FDMA vs TDMA Underground: Why Amplifier Group Delay Favours FDMA on Leaky Feeder and DAS Cascades

Ask a radio supplier which digital standard to put underground and the answer usually starts with spectrum. TDMA puts two calls on one 12.5 kHz carrier in DMR and P25 Phase 2, and four calls on a 25 kHz carrier in TETRA, so it halves the carriers you licence and the transmitters you combine. On the surface that argument is strong, and we have run the numbers for it.

Underground, the argument runs into a different problem. Everything a radio hears in a decline reaches it through a cascade: a head end, kilometres of radiating cable, a line amplifier every few hundred metres, sometimes a fibre hop and a remote, sometimes an off-air repeater at the portal. Every one of those elements delays the signal, and the amplifiers delay it in proportion to how narrow their filters are. FDMA does not care much about that delay. TDMA cares about it in three separate ways.

This article works those three ways with the numbers the standards actually publish: the ETSI specifications for DMR and TETRA, the TIA material that is public for P25 Phase 2, the delay tolerances that consultants and manufacturers have put in print, and the delay figures that amplifier makers put on their datasheets. The conclusion is not that TDMA cannot be used underground. It is used underground every day. The conclusion is that FDMA has margin where TDMA has a budget, that the budget is consumed by amplifier filtering far faster than by cable length, and that a designer who has not worked out which of the three delay questions applies to their mine tends to find out at commissioning.

The Short Answer

There are three questions, and they have different answers for the two access methods.

And one equipment fact sits underneath all three. By published datasheet, the latency through a channelised amplifier filtering a single 12.5 kHz channel is 60 to 90 µs per pass. Through a band-selective amplifier it is a few microseconds. Through a wideband mining line amplifier with a passband several megahertz wide it is a small fraction of a microsecond. Radiating cable is about 3.75 µs per kilometre. So one channelised repeater in the chain adds the delay of about 16 km of cable, in one pass, in each direction.

What Each Access Method Asks of the Channel

The difference between FDMA and TDMA is not the modulation. It is what the base station receiver needs to know about time.

FDMA gives each call its own carrier and transmits it continuously for as long as the radio is keyed. P25 Phase 1 sends C4FM at 4800 symbols per second in 12.5 kHz. NXDN, which Icom sells in Australia as IDAS, sends 4-level FSK at 4800 bit/s in a 6.25 kHz channel, which is 2400 symbols per second and a symbol about 420 µs long. Analogue FM has no symbol at all. In every case the receiver recovers whatever timing it needs from the signal itself, and it does not matter when the signal arrives.

TDMA puts two or four calls on one carrier by giving each a slot, and that only works if every radio transmits inside its slot as seen at the base station. The radio cannot know how far away the base station is, so the standards handle it the same way: the radio sets its transmit clock from the downlink it hears, and the base station leaves a guard time in each inbound slot to absorb the round trip.

For DMR, ETSI TS 102 361-1 states that the outbound channel is transmitted continuously while the base station is active, and that “the inbound channel has an unused guard time between bursts to allow Power Amplifier (PA) ramping and propagation delay”. The slot is 30 ms, the burst is 264 bits and 27.5 ms, and “on the inbound channel, the remaining 2,5 ms is used for guard time”. Within that guard, the standard sets aside a specific budget. Clause 10.2.3.1.3 says that “a 1 ms slot timing variation allowance is built in to the Normal Burst structure”, that “each MS shall time synchronize with the BS before transmitting”, and that the allowance lets a radio on the second slot transmit “up to 150 km from the BS without inter-slot interference” while another radio sits next to the base station on the first slot. The arithmetic is a round trip: 2 × 150 km at the speed of light is 1.0 ms.

For TETRA, ETSI EN 300 392-2 divides a 56.67 ms frame into four slots of 14.167 ms, each of 255 symbols at 55.56 µs. Annex O works the delay tolerance for the normal uplink burst as the slot length less the information length, the ramp time and a filter allowance, and arrives at 7 symbol periods, which is 389 µs. It then writes the range limit out explicitly: “R Terrestrial_max = 0,5 × Td × Ts × c = 58,3 km”, noting that “each symbol represents a distance of 8,33 km”. The MS side is just as specific. Clause 7.8 requires the mobile to adjust its internal timebase to the signals it receives from the base station, and to correct in steps of no more than 125/9 µs, about 13.9 µs, whenever the difference exceeds that.

