Part 7 of The Analog Edge, a 14-part field guide to the analog half of a GopherTrunk installation. Part 6 finished the desk work: our reader with the marginal system has now staged gain, ruled out overload, checked the oscillator, and settled the rate — all without touching hardware. What’s left is the hardware, starting where the signal does. The antenna is the one component that can *add signal-to-noise ratio instead of merely preserving it, which is why it’s the first dollar this series tells anyone to spend. It is also — on the evidence of the project’s own weak captures — the dollar most often left unspent.*
TL;DR: Trunked systems live in a handful of bands — VHF-high, UHF, and the 700/800/900 MHz cluster — and an antenna is only “good” at a frequency, in a direction. Gain is a shape, not a magnitude: a high-gain vertical buys its dB by flattening its pattern toward the horizon, which is exactly wrong for a close, elevated tower. Match vertical polarization (land-mobile is vertical; cross-polarization costs real dB), put height before gain until feedline loss eats the difference (Part 8), and choose by situation: discone to survey everything, tuned vertical to monitor one band well, yagi to reach one distant system. The tracker’s marginal captures — the −44 dBFS TETRA sessions, the −75 dBFS DMR file no decoder could use — are what under-antennaed systems look like from the samples’ side.
Key takeaways
- The antenna is the only free SNR in the chain. Every later stage — LNA included — adds its own noise while amplifying; a better antenna delivers more signal and no more noise. Nothing downstream can match that trade.
- Gain redistributes; it doesn’t create. An antenna’s dB come from squeezing its radiation pattern toward the horizon. Whether that helps depends entirely on where your towers are — including the close one a flattened pattern can miss.
- Polarization is a silent tax. Land-mobile trunking is vertically polarized; a horizontally-mounted element or a random indoor wire pays a large cross-polarization penalty before any other factor is counted.
- Buy for the band you actually monitor. A wideband discone hears everything adequately; a band-tuned vertical hears your system well. Survey first, then specialize — the two-antenna strategy this series’ diversity parts (11–13) will formalize.
Cheat sheet
| Concern | The short answer | Where to go deeper |
|---|---|---|
| Which antenna to buy | discone to survey, tuned vertical to camp, yagi to reach | Best scanner antennas, best SDR antennas |
| What “dBi/dBd” mean | gain relative to isotropic / dipole (dipole = 2.15 dBi) | Antenna gain |
| Wideband survey antenna | the classic broadband vertical monitor antenna | Discone |
| High-gain vertical mechanics | stacked elements, flattened elevation pattern | Collinear |
| One distant system | directional gain toward a single azimuth | Yagi-Uda |
| Mounting, masts, grounding | doing the height part safely and legally | Antenna mast & mounting guide |
| Whether it worked | level and decode metrics, before vs after | Parts 2–3 of this series — re-run the same measurements |
In this post
- The bands trunking lives in — where to aim the whole exercise.
- Gain is a shape, not a magnitude — patterns, and the close-tower trap.
- Polarization — the cheapest dB you’ll ever recover.
- Height beats gain — until it doesn’t — the feedline caveat.
- Choosing: discone, tuned vertical, or yagi — a decision table.
- What the tracker’s weak captures teach — the evidence from our side.
The bands trunking lives in
Antennas are frequency-selective, so the first question is never “which antenna?” but “which frequencies?” — and trunked radio concentrates into a few well-known neighborhoods (US-centric here; check your region’s band plan):
| Band | Rough range | Who you’ll find there |
|---|---|---|
| VHF-high | 150–174 MHz | statewide/rural public safety, utilities |
| UHF | 450–512 MHz | municipal systems, businesses; DMR and NXDN country |
| 700 MHz | 769–806 MHz | modern P25 public-safety systems |
| 800 MHz | 851–869 MHz | the classic trunking band — P25, EDACS, Motorola legacies |
| 900 MHz | 935–940 MHz | business/industrial trunking, DMR |
Look up your local systems before buying anything: a region whose activity sits entirely at 700/800 MHz wants a very different antenna than one on VHF. The wavelength spread is the point — a quarter-wave element is ~48 cm at 155 MHz and ~9 cm at 860 MHz, and no single passive element is optimal across that ratio. Wideband antennas exist by accepting compromise everywhere; tuned antennas excel by refusing it in one place. (For the propagation background — why 800 MHz is line-of-sight-ish while VHF bends farther — the propagation lesson is the companion read.)
Gain is a shape, not a magnitude
Antenna gain is the most mis-sold number in radio. An antenna is passive — it cannot amplify. A “6 dB gain” vertical delivers more signal from the horizon by taking it from everywhere else: picture the radiation pattern as a donut around the element that higher gain squashes flatter and wider. Those dB are real if your towers sit near the horizon. They are negative if a tower is close and elevated — a downtown high-rise site a kilometer away can sit meaningfully above a flattened pattern’s main lobe, and operators genuinely see a cheap unity-gain whip beat an expensive high-gain collinear on exactly that one system. The antenna gain entry covers the units (dBi vs dBd — a half-wave dipole is the 2.15 dBi reference); the operational rule is simpler: know your towers’ directions and elevations before buying gain, because gain is a bet on where the signal comes from.
Polarization
Land-mobile radio is vertically polarized — mobile whips point up, so the infrastructure does too. Receive with a matched vertical element and you collect the full field; turn the element horizontal and, in the textbook case, the cross-polarization loss is severe (tens of dB in free space; scattering in real environments softens it to “merely large”). This is the cheapest audit in the series: telescopic antennas angled for looks, mag-mounts on their side on a windowsill, random lengths of wire draped where convenient — each is quietly paying a polarization tax on every tower at once. Stand the element vertical, with a decent ground plane under a mag-mount (a cookie sheet genuinely works), before spending anything. The polarization entry has the theory; the audit takes thirty seconds.
