How many radios can one repeater channel support before calls start getting blocked?

A single conventional repeater channel comfortably carries roughly 20 to 50 radios, and the spread is that wide because capacity is set by how much people talk, not by how many radios you buy. The engineering measure is offered traffic in Erlangs: one user making six transmissions an hour at eight seconds each offers 0.013 Erlangs, and a conventional channel starts to feel congested above about 0.20 to 0.25 Erlangs, roughly 19 such users. A DMR Tier II repeater gives you two voice slots on the same 12.5 kHz channel, so the same box carries around double an analogue one. A busy construction or event fleet transmitting every four minutes will block that same channel at ten to fifteen radios.

Why the answer is Erlangs, not a headcount

I get asked for a number, and the honest answer is that a headcount on its own tells me nothing. Fifty condominium security officers who key up twice an hour generate less load than twelve hotel banqueting staff during a function changeover. What matters is offered traffic, which is calculated as: A = (transmissions per user in the busy hour x mean transmission length in seconds) / 3600.

That gives a figure in Erlangs, where 1.0 Erlang means the channel is occupied continuously. Multiply by your radio count and you have the total load the channel has to absorb in its worst hour. Everything downstream, whether you need a second channel and whether trunking is worth the money, falls out of that one number.

A conventional (non-trunked) channel is contention-based. There is no queue: a user who keys up on a busy channel either talks over someone or waits and tries again. In practice, congestion becomes noticeable to users somewhere around 20 to 25 percent channel occupancy, and becomes an operational complaint above 30 percent. That is a much lower working point than people expect, and it is why the radios keep getting cut off is often a capacity problem being misdiagnosed as a coverage one.

• Analogue FM, 12.5 kHz: one voice path per repeater channel

• DMR Tier II, 12.5 kHz: two voice paths (two TDMA timeslots) per repeater channel

• Offered traffic per radio: A = (calls per busy hour x mean seconds) / 3600

• Practical conventional working point: 0.20 to 0.25 Erlangs per slot

• Erlang B at 2% blocking, for trunked systems: 1 channel 0.02 E, 2 channels 0.22 E, 4 channels 1.09 E, 8 channels 3.63 E

• Typical push-to-talk transmission: 6 to 10 seconds; a full exchange runs 3 to 4 transmissions

How many radios per channel in practice?

Run the arithmetic against three real fleet profiles and the answer separates cleanly.

A quiet fleet, such as condominium or office building security at three transmissions per user per busy hour at six seconds each, offers 0.005 Erlangs per radio. At a 0.25 Erlang working point that is around 50 radios on one slot.

A moderate fleet, such as hotel front-of-house, engineering and housekeeping at six transmissions an hour at eight seconds, offers 0.013 Erlangs per radio, giving about 19 radios per slot.

A busy fleet, such as a construction site during a lift or event operations, runs 15 transmissions an hour at ten seconds, or 0.042 Erlangs per radio. That is ten to twelve radios before the channel is visibly congested, and it explains why a site that works fine at 60 radios in the morning falls over with the same 60 radios during a concrete pour.

Does DMR give you twice the capacity?

Close to it, and this is the strongest single argument for digital on a capacity-limited site. DMR Tier II splits one 12.5 kHz channel into two TDMA timeslots, so one repeater carries two simultaneous conversations where an analogue repeater carries one. That is a genuine doubling of voice paths on the same licensed frequency pair, and it does not require a second frequency assignment from IMDA, which is usually the harder thing to get. It is worth being clear that this is a capacity gain, not a coverage gain: switching to digital DMR does not by itself fix a building's dead zones.

Trunking, whether Capacity Plus or DMR Tier III, pools all available slots so any user takes the next free one. The gain is real but it scales with channel count, not with the decision to trunk. At 2 percent blocking, four trunked slots carry 1.09 Erlangs, or 0.27 per slot; eight trunked slots carry 3.63 Erlangs, or 0.45 per slot. Below about four slots the trunking gain barely covers the extra infrastructure cost. Above six or eight it is decisive.

What GPS and data traffic do to your channel

This is the failure mode that catches people out. On a DMR system, ARS registration, text messaging and GPS location updates share the same timeslots as voice. Thirty location updates per radio per hour across a 100-radio fleet is not background noise. It is a substantial and continuous load sitting underneath the voice traffic, and it does not appear in anyone's capacity estimate because nobody counts it as a call. I have seen multi-channel systems where users needed three or four attempts before a call went through, and the arithmetic showed the combined voice and location traffic simply exceeded what the system could carry. If you are enabling GPS on more than about 30 radios, budget a dedicated data slot or drop the polling interval to five minutes or longer.

When do you add a channel, and what does that cost in Singapore?

Add a channel when busy-hour occupancy on the existing one is sustained above 25 to 30 percent, or when users routinely report needing a second attempt. The constraint in Singapore is rarely the hardware. It is the frequency. Each additional channel needs its own assignment under an IMDA localised radio-communication station licence, with the licence fee applying per channel, so the recurring cost scales with channel count. That is exactly why exhausting the two DMR timeslots on your existing pair before asking for a new frequency is the sensible order of operations. The licensing side is covered in more detail in our note on IMDA licensing and type approval for in-building repeaters.

Related questions

Does adding a second repeater double my capacity? Not on its own. Two conventional repeaters on separate channels only help if you split the fleet by department or talkgroup, and users left on the original channel see no improvement at all. Trunking the two together pools the slots and does deliver close to double.

Will a DAS fix a channel that is always busy? No. A distributed antenna system fixes coverage, meaning where the signal reaches, not capacity. Improving coverage on a congested channel usually makes congestion worse, because more radios can now reach the repeater.

How do I confirm the channel is genuinely overloaded? Pull the repeater's call log for the busiest hour and compute channel occupancy directly. Sustained occupancy above 25 to 30 percent, or a measurable rate of failed first attempts, confirms a capacity problem rather than a coverage one.

Suneast designs and installs two-way radio and in-building coverage systems in Singapore, and we size the channel plan from measured busy-hour traffic rather than a headcount. That means pulling call logs from an existing repeater where one is in place, or building an Erlang estimate from the actual operational pattern where it is a new system, then setting the timeslot and talkgroup plan and the IMDA licence application against that figure. Where the problem turns out to be coverage rather than capacity, we say so. Adding channels to fix a dead zone is an expensive way to solve the wrong problem.