For P25 Phase 2, the physical layer document TIA-102.BBAB is not public, so this article does not quote a timing allowance for it. What is public is the structure. Viavi’s application note gives 12.5 kHz, 12 000 bit/s, 6000 symbols per second and a 30 ms slot, with H-DQPSK outbound and H-CPM inbound. Anritsu’s test description of the inbound burst shows a 12-bit ramp and an 8-bit pilot at each end of a 320-bit information field, and a power envelope with a 28 ms steady period between 1 ms ramps. Whatever the allowance is, it has to live in that structure, and the symbol is 167 µs.

The key sentence for underground work is the one in the DMR standard that says the mobile “shall time synchronize with the BS before transmitting”. The mobile has no clock of its own worth speaking of; it takes the downlink as time zero. Whatever delays the downlink delays the mobile’s clock by the same amount, and the uplink is delayed again on the way back through the same equipment. The round trip is twice the one way delay of the cascade, exactly the arithmetic the standards use for distance. Underground, distance is replaced by equipment.

Where the Delay Comes From Underground

The delay through an underground distribution system comes from four places, and they are not equal.

Radiating cable. The velocity of propagation on RFS RADIAFLEX cable is 88 to 90 percent of free space by datasheet, which for a 7/8 inch RLKU78-50JFNA at 89 percent is 3.75 µs per kilometre. Six kilometres of decline is 22.5 µs one way. This is the number people reach for first, and it is usually the smallest of the four that matters.

Passives. Splitters, taps, couplers, power inserters and terminations add nanoseconds. Ignore them.

Line amplifiers and band-selective repeaters. Group delay is the negative rate of change of phase with frequency through a device, and for a filter it scales with the inverse of the passband width: the narrower and steeper the filter, the longer the delay, and the more the delay rises toward the band edges. The line amplifiers used on Australian mine leaky feeders are band-selective with passbands measured in megahertz. Mine Site Technologies’ VDV amplifier, for example, passes 152 ± 7 MHz on the downlink and 177 ± 9 MHz on the uplink, and PBE’s BDA-4 VHF variants pass 145 to 160 MHz down and 170 to 185 MHz up. Filters that wide have group delay of a small fraction of a microsecond, which is why twelve of them in a decline add less than the cable does. An off-air BDA at the portal that selects a sub-band rather than a channel, which the US public safety market calls a Class B device, is narrower and slower. Comba’s public safety programme manager quotes Class B delays “as low as 6.5 µs”.

Channelised and digital repeaters. This is where the delay lives. A channelised repeater, a Class A device in US terminology, filters each 12.5 kHz channel individually in DSP, and the filter latency is on the datasheet. TX RX Systems publishes it as a table against filter bandwidth, and the same table appears on the Avire Safe-Com SAFE-1030 datasheet.

Two further figures make the same point. A reseller listing for CommScope’s 700/800 MHz Class A public safety BDA quotes a maximum group delay of 90 µs at 12.5 kHz and 3 µs at full band, a factor of thirty between the two filter settings on one product. And Honeywell’s Gamewell-FCI datasheet lists “selectable channel filter latency / group delay per channel” of 12, 20 or 48 µs, with the 12 µs option described as “low processing delay” and the alternatives as “high selectivity”. You buy selectivity with delay, and there is no third option.

Fibre and digital transport. Optical fibre carries a signal at roughly 5 µs per kilometre, and a radio over fibre remote or a digital repeater adds a processing floor on top of its filter, typically tens of microseconds. A remote that serves a second level from the head end via 3 km of fibre has added 15 µs of fibre and its own filter latency before its cable starts.

There is one more thing the DMR standard says that matters here. In defining its own transmit filter, TS 102 361-1 states that “the group delay of the filter is flat over the pass band for |f| < 2 880 Hz”. The modulation is designed on the assumption of flat delay across its occupied bandwidth. Absolute group delay through the distribution system shifts the whole signal in time, which is what the slot budget and the differential delay budget are about. Group delay variation across the channel is a different effect: it delays the edges of the signal relative to the centre and distorts the symbols. A 12.5 kHz channel filter steep enough to give 60 µs at band centre has more than that at ±2.88 kHz, which is exactly where a 4FSK signal at 4800 symbols per second carries its energy, and every filter in the cascade adds its own ripple to the total. That effect applies to every digital modulation, FDMA or TDMA, and it is the second reason not to cascade channelised stages.

Question 1: The Round Trip Against the Slot Budget

Take a mine with a 6 km decline on 7/8 inch radiating cable at 89 percent velocity and twelve wideband line amplifiers at 500 m spacing. The cable is 22.5 µs one way. The amplifiers, at a fraction of a microsecond each, add about 2 µs. That gives a little under 25 µs one way, or about 49 µs round trip, for a radio at the bottom of the decline.

Now vary the head end.

The remote in row C2 is 3 km of fibre at 5 µs per kilometre, a 60 µs channel filter, and 3 km of its own cable with six amplifiers. Row C3 repeats that once more. The budgets are the 1 ms allowance from TS 102 361-1 and the 389 µs delay tolerance from EN 300 392-2 Annex O. Both are round trip figures, and both are shared with the radio’s clock drift, which the DMR standard says has to fit inside the same 1 ms.

Three things follow from the table.

First, for a single-feed leaky feeder with wideband line amplifiers and no repeater at the head end, the round trip is tens of microseconds against a budget of hundreds or a thousand. TDMA is comfortable. Anyone who tells you DMR cannot run on a leaky feeder because of cable delay has not done the arithmetic.

Second, the budget is not consumed by cable. It is consumed by channelised equipment, 120 µs of round trip per stage, and TETRA runs out after three stages. The same equipment on an FDMA system consumes nothing, because there is no budget to consume.

Third, the budget is shared. The DMR standard is explicit that the 1 ms covers propagation and clock drift together, and it works an example in which 135 km of range leaves 0.1 ms for drift. A cascade that uses half the allowance has halved the time a radio can stay keyed before its ±2 ppm clock walks it out of its slot.

Question 2: Two Copies of the Same Signal

The round trip is a TDMA problem. The second question applies to everything digital, and it is the one that actually bites in practice.

Wherever a radio can hear the same carrier by two paths with different delays, the two copies arrive out of step. If they are comparable in strength and the time difference is a large enough fraction of a symbol, the receiver cannot decide where the symbol boundaries are and the audio breaks up regardless of signal level. In the simulcast literature this is called time delay interference, and the same mechanism appears whenever an in-building or underground distribution system re-radiates a signal that is also arriving directly.

Underground, the two copies arise in a handful of predictable places:

• The portal, where a radio hears the surface site directly and also hears the same channel from the leaky feeder, delayed by the head end and the cable.
• Shafts, open pit rims and shallow workings, where surface coverage leaks in.
• Two branches of cable that meet, at a level access, at a loop, or on adjacent turns of a spiral decline, each carrying the signal with a different cumulative delay.
• A leaky feeder plus a repeater chain on the same frequencies, which is the configuration frequency sharing on the DAMM BS422 makes attractive.
• Surface and underground on one channel plan, which is simulcast whether or not anyone calls it that.

How much differential delay a receiver tolerates scales with its symbol period. Nelson and Horden, writing for MissionCritical Communications in 2014, put the commonly used figure for P25 Phase 1 linear simulcast at 60 µs against a 208 µs symbol, worked that as 0.288 of a symbol, and applied the same fraction to the 167 µs symbol of Phase 2 to get 48 µs. They are careful to say that “the exact tolerance for delay spread varies by manufacturer”. Icom’s own page for its IDAS simulcast product says the 6.25 kHz FDMA system with its 420 µs symbol tolerates “about 60 µs” and that “by comparison, the delay spread tolerance of a DMR/P25 Phase 1 system is about 30 µs”. Comba, writing about in-building systems, quotes “a maximum allowable delay between 15 µs for Phase II to 33 µs for Phase I”. And TETRA’s own standard tests its receivers against an equaliser profile with two equal-power paths 11.6 µs apart, which is 0.21 of its 55.6 µs symbol.

The numbers disagree in detail because they come from different vendors and different assumptions, but they agree in shape: the tolerance is a fifth to a third of a symbol, and the symbol is what the access method and the modulation set.

One honest nuance sits in that table. DMR carries its two slots at the same 4800 symbols per second as P25 Phase 1, because it accepts a lower payload per slot rather than a faster symbol. On differential delay alone, DMR and P25 Phase 1 are peers. P25 Phase 2 gave up 20 percent of its symbol to fit two slots at a higher payload, and TETRA gave up most of it to fit four. The FDMA systems most often specified in Australian mines, analogue FM and P25 Phase 1, sit at the tolerant end, and NXDN sits beyond them.

TDMA adds one more constraint to the overlap that FDMA does not have. Two feeds carrying the same TDMA carrier have to agree not only on the symbol clock but on the slot clock. TETRA’s clause 7.7 requires that where neighbouring cells are declared synchronised, “the timing difference between the start of timeslots transmitted by those BSs should be less than 250/9 µs”, about 27.8 µs, and DAMM quotes ±14 µs against the synchronisation source for the BS422. In FDMA simulcast you align the modulation. In TDMA simulcast you align the modulation and the slots, and a cascade whose delay differs between feeds by more than a few tens of microseconds has broken the second alignment even if the first is perfect.

Question 3: Bursts Through Gain Control

The third question is smaller and easier to check. A DMR radio on one slot transmits a 27.5 ms burst and then nothing for 32.5 ms, in every 60 ms frame, with the transmitter power mask in TS 102 361-1 allowing 1.5 ms ramp regions at each end. An FDMA radio transmits continuously for as long as it is keyed.

A line amplifier’s automatic gain control was designed for one of those two signals. The mining line amplifiers in our catalogue reference their AGC to a pilot: PBE’s BDA-4 specifies “AGC with Remote Tone Generator”, Becker’s smartcom RNG-AMP describes its AGC as avoiding return-pilot noise build-up, and MST’s VDV amplifier quotes a 20 dB AGC range. An AGC that watches a pilot tone is looking at the pilot, not the traffic, and does not react to a burst appearing and disappearing on a traffic channel. An AGC that levels on the composite signal will see the burst come and go every 60 ms and do whatever its attack and release time constants tell it to do, which at best means the level control is working against the modulation and at worst means the burst edges are distorted. This is not a difference in the standards. It is a difference in the amplifiers, and the fix is to ask the supplier which kind you have before you commission a TDMA fleet on a cable that was equalised for analogue.

The Surface Is Different

On the surface, all three questions are still there, but the delays come from geometry rather than equipment. The round trip is set by how far the radio is from the site, and the standards’ budgets are generous: 150 km for DMR, 58 km for TETRA. The differential delay comes from simulcast site spacing, and Nelson and Horden’s mitigations for it are all geometric: directional antennas, adjusted ERP, lower antennas, and staggered launch delays that move the interference zone out of the service area.

None of those levers exist underground. The cable goes where the ore body sends it, the branches meet where the levels meet, and the portal is where the surface signal leaks in. The delay is built into the equipment, and the only design choices are which equipment to use and which access method to carry over it. That is why an argument that is finely balanced on the surface, where TDMA’s spectral efficiency is worth real money in licence tax and combiner ports, tilts toward FDMA underground: a mine needs two to four carriers, not twenty, so the spectral saving is small, and the delay budget it gives up is not.

Where TDMA Underground Is Perfectly Fine

It would be wrong to leave the impression that TDMA does not belong underground. The table in Question 1 says the opposite for the common case.

A leaky feeder fed by its own base station at the head end, with wideband line amplifiers, band-selective BDAs at most, and no overlap with surface coverage or between branches, gives DMR a round trip in the tens of microseconds against a 1 ms allowance and no second copy of the signal to argue with. That is a comfortable system, and it describes many operating mines. TETRA on the same cascade uses a larger share of a smaller budget but is still well inside it.

The systems that struggle are the ones with channelised repeaters in series, radio over fibre remotes chained behind other remotes, off-air channelised head ends re-radiating a surface channel into the portal, or a single channel plan across surface and underground without the synchronisation to support it. On those systems FDMA is not a preference. It is the access method that removes two of the three questions and softens the third.

Design Rules

• Know which of the three questions your mine has. A single-feed cable with its own base station has only the round trip to check. A portal repeater or a surface-and-underground channel plan has the differential delay to check, and it is the harder one.
• Add up the cascade delay from the datasheets, not from the cable length. Cable is 3.75 µs per kilometre. A channelised stage is 60 to 90 µs per pass. One stage outweighs 16 km of cable.
• Never put channelised stages in series on a TDMA system without working the budget. DMR tolerates several. TETRA tolerates three at most, and less once clock drift is included.
• Use band-selective or wideband equipment wherever the same channel can be heard twice. At the portal, at a shaft, where branches meet. A 6.5 µs BDA keeps every overlap inside the P25 and DMR tolerances; a 60 µs BDA puts it outside all of them.
• Treat surface and underground on one channel plan as simulcast. It needs a common timing reference, a delay-equalised feed, and for TDMA a slot alignment inside the standard’s figure. If you cannot provide that, use different channels.
• Ask what the AGC references. Pilot-referenced AGC does not care about bursts. Composite-referenced AGC does.
• If the fleet is FDMA, spend the margin on carriers. Every extra carrier on the cable raises the composite power the amplifiers have to carry linearly, and that is the intermodulation and headroom problem that FDMA brings in exchange for the timing margin it gives you.

