I'm thinking of buying PowerBeam ac. It's a wireless device to connect two computer networks that are about 25 kilometers apart. But it depended on line-of-sight they say. So is there a cheap way to test for line-of-sight at that distance and frequency? Like a homemade thing or something?

  • 1
    $\begingroup$ Binoculars and a signal mirror work pretty well. $\endgroup$ May 18, 2017 at 15:19
  • $\begingroup$ @PhilFrost-W8II if you look at this image from 20km you would see that you want more than a mirror. commons.wikimedia.org/wiki/… $\endgroup$ May 19, 2017 at 0:36
  • $\begingroup$ @RowanHawkins If you look at this video from 43 miles (70 km) you can see the mirror works great. And that's without the binoculars... youtube.com/watch?v=GwCbgQGmID4 $\endgroup$ May 20, 2017 at 15:49

5 Answers 5


5 GHz and 25 km range is hard.

Let's do a quick free space loss calculation:

$$\begin{align} a &= \left(\frac{4\pi d}{\lambda}\right)^2\\ &= \left(\frac{4\pi dc}{f}\right)^2\\ &= \left(\frac{4\pi \cdot 2.5\cdot 10^4 \text{ m} \cdot 5\cdot 10^9\text{ Hz}}{3\cdot 10^8\frac{\text m}{\text s}}\right)^2\\ &= \left(\frac{5\pi}{3}10^6\right)^2\\ &\approx {5.2}^2\cdot10^{12}\\ &\approx 130\text{ dB} \end{align}$$

130 dB is a lot. Remember, in the ISM band you're normally not allowed to send more than 20 dBm, which leaves you with but -110 dBm at the receiver – not a good thing, considering that thermal noise in a 20 MHz WiFi channel is already $-174 \frac{\text{dBm}}{\text{Hz}}\cdot 20\text{ MHz}= -101\text{ dBm}$, i.e., you'll get ten times the power of the signal in thermal noise.

Hence, your antenna gains must be very high on both receiver and transmitter side, and hence, your equipment must be very well-aligned.

Furthermore, adding to the thermal noise mentioned above, you got the receiver's Noise Figure (NF) – building a low-NF receiver at 5 GHz isn't easy, either.

So, physics and legislation are stacked against you here: If non-LOS communication over microwave over 25 km would easily work, everyone and their mom would run to use it. But then, the spectrum would be crowded – because each station would "spoil" the spectrum for a 25 km radius.

The fact that the higher the frequency, the higher the path loss, is the only reason we actually have these large ISM bands in the 2.4 and 5 GHz region.

Directional links in the Gigahertzes are thoroughly common; look for the horizontally pointing dishes (typically encased in white plastic) on larger cell phone towers. However, these are usually run with an explicit license for stationary earth-earth directional links, and hence legally do much more power than your AC router. Now, you could possibly build a similar link yourself with off-the-shelf satellite dishes, but I've got no experience nor appropriate license knowledge on doing so on ham bands.

So, coming back to your original question:

so is there a cheap way to test for line-of-sight at that distance and frequency?

well, line-of-sight can be tested by sight! So, good binoculars are your weapon of choice.

Now, I feel like I should also point you to Fresnel zones: it's recommendable to not only have clean LOS, but also a clean first Fresnel zone, to avoid diffracting/reflecting your microwave beam. In other words: this is something to be mounted high if you want to go the full distance!

Aligning those dishes won't be easy, but maybe they come with some straight piece that points exactly parallel to the beam direction, on which you can mount a laser pointer, and tell a friend over the phone to turn the dish until you can see the laser pointer pointing at you.

Another thing to consider: WiFi was never meant to be used over huge distances. It's a carrier-sense system, ie. a station must listen for a predefined time, before it considers the channel free and may device send. That way, the chance of packets colliding in the air is reduced significantly enough to make such systems work without a central instance to coordinate spectrum access in fixed time slots.

This is achieved by first sensing, then sending a short "Ready To Send" packet containing the coming packet's length, which the intended receiver detects, and then again waits for a fixed period, to make sure no one missed the original RTS, before replying with a "Clear To Send". The idea is that all other stations hear at least either the CTS or RTS and know that they need to stay silent for the specified amount of time.

However, if the CTS->RTS mechanism, being completely superfluous for a fixed length, takes 2 times the sensing period in addition to 160 µs of signal flight, the consequences for throughput are dire, because the channel will be used nearly only for waiting and sensing – a full IEEE802.11n symbol is about 4µs the most, and there's about 30 of those in a frame; though a couple of frames can be aggregated under 802.11n, the relation of flight time to active data transmission is unnecessarily bad:

If your wave has to travel 25 km, it'll be in flight for $\frac{d}{c}= \frac{2.5\cdot 10^4\text{ m}}{3\cdot 10^8\frac{\text m}{\text s}} \approx 80\, \mu\text{s}$.

In this scenario, having a system with separate up- and downlink frequencies (FDD), halving the available bandwidth, is significantly more efficient.

