698-960MHz vs 3300-3800MHz 5G Antennas: Choosing the Right Band for Rural vs Urban
Every few weeks an operator or integrator asks us the same question in different words: should this site be low band or mid band? The answer is almost always "both, eventually" — but the order you deploy them in, and the antenna you choose for each, decides whether the site pays for itself in two years or five.
We build both families in our factory in Zhaoqing, so this is not a sales pitch for one side. It is the comparison we walk through with customers, with real numbers from link budgets we have run.
What 698-960MHz buys you: Distance
A low band 5G antenna working between 698 and 960MHz covers the LTE B28/B20/B8 bands and their 5G NR equivalents, n28 and n20 among them. The physics is simple and unforgiving: lower frequency means lower path loss, less foliage attenuation, and better building penetration.
The rough field numbers we work with: a 700MHz sector on 30m tower reaches 5 to 8 km in flat rural terrain, sometimes more. The same site at 3.5GHz manages 0.8 to 1.5km. That is not a 2x difference — it is a 5x difference in covered area per site, which multiplies out when you are blanketing farmland, highway corridors, or island communities.
But low band exacts its price at the antenna. A half-wave dipole at 700MHz is about 21cm; at 3.5GHz it is 4cm. That is why a 698-960MHz antenna element is physically large, why the panels are longer, and why element-level gain tops out around 9 to 11dBi for a practical sector panel. Push a low band 5G antenna design beyond that and you fight sidelobes that waste energy above the horizon — energy a rural site cannot afford to lose.
Bandwidth is the second constraint. Squeezing 698 to 960MHz into one radiating structure is a 37% relative bandwidth. We handle it with log-periodic or coupled dipole arrangements, and the honest trade-off is that return loss at the band edges (698MHz, 960MHz) is always the hardest spec to hold. Our production units hold VSWR under 1.5 across the full range, but margin at 960MHz is thinner than at 800MHz. Every designer who tells you otherwise is rounding in their favor.
What 3300-3800MHz buys you: Capacity
The n78 band, 3300 to 3800MHz, is where most of the world's actual 5G traffic lives. It is not because the band is special — it is because 500MHz of spectrum supports the channel widths that make 5G worth building.
An n78 antenna is a different animal entirely. Element size shrinks, so a 0.3 x 1.2m panel holds 192 dual-polarized elements instead of a few dozen. That is what makes 32T32R and 64T64R beamforming possible: the antenna steers energy per user, which is where the capacity comes from. A well-designed 3300-3800MHz array reaches 24 to 25dBi with beamforming gain included, against 15 to 17dBi passive.
The penalty is reach and penetration. 3.5GHz loses 15 to 20dB more through a concrete wall than 700MHz. Indoors, deep inside a shopping mall or a multi-storey car park, an n78 signal from a macro site is often gone. This is why urban 5G is a densification story: small cells, indoor panels, one sector per few blocks rather than per district.
PIM discipline also tightens. Our low band panels ship with PIM around -155dBc; for 3300-3800MHz units connected to 64T radios, we hold -160dBc or better, because at these power levels and channel widths, even minor intermodulation shows up as throughput noise. Connectors, jumper quality, and installation torque matter as much as the design itself — we have seen more field PIM problems from a loose DIN connector than from any PCB we ever shipped.
How the choice actually gets made
In practice, the decision tree looks like this:
Rural, low density, wide area: low band carries the rural 5G coverage layer. One 698-960MHz panel per sector, 2T or 4T, and the site covers a village cluster. Add mid band later at the same site when traffic justifies it.
Urban core, high density: n78 is the capacity layer. 64T64R arrays on rooftops, small cells below for the streets the macros cannot reach. Low band here is only a fallback layer for deep indoor.
Suburban and highway corridors: this is the messy middle, and honestly where most of our multiband antenna volume comes from. A quad-band panel covering 698-960 and 3300-3800MHz in one radome saves a second mast position, a second set of feeders, and a second round of structural analysis.
That last point deserves emphasis. Tower space rental is frequently the largest recurring cost after power. A multiband antenna that merges both bands into a shared aperture is not an engineering luxury — it is the difference between upgrading a site and being denied the lease for a second antenna position.
Where we admit the limits
The coverage figures above assume flat rural terrain and standard propagation models. Hilly terrain, coastal humidity, and dense tropical foliage all cut low band range; we have seen 700MHz sectors in mountainous provinces behave closer to the 3.5GHz numbers than anyone planned for. Run your own link budget before ordering. We will supply the pattern files and the tilt tables, but we cannot see your terrain from here.
Also worth saying: "low band for rural, mid band for urban" is a starting rule, not a law. Some of our best rural deployments are n78 fixed wireless access to a village with fiber backhaul, because one high-capacity sector served 200 households better than a thin low band layer ever could.
One band is not a strategy
The operators doing well in 2026 stopped treating this as a choice years ago. Low band 5G antenna coverage gets the network everywhere; the n78 layer makes it profitable where the users are. Your antenna supplier's job is to make both layers — and ideally one radome that carries both — with specs that hold in the field, not just on a datasheet.
If you are planning a deployment and want a second opinion on band choice or antenna selection, send us your site parameters. We answer with numbers, not adjectives.
TAG：698-960MHz vs 3300-3800MHz 5G Antennas: Choosing the Right Band for Rural vs Urban https://www.rfelement.com