Frequently Asked Questions

Can DMR run on a leaky feeder? Yes. For a cable fed by its own base station with wideband line amplifiers, the round trip delay is tens of microseconds against a 1 ms inbound timing allowance in the standard. The problems arise with channelised repeaters in the chain, with overlaps where a radio hears the same carrier twice, and with amplifiers whose AGC responds to the bursts.

Why is TETRA more sensitive than DMR underground? Three reasons, all from the standard. Its inbound delay tolerance is 389 µs round trip against DMR’s 1 ms. Its symbol is 55.6 µs against DMR’s 208 µs, so it tolerates less differential delay where two copies of a signal meet. And its receiver performance is specified against an equaliser profile with paths 11.6 µs apart, so anything beyond that is outside the tested condition.

Does cable delay matter? Less than people expect. Radiating cable at 88 to 90 percent velocity is 3.7 to 3.8 µs per kilometre, so even a 10 km run is under 40 µs one way. The equipment in the chain is what consumes a TDMA budget, and a single 12.5 kHz channel filter is worth 16 km of cable.

What is group delay and why does a narrow filter have more of it? Group delay is the time a signal’s envelope takes to pass through a device, defined as the negative rate of change of phase with frequency. A filter that has to roll off steeply over a few kilohertz needs many poles and long time constants, and its delay scales with the inverse of its bandwidth. Published figures run from 60 µs at 12.5 kHz to 8 µs at 500 kHz on the same product, and 3 µs at full band against 90 µs at 12.5 kHz on another.

Is FDMA always the right answer underground? No. FDMA costs carriers, and every carrier adds to the composite power the amplifiers have to handle linearly, which is the intermodulation and headroom constraint of a leaky feeder. Where the fleet is already TDMA and the cascade is simple, TDMA is fine. Where the cascade has channelised stages, overlaps or a shared surface channel plan, FDMA removes two of the three delay questions.

Do these numbers apply to a building DAS? Yes, the mechanism is identical, and most of the published delay figures in this article come from building systems. A DAS that re-radiates a signal also present outside the building is the portal case, and the Class A and Class B latency figures are the same equipment.

Try the Numbers

The Leaky Feeder Designer has a time domain panel that runs this arithmetic on the network you draw. It computes the cable delay from the velocity factor of the selected cable and the length of each span, adds a small fixed allowance for each wideband amplifier and passive, finds every path from the head end to each point on the network, and reports the delay spread between those paths as a percentage of the symbol period for TETRA, P25 Phase 1, P25 Phase 2 or DMR. What it does not do is model channelised or digital repeater stages, whose latency you have to take from the datasheet and add to the result by hand, and it does not check the round trip against the inbound slot budget, which is a division you can do from the figures above. For a building rather than a mine, the In-Building DAS Designer handles the coverage side of the same problem.

Related reading: Designing Leaky Feeder Coverage for Underground Mines for the loss and amplifier spacing design that this article assumes is already done, P25 vs DMR vs TETRA for the carrier and licence arithmetic behind the spectral efficiency argument, and the DAMM BS422 review for the repeater chain alternative to radiating cable and the synchronisation it needs.

Notes on Sources

The timing figures for DMR and TETRA were read from the ETSI documents named below, and the quoted sentences are as they appear in those editions. The P25 Phase 2 structure is from public test equipment application notes because the TIA physical layer document is not freely available; no timing allowance is quoted for it for that reason. Delay tolerances are attributed to their authors because they are engineering judgements that differ between vendors, not values from a standard, with the exception of the TETRA equaliser test profile. Amplifier latencies are from manufacturer datasheets. The worked cascades and the round trip percentages are our own arithmetic from those published figures, and the group delay of a wideband line amplifier is an inference from its passband width rather than a datasheet value.