  • $\begingroup$ Good points, I added an config example for RTS/CTS control which may be useful $\endgroup$ Jun 27, 2016 at 11:52
  • $\begingroup$ I think that particular device, when running in point-to-point mode, alters the standard 802.11ac protocol to be more amenable to high latency. $\endgroup$ May 17, 2017 at 11:19

One way to estimate line-of-sight before you even step outside is to calculate the radio horizon. For the horizon distance in miles and antenna height in feet, the formula is

$$ \text{horizon} = 1.23 \sqrt{\text{height}} $$

For kilometers and meters, it is:

$$ \text{horizon} = 3.57 \sqrt{\text{height}} $$

The horizons of each station add. If you need a link distance of 25km, then you might aim for a horizon of 12.5km on each side. That would require an antenna at height:

$$ \left(12.5 \over 3.57\right)^2 = 12.26\:\mathrm m $$

This is an approximation based on a smooth Earth with no terrain. If a station is on a hill, add that to the height. If there's a hill between the two stations, that may be a problem.

If you want a quick and easy real-world check, a signal mirror and binoculars work well. A true optical line of sight isn't strictly necessary for a working RF link, but it's nice to have if the objective is a high quality, reliable path.

There are other complicating factors which must ultimately be considered, such as atmospheric effects, Fresnel zones, interference, and other practical issues other answers have mentioned. However, if this estimation shows the radio horizon is grossly insufficient, there's little reason to continue your endeavor unless you can secure taller towers or better sites.


To add to the already given answer.

Most countries have a 100mW / 20dBm ERP maximum for license free usage of those bands. (You will have to check if this is your case, but chances are)

As this is usually expressed in "ERP" which is "Effective Radiated Power" it will take antenna gains into account. Example by using an antenna with 3dBi gain, you will have to compensate your TX-output power to 50mW, to ensure you comply with legislation. This is probably the single most limiting aspect of what you are asking.

However (!!) If you can design your system, where you stay within your 100mW ERP for TX, but you somehow can improve your RX by means of an antenna with extreme directional gain, you may have a chance.

For this to be successful, you need to use devices with separate TX antenna and RX antenna connections, or use of a "Circular Coaxial Coupler" (for the bands you intend to use)

Lets use a calculation from another answer already given:

  • So we know that path-loss is about 130 dB.
  • We have seen the noise figure at -101 dBm.
  • And we take a maximum of 20 dBm.

But now we take the RX antenna at 40 dBi (huge parabolic dish).

20(dBm) - 130(dB) + 40(dBi) = -70dBm for RX with a -101 dBm noise figure.

Theoretically this should be possible, but you will have to precisely line up the parabolic dishes, you need to have the correct equipment, and-so-on-so-forth.

It would be a nice experiment if you have the equipment laying around... If you have to buy it all from scratch, then my practical advice is: 25 km is just a little to far for bridging networks with "consumer" equipment and staying within the confines of the law.

[EDIT] based on the answer already given: regards "Timing"

Some equipment may have settings to further control the RTS/CTS and other parameters of the protocol, to enable longer distances.

Here a typical screenshot of an EnGenius device:

EnGenius advanced settings

There are various settings which would help longer distances:

"Data Rate" - the lower the data rate the longer the distance, experimentation needed

"Transmit Power" - the higher the Transmit Power the longer the distance, this should be as high as (legally) allowed (taking antenna gain, coax/feeder loss, into consideration)

"RTS/CTS Threshold" - the lower the threshold the longer the distance, at a cost of bandwidth, experimentation needed: set as high as possible without collisions. This might require monitoring of re-transmits

"Distance" - this should be set as accuarately as possible, as according to this document it affects the ACK timeout on longer links.

"Short GI" - should be disabled, this affects the timing between "characters" on the link, and should be set to standard/disabled (timing 800ns) which prevents/avoids collisions within the same character due to reflections. The longer the distance the more risk of reflected signals you have

"Aggregation" - this should probably be disabled, but experimentation may be needed.

Obligatory disclaimers:

  • other brands may call settings differently
  • EnGenius has different models, not all of them available to consumers
  • I have no affiliation with EnGenius, this is only an example
  • and... ofcourse... YMMV
  • $\begingroup$ good points! Another aspect is timing... added that to my answer. $\endgroup$ Jun 24, 2016 at 16:31

If you want to calculate the height required for a 25 km radio link. Check out this link.


This link gives not visual line of sight but also the radio horizon which is slightly longer. At 5 Ghz with sufficient gain antenna to direct the signal as close to the horizon as possible and height above ground you should be able to establish a good strong link. This should be at both sites. At 5 Ghz the radio noise and interference would be quite low which is a bonus.

40 meters above ground should do it. If you have an unobstructed point to point link you could probably do it on relatively low power.


25KM is a really long distance. You need to be up fairly high to beat the curve of the earth. If you are at sea level on a beach, and your eyes are 1.7m above the ground, the horizon would be at 4.7km. Source wikipedia:Horizon

In actuality both ends would need to be elevated so that the beam path passes high enough over terrain between the 2 points.

A cheap way to test line of sight is binoculars and a really large white or Black sheet depending on the color of the surrounding terrain. A signal mirror would need to be aimed and it would need to be huge to be really visible. If you want to use a light, do it at night, and get one of the aircraft approved Xenon strobes.

You wouldn't want to use a laser with binoculars for 2 reasons. 1) coliniated light can damage your retina when intensified by a lens. 2) the divergence of a non-commercial laser point over 25 Km probably wont be visible anyway.

A bunch of things which other answers have hit on are various type approval issues with using this for WIFI. If anyone had looked at the link, they would see that the PowerBeamAC is not working under the 802.11ac specification. It has a separate 802.11 antenna for configuration, but the 25km, link is advertised using a supposedly proprietary TDMA signalling for point to point. and is advertising 450Mbit/s bandwidth. It is separately type approved by the CE, FCC and IC. All of this is in the DATASHEET


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .