Industry Insights

How far does 470 MHz actually reach? A link budget walkthrough

Free-space math promises hundreds of kilometres at 470 MHz; deployed meter networks get 2-10 km. Here is the budget, the derating and how to size a site.

Link budget

The two numbers that never agree

Ask two engineers how far a 470 MHz link reaches and you get answers five orders of magnitude apart. Ask a spreadsheet and you get one answer: with a 26 dBm transmitter and a receiver that hears −116 dBm, the path budget is 142 dB, and 142 dB at 470 MHz is 639 km of free space. Ask a technician standing in a basement meter room and you get the other answer: "about three kilometres, and not through that wall."

Both are correct. The spreadsheet is answering a question nobody asked — how far the link goes when the earth is flat, the air is dry, both antennas are 30 m up, and nothing stands between them. The technician is answering the real question.

This walkthrough is about the gap between those two numbers, because that gap is where projects are won and lost. Every figure below comes from either a published formula or the parameter table of a shipping product, and the arithmetic is shown so you can disagree with it.

Free-space path loss against distance at 470, 868 and 2400 MHz

Build the budget in one line

A link works when the received power is above the receiver's sensitivity. Written as a sum, everything that helps is positive and everything that hurts is negative:

P_rx = P_tx − L_cable + G_tx − L_path + G_rx
Link works when:  P_tx − L_cable + G_tx + G_rx + |Sensitivity|  ≥  L_path

That last line is the useful one. The left side is a property of your hardware — it does not change when you move the site. The right side is a property of the path — it changes with every wall, tree and roof you add. Design work consists of buying enough of the left side to cover the right side with margin left over.

Term

Symbol

Typical value in a metering project

Source

Transmit power

P_tx

17 dBm (indoor DTU) to 26 dBm (outdoor DTU)

product parameter table

Cable and connector loss

L_cable

0 dB (direct SMA whip) to 2 dB (5 m of RG-58)

cable datasheet

Antenna gain

G_tx + G_rx

2 dBi whip typical; 0 dB for a stub antenna inside a cabinet

antenna datasheet

Receiver sensitivity

S

−111 dBm to −116 dBm at 19.2 kbps

product parameter table

Path loss

L_path

computed, then derated

ITU-R P.525 for the free-space term

The free-space term is the only part with a closed-form answer, and it is older than the transistor: ITU-R Recommendation P.525 gives 20·log₁₀(d) + 20·log₁₀(f) + 32.44 for distance in km and frequency in MHz. At 470 MHz that is 85.9 dB at 1 km, 99.9 dB at 5 km and 105.9 dB at 10 km — an unglamorous 6 dB per doubling, which is exactly why narrowband sub-GHz links outrun 2.4 GHz links on the same power.

What the catalogue actually gives you

Here is the same arithmetic applied to real parameter tables rather than to a hypothetical radio. Every figure in the last three columns is computed from the first two:

Model

Type

P_tx

Sensitivity

Budget

Free-space distance

With 10 dB fade margin

TP5500

outdoor terminal DTU

26 dBm

−116 dBm

142 dB

640 km

202 km

TP5510

gateway DTU

26 dBm

−116 dBm

142 dB

640 km

202 km

TP5603

point-to-point DTU

17 dBm

−116 dBm

133 dB

227 km

72 km

TP5301

RS485 bridge module

20 dBm

−111 dBm

131 dB

180 km

57 km

TP6306

RS485 collector

20 dBm

−111 dBm

131 dB

180 km

57 km

Read the table for what it is worth and then throw the distance column away. It is not wrong; it is unconditioned. Its real use is comparison: the outdoor DTU carries 11 dB more budget than the RS485 bridge, and 11 dB buys you a factor of 3.5 in free-space distance — or, more realistically, one extra concrete floor.

Link budget at 470 MHz comparing 17, 20 and 26 dBm transmitters against three receiver sensitivities

Where the free-space answer dies

Four things eat the margin, in roughly this order of damage:

  • Fade margin. Radio paths are not constant. Rain, leaf wetness, a truck parked in the Fresnel zone and multipath all move the received level by 6–20 dB over a season. Reserve 10 dB as a floor and 15 dB if the link must hold through a wet winter.
  • Building penetration. A meter inside a reinforced-concrete basement is not 5 km from the antenna; it is 5 km plus a wall. Published penetration budgets for UHF into buildings run from 8 dB for a light interior wall to 25 dB and beyond for a below-grade pit with a metal cover. Measure one, do not guess: it is the single largest term in most failed designs.
  • Antenna height and Fresnel clearance. Ground reflection is not optional at 470 MHz. The first Fresnel radius at 5 km is roughly 18 m, and picking up 10 dB by lifting an antenna from 2 m to 8 m is routine. A pole is usually cheaper than a bigger radio.
  • Interference and duty cycle. A link that works at 10 % duty cycle fails at 60 % if the channel is shared. This is a scheduling problem rather than a physics one, which is why 500-channel planning exists on point-to-point DTUs.

