Most point to point radios let you set the channel to 20, 40 or 80 MHz, and some go wider. The widest setting looks like free speed, but channel width for point to point link planning changes far more than throughput: it moves the noise floor, sensitivity, fade margin and interference. Here is what each step costs and how to choose.
What channel width should I use for a point to point link?
Use the widest channel your fade margin and the spectrum allow. Each doubling, from 20 to 40 to 80 MHz, roughly doubles throughput but raises the noise floor about 3 dB, so sensitivity and fade margin drop by 3 dB. Long, noisy or marginal links are often more stable on 20 or 40 MHz.
What channel width does on a wireless bridge
Channel width is the slice of spectrum the radio spreads its signal across. Modern 5 GHz radios split it into many narrow OFDM subcarriers; a wider channel holds more of them, so at the same modulation it carries more data.
The 802.11ac figures show how directly this scales. A 20 MHz channel has 52 data subcarriers, 40 MHz has 108, 80 MHz has 234 and 160 MHz has 468. Going from 20 to 80 MHz multiplies the data subcarriers by 4.5, slightly better than the fourfold increase in width.
That is only half of the trade. The other half is the link budget, which our guide to planning a point to point wireless link builds step by step.
Why each doubling costs about 3 dB
A receiver hears thermal noise across the whole channel it listens to. At room temperature that noise is about -174 dBm in each hertz of bandwidth, and the total in a channel is -174 + 10 log10(bandwidth), plus the receiver’s own noise figure. Doubling the bandwidth doubles the noise power, and 10 log10(2) is 3 dB.
From that formula, before any noise figure is added, the thermal floor is about:
- 20 MHz: -101 dBm
- 40 MHz: -98 dBm
- 80 MHz: -95 dBm
- 160 MHz: -92 dBm
The signal has not changed, so every 3 dB the floor rises comes straight out of the signal to noise ratio. The radio needs a stronger signal for the same modulation, or drops to a slower one.
20 vs 40 vs 80 MHz channel compared
The IEEE 802.11 receiver specification sets a minimum input sensitivity for each modulation and width, measured at under 10 percent packet error. The table below takes figures from the IEEE 802.11 working group’s receiver minimum input level sensitivity table; every row steps by exactly 3 dB per doubling.
| Width | Thermal floor | Data subcarriers | BPSK 1/2 | 256-QAM 5/6 |
|---|---|---|---|---|
| 20 MHz | -101 dBm | 52 | -82 dBm | -57 dBm |
| 40 MHz | -98 dBm | 108 | -79 dBm | -54 dBm |
| 80 MHz | -95 dBm | 234 | -76 dBm | -51 dBm |
| 160 MHz | -92 dBm | 468 | -73 dBm | -48 dBm |
These are the minimums the standard requires. Real radios usually do better and publish their own table for each width and modulation: plan with that data sheet.
Fade margin, modulation and link distance
Fade margin is the gap between the signal you expect at the receiver and the sensitivity the data sheet lists for the modulation you need. Rain, multipath and seasonal changes eat into it.
Here is how width moves that margin. Assume a link budget that predicts -55 dBm at the receiver, an assumed figure, and use the standard’s minimums from the IEEE table:
- 80 MHz: 256-QAM 5/6 needs -51 dBm, so it is 4 dB short. 64-QAM 5/6 needs -58 dBm, leaving 3 dB of margin.
- 40 MHz: 256-QAM 5/6 needs -54 dBm, leaving 1 dB. 64-QAM 5/6 needs -61 dBm, leaving 6 dB.
- 20 MHz: 256-QAM 5/6 needs -57 dBm, leaving 2 dB. 64-QAM 5/6 needs -64 dBm, leaving 9 dB.
The wider channel still carries more data, but on a thinner margin. As the path gets longer the received level falls, and the extra 3 or 6 dB of a narrower channel becomes the difference between a link that holds and one that keeps renegotiating. Gaining that margin back with hardware means more antenna gain, covered in antenna gain in dBi explained, or a larger aperture, compared in dish vs panel antenna for wireless links.
Interference and spectrum use
A receiver listens to everything inside its channel, not only your far-end radio. An 80 MHz channel covers four times the spectrum of a 20 MHz one, so it has four times the chance of overlapping a neighbouring access point, another bridge or a sector on a nearby tower. A single strong interferer anywhere in that span degrades the whole channel.
Width also limits reuse. Each doubling halves the number of non-overlapping channels in the band, which matters on a tower carrying several links. When the band is crowded and the hop is short, a higher band may suit better; see 5 GHz vs 60 GHz wireless bridge. Where interference cannot be tolerated at all, licensed vs unlicensed backhaul covers the protected options.
DFS channels
Parts of the 5 GHz band are shared with radar, and radios there must use dynamic frequency selection (DFS). An NTIA technical report describes DFS as technology that is supposed to detect nearby radar systems and change operating frequencies, and documents interference to Terminal Doppler Weather Radars, which operate in the 5600 to 5650 MHz band. A wide channel is more likely to span DFS spectrum, and a radar detection moves the whole link, so plan a fallback channel.
How to choose a channel width for a point to point link
- Size the throughput you need. Total the traffic the link will carry, with headroom, before choosing a width.
- Run the link budget. Predict the received level from the path, antenna gain and transmit power.
- Read the data sheet table. Find, for each width, the fastest modulation that still leaves the fade margin your path and climate require.
- Survey the spectrum at both ends. Scan from each radio, at mounting height, and note which channels are clear at each width.
- Pick the narrowest width that meets the throughput. Spare capacity on a wide channel is worth less than margin you will need on a wet day.
- Confirm after installation. Watch the received level, modulation and error counters for a few days.
When narrowing the channel fixes an unstable link
Symptoms of a link running too wide include modulation that keeps stepping up and down, rising latency, retransmissions, and throughput that falls well short of the configured rate. Installer forums report long links on crowded channels that gained no speed and lost stability when widened.
Before changing hardware, try one step narrower. The 3 dB regained and the smaller slice of spectrum often let the radio hold a steady modulation. If it still drops, check the path, alignment and interferers next.
Where to find point to point radios and antennas
The point to point and wireless radios category and the wireless network antenna category sit in the Broadband and WiFi division. For pricing, data sheets and availability, ask your sales representative through the Bursma contact page, call the Columbus, Ohio low voltage branch, or open a Bursma dealer account.
Frequently asked questions
Does a wider channel reduce range?
Yes, in effect. Each doubling raises the noise floor about 3 dB, so the radio needs a stronger signal for the same modulation, which shortens the distance at which a given speed holds.
Is 80 MHz good for a long point to point link?
Only if the link budget leaves enough fade margin at 80 MHz and the spectrum is clear at both ends. On long or crowded paths, 20 or 40 MHz is often more stable.
Why is my wireless bridge throughput lower than the channel rate?
The configured rate is a maximum. Interference, a weak signal, a falling modulation and protocol overhead all cut real throughput.
Should both ends of the link use the same channel width?
Yes. Both radios must be set to the same channel and width to form the link, so change them together, far end first if you are working remotely.