Height beats gain — until it doesn’t
At trunking frequencies, propagation is dominated by what stands between you and the tower. Height fixes obstruction directly: getting an antenna above the roofline changes the path, which routinely dwarfs any achievable pattern gain — going from an indoor desk antenna to a modest outdoor mount is frequently a 10–20 dB swing at UHF and up, whereas the best consumer verticals advertise 6–9 dB against a reference dipole. If you can only change one thing about a marginal installation, raise the antenna. Indoors, even a window facing the tower instead of an interior room can be the difference the decoder needed.
The caveat that pairs with this part: every meter of height is a meter of coax, and at 800 MHz cheap coax charges by the meter — enough that a long run of the wrong cable can spend the entire height dividend before it reaches the tuner. That arithmetic (loss per cable class, and why receive loss before the first amplifier is especially costly) is precisely Part 8, so plan mast and feedline as one decision — the mast & mounting guide covers the mechanical and safety side.
Choosing: discone, tuned vertical, or yagi
For a fixed monitoring installation, three archetypes cover nearly every case:
| Discone | Tuned vertical (¼-wave / collinear) | Yagi | |
|---|---|---|---|
| Bandwidth | very wide (an octave-plus) | one band, done well | narrow, one band |
| Gain | ~unity | unity (¼-wave) to ~6–9 dBd (collinear) | high, in one direction |
| Pattern | omnidirectional | omnidirectional (flatter with gain) | one azimuth lobe |
| Best for | surveying, hunting, many bands at once | camping on your area’s main band | one distant/weak system |
| Watch out | mediocre everywhere by design | close-elevated-tower trap above | must be aimed; everything else fades |
The strategy that falls out: survey wide, then camp tuned. A discone plus the hunt tells you what’s actually receivable at your site; once you know which system you care about, a vertical tuned to its band converts that discovery into decode margin. The yagi is the specialist’s move — one distant system, direction known, everything else sacrificed. If you run more than one dongle, run more than one antenna philosophy at once (the multi-dongle guide covers the plumbing): discone on the hunting dongle, tuned vertical on the production system. Specific current models are kept in the scanner antenna and SDR antenna guides rather than here, so this post can age gracefully.
What the tracker’s weak captures teach
This series keeps its claims tied to the project’s evidence, so: what does an under-antennaed system look like from the samples’ side? Like the captures our decoders keep meeting. The TETRA control-channel sync-loss investigation ran on captures peaking around −44 dBFS — weak enough that the marginal-SNR regime caused hard sync losses, and the fix notes say it plainly: the equalizer mitigates a weak front end but the residual condition is RF/gain/antenna — raise the signal level too. The DMR two-slot verification from Part 1 is still parked on a −75 dBFS capture with no recoverable frame sync at all. In both cases sophisticated software (equalizers, soft decision) recovered real margin, and in both cases the notes end by pointing back across Part 1’s line. The decoder can only be as good as the samples — and the antenna is where the samples are born. The antennas lesson makes a good deeper companion to this whole part.
Where this goes next
A better antenna earns you dB at the top of the mast; the feedline decides how many survive the trip down. Part 8 follows the coax — loss per hundred feet by cable class at 800 MHz, why every adapter is a small tax, the SMA/BNC/N/F connector ecosystems, and why a dB lost before the first amplifier is the most expensive dB in the whole chain.
FAQ
Will a better antenna fix my garbled audio? If the problem is signal-starvation, it’s the single most effective fix available — and Parts 2–3 give you the before/after instruments to prove it (level regime, decode error rate, demod SNR). If your levels are already healthy and the decode error rate is low, your problem is elsewhere in this series, and a new antenna will disappoint. Measure first.
Is the antenna that came with my dongle good enough? For strong local systems, sometimes. Structurally it’s a short, often poorly-grounded whip at desk height — the polarization, ground-plane, and height mistakes bundled together. It’s fine for proving the software works; treat it as the baseline the real antenna gets measured against.
Can one antenna cover VHF and 800 MHz? A discone genuinely can, at the price of being merely adequate at both — that’s its design contract. What can’t cover both well is a tuned antenna: a vertical cut for 155 MHz is far off-resonance at 860 MHz. If you monitor two distant bands seriously, that’s two antennas (and ideally two dongles) — not one compromise.
Do amplified (“active”) antennas help? They move Part 9’s problem into the antenna housing: an amplifier helps only if it comes before significant loss and after adequate selectivity, and a wideband amp at the antenna also amplifies every pager and broadcast blaster in view (Part 4). Passive antenna + separate, chosen LNA where the math says it belongs (Part 9’s whole subject — funnel: best SDR LNAs) beats an opaque bundled amp you can’t reason about.
How do I verify a new antenna actually improved things?
Re-run the Part 2/Part 3 measurements you already know: same channel, same
gain staging procedure, compare iq_power_dbfs regime, clip ratio, decode
error rate at the chosen rung, and demod SNR on a replayed capture. A real
upgrade shows up as decode-quality improvement at equal or lower gain —
if you had to raise gain to see a difference, you measured the knob, not
the antenna.
Series navigation
Part 7 of 14 · ← Part 6: Sample Rate — The Decode Path Doesn’t Care; the Front End Does · Next → Part 8: Feedline & Connectors — Where dB Go to Die