The honest summary: free-space distance divided by 10 to 30 is the number to plan with. For the outdoor DTU that lands at 20–60 km line-of-sight and 2–8 km urban — which matches what field teams report for sub-GHz metering networks, and lines up with how long-range LPWAN operators describe their coverage.

A worked example: 5 km to a water-meter cluster

Take a distribution substation with a TP5510 gateway on the roof and a cluster of RS485 meters 5 km away in a residential block. Required path loss = free-space term + fade margin + penetration:

Term

Value

Note

Free-space loss at 5 km, 470 MHz

99.9 dB

computed

Fade margin

+10 dB

design floor

Penetration into a meter cabinet

+12 dB

assumed, measure it

Required path loss

121.9 dB

Budget, TP5301 (20 dBm / −111 dBm)

131 dB

9.1 dB spare

Budget, TP5500 (26 dBm / −116 dBm)

142 dB

20.1 dB spare

So the cheap bridge module works — with 9 dB of headroom, which is one bad week of weather away from not working. The outdoor DTU keeps 20 dB in hand, enough to absorb a second wall or a poorly aimed antenna. If the site has any obstruction you did not survey, that 11 dB difference is the whole decision, and it costs less than the truck roll it prevents.

Transmit power against receiver sensitivity for every product in the catalogue

Airtime: why the fastest symbol rate is not always the right one

Range and rate trade against each other, and the trade is quantified easily. A 100-byte reading is 800 bits, so the time on air is 800 ÷ bit rate:

Symbol rate

Airtime for a 100-byte reading

Sensible use

2.4 kbps

333 ms

maximum range, sparse readings, mains-powered nodes

19.2 kbps

41.7 ms

default for metering: good range, low collision risk

76.8 kbps

10.4 ms

dense clusters, firmware transfer, short range

Nothing is free here: the higher rates need better SNR, and the sensitivity column in any parameter table is quoted at a specific rate. Halving the airtime by quadrupling the rate typically costs 3–6 dB of sensitivity — the reason two figures in a datasheet always belong together.

How to verify a site in one afternoon

  1. Install one node at the worst meter, not the nearest. The link that matters is the last one, and it is always underground.
  2. Log RSSI at the receiver while the transmitter sweeps its channels. A clean channel shows a flat noise floor; a busy one shows spikes you cannot hear in the datasheet.
  3. Walk 200 m past the point where packets stop arriving. Fading is Rayleigh-distributed, so the edge of coverage breathes — record where packets arrive at 50 % and where at 99 %.
  4. Write down antenna height, cable length and wall type. In six months this log is the only thing that explains why one cluster needs retries.
  5. Repeat at midday and at midnight. Temperature drift moves crystal references and, on marginal links, changes the answer.

Frequently asked

Does 26 dBm really double the range of 17 dBm?

No — 9 dB of extra power buys a factor of about 2.8 in free-space distance and rather less in clutter, because beyond a certain point the limit is the environment, not the transmitter. Above roughly 20 dBm, antenna height and placement beat power every time.

Can I use a higher-gain antenna instead of a higher-power radio?

Usually yes, and it is the cheaper upgrade: 6 dBi of antenna gain adds the same 6 dB as quadrupling transmit power, without increasing the current draw or the regulatory exposure. The catch is that a narrow beam needs aiming.

Why does the same module reach 8 km on one site and 900 m on another?

Because one path had line of sight and the other had a floor slab, a metal cabinet and a hill. Sensitivity and power are fixed; path loss is not. This is why the parameter table should be read as an input to a calculation, not as a range claim.

What sensitivity figure should I plan with?

The one quoted at the symbol rate you will actually run. A −116 dBm figure at 19.2 kbps becomes roughly −111 to −113 dBm at 76.8 kbps. Using the best number in the table with the fastest rate is the most common way to lose 5 dB before the first packet is sent.

If you want the short version: put the numbers in a table, subtract the environment, and reserve 10 dB. Then pick the model whose remaining margin still looks comfortable — the catalogue lists power, sensitivity and interface for every module, gateway and terminal, and an engineer can tell you which two candidates fit a specific site.

More in Industry Insights