130 km link.!

[gerencia@e-digitales.com]

is me again..
i´m interested in make a long distance link for about 130 km with clear LOS and between 2 hills 1800m point1 and 2900m point2 (Bogota-Ibague cOLOMBIA). I saw Orthogon radios but are very expensive .

In the V3 page says that max distance are about 230km, if i use 400 mw ubiquiti sr5 and 4´ (1.20m or homeMade) dish antennas with V3 OS, will be possible 130km link with at least 24mbps?

Thanks

dimas perez

[ninedd]

In the V3 page says that max distance are about 230km, if i use 400 mw ubiquiti sr5 and 4´ (1.20m or homeMade) dish antennas with V3 OS, will be possible 130km link with at least 24mbps?
Yes, but you might not even need to run those cards - the math says you can probably use CM9's, and take advantage of thier frame / compression / burst features for even more speed. We've seen CM9's at 11Mbit outperform other cards linked at 36 Mbit.

At 130Km, the free space loss is about 150 dB at 5.8 Ghz. With 17 dBm radios, and 1 Meter (32 dB gain) dish's, your should still get around -69 dBm, which should be fine to make a good link.

At 2.4 Ghz (if you can spare the channel space and have