare is still there.

I_module = I_load / N_on N_needed = ceil(P_dc / P_module) + 1

Float voltage against battery temperature

24-cell VRLA string, −3 mV per cell per °C

Float voltage against battery temperature52.052.553.053.554.054.555.055.556.0101520253035404554.0 V at 25 °C53.28 V should be set at 35 °CBattery temperature (°C)Float voltage (V)
  • Compensated set point
  • Actual, compensation off
Overcharge at 35 °C
0.72 V (30 mV per cell)

A warm battery needs a lower float. Without compensation the string is overcharged on hot days and undercharged on cold ones.

V_float = 24 · (2.25 − 0.003 · (T − 25))

Failure modes

What breaks, what it looks like, what the twin does about it.

Failure modeThe physicsSignals that show itHow the twin catches it
Module failureA module stops converting. The others pick up its share.Module current falls to zero. The other modules’ currents step up.Sees the sharing shift and checks there is still a spare module for N+1.
Fan failure and thermal deratingWithout airflow the module heats and cuts its own output to protect itself.Module temperature up. Its output down while the load is the same.Compares each module’s temperature and output with its neighbours at the same load.
Uneven sharingA module set slightly high carries more than its share, runs hotter and ages first.One module’s current outside ±10% of the average.Checks sharing every reading and names the module that drifts.
Wrong float settingFloat too high overcharges and dries the battery. Too low and it never gets full.Float voltage off the temperature-compensated curve. Float current high or near zero.Sets the float voltage against the compensation curve for today’s battery temperature.
Low-voltage disconnect set too highThe load is dropped while the battery still has charge left.Radios drop early in an outage with battery capacity still in hand.Checks the disconnect setting against 43.2 V and the autonomy the site should have.

What the twin does not do

  • The twin does not change rectifier settings itself. Efficiency mode and float changes go through your controller, with a person approving.

How we catch it early

A physics twin and a PINN, working on one rectifier system.

The twin is a set of equations for this machine, tuned to its nameplate. Give it the flow, speed and conditions and it returns what every reading should be. It always gives the same answer for the same inputs.

A physics-informed neural network, or PINN, sits on top. It is trained on the plant’s own data and it is penalized whenever it breaks a physics rule. So it can only learn what the equations leave out: wear, fouling, friction. It cannot invent an answer that the machine could never produce.

  1. Step 1

    Measure

    Flow, pressure, current, temperature, vibration. Whatever the asset already has.

  2. Step 2

    Physics twin

    The equations of this machine, set to its nameplate. It says what the readings should be right now.

  3. Step 3

    Residual

    Measured minus expected. Demand, weather and speed are already taken out.

  4. Step 4

    PINN

    A small neural network that learns the cause of the gap, under the physics rules.

  5. Step 5

    Finding

    A named cause, a date, and the kWh or hours at stake. An engineer reviews it.

loss = (measured − predicted)² + λ · (physics residual)²

The second term is the rulebook. It is zero when the network obeys the equations.

Why this beats black-box AI

A black-box model learns patterns from history. It needs examples of each failure, and it goes quiet when the pump runs at a point it has not seen. The twin starts from physics, so it works on day one. The PINN adds what physics cannot know about this particular machine.

That is also what makes the savings countable. The baseline is a calculation, not a guess, so the gap between baseline and meter is a number a customer can check.

What the twin checks on this asset

  • Efficiency against the curve at today’s load: 93.4% with all on
  • Module sharing within ±10%
  • Float voltage against the temperature-compensated set point
  • N+1 kept after a module fails
Fixed alarmBlack-box AITwin + PINN
Needs failure history to learnNoYes, lotsNo
Says what is wrongNoRarelyYes, by name
Works at a flow it has never seenNoOften failsYes, the equations still hold
Puts a number on the savingNoNoYes, in kWh and dollars
Lead time on slow wearLateVariesWeeks to months
Every number can be tracedYesNoYes, to an equation

What the twin puts a number on

Energy and cost, from the equations.

Efficiency mode at 2.4 kW

57 W per site

170 W of losses with all 6 on, 113 W with 2 on

Energy per site

501 kWh a year

Running all year

Across 1,000 sites

501 MWh a year

Same load at every site

Less cooling to do

57 W of heat

19 W less air conditioner power at 35 °C outside

What we do for you

We watch it. We catch it. We prove it.

Our reliability engineers run the twin on your rectifier system, review every finding, and help your crew fix it. After the repair we compare the meter to the baseline and write the result in the ledger.

Start with one cluster of sitesSee the 90-day pilot

More assets

More from telecom towers, then the lead asset of each other industry.