• ETSI TS 102 361-1 V2.7.1 (2026-05), Electromagnetic compatibility and Radio spectrum Matters (ERM); Digital Mobile Radio (DMR) Systems; Part 1: DMR Air Interface (AI) protocol. Clause 4.2.1 for the continuous outbound and the inbound guard; clause 4.2.2 for the 30 ms slot, the 264-bit 27.5 ms burst and the 2.5 ms inbound guard; clause 10.1.4 for the ±2 ppm clock drift; clause 10.2.1 for 4800 symbols per second; clause 10.2.2.2 for the transmit filter with flat group delay for |f| < 2880 Hz; clause 10.2.3.1.1 for the power mask with 1.5 ms ramp regions; clause 10.2.3.1.3 for the 1 ms slot timing variation allowance, the 150 km range and the 135 km worked example.
• ETSI EN 300 392-2 V3.8.1 (2016-08), Terrestrial Trunked Radio (TETRA); Voice plus Data (V+D); Part 2: Air Interface (AI). Clause 4.5.3 for the 14.167 ms timeslot of 255 symbols at 55.56 µs; table 6.60 for the propagation models including the equaliser test profile with taps at 0, 11.6, 73.2 and 99.3 µs; clause 7.7 for the 250/9 µs base station synchronisation figure; clause 7.8 for the MS timebase adjustment in steps of 125/9 µs; Annex O, clause O.5.1, table O.2 and equations O.1 to O.3 for the 7 symbol delay tolerance and the 58.3 km terrestrial range.
• Viavi Solutions (Aeroflex), Understanding and Testing P25 Phase 2 TDMA, application note, for the 12.5 kHz, 12 000 bit/s, 6000 symbols per second and 30 ms slot figures, the H-DQPSK and H-CPM modulations, and the statement that H-DQPSK was chosen partly for its delay spread characteristics.
• Anritsu, P25 Phase 2 Tx Test Solution MS2830A, product introduction, for the inbound burst structure of 12-bit ramps, 8-bit pilots and a 320-bit information field, and the H-CPM power envelope with a 28 ms steady period between 1 ms ramps, both referenced there to the TIA-102 standard.
• Adam Nelson and Neil Horden, “How Time Delay Interference Affects P25 Coverage”, MissionCritical Communications, April 2014, reprinted by Federal Engineering, for the 208.333 µs and 166.667 µs symbol periods, the 60 µs and 48 µs delay spread figures, the 0.288 symbol fraction, and the mitigation methods for surface simulcast.
• Icom Australia, Simulcast system, product page, for the IDAS figures of 4800 bit/s, 4-level FSK, 6.25 kHz FDMA, a symbol duration of about 420 µs, a delay spread tolerance of about 60 µs, and the comparison to “about 30 µs” for DMR and P25 Phase 1.
• Don Henry, Comba Telecom, “Reducing Time-Delay Interference in Mission-Critical Situations”, Microwaves & RF, and the corresponding Comba article “Mitigating TDI on Public Safety DAS Systems”, for the 15 µs (Phase II) and 33 µs (Phase I) maximum allowable delay figures, the Class A 15 µs and Class B 6.5 µs BDA delays, and the observation that delay comes from active equipment, passives and “especially from filters”.
• TX RX Systems, Public Safety Distributed Antenna System, Bi-directional Amplifier 700/800 MHz and VHF, datasheet DS039623, 13 June 2024, for the Class A filter latency table: 60 µs at 12.5 kHz, 35 µs at 25 kHz, 25 µs at 50 kHz, 15 µs at 75 kHz, 10 µs at 200 kHz and 8 µs at 500 kHz. The same table appears on the Avire Safe-Com SAFE-1030 BDA datasheet.
• Honeywell Gamewell-FCI, Channelized Digital Class A and Class B BDA, datasheet 9021-62042, for the selectable channel filter latency of 12, 20 or 48 µs and the “low processing delay” and “high selectivity” filter options.
• CommScope Public Safety BDA Class A, 700/800 MHz, part 7831758-0021, as listed by a reseller quoting the manufacturer’s specification of a maximum group delay of 90 µs at 12.5 kHz and 3 µs at full band.
• RFS RADIAFLEX radiating cable datasheets, including RLKU78-50JFNA REV P2 (November 2025), for velocity of propagation of 88 to 90 percent across the range.
• Mine Site Technologies VDV Line Amplifier, PBE Group BDA-4 and Becker Mining Systems smartcom RNG-AMP datasheets, as held in the noIM₃ component catalogue, for the passband widths and the AGC descriptions quoted.
• DAMM MultiTech Outdoor Base Station BS422 brochure, version 3.7, for the ±14 µs synchronisation figure, as discussed in our BS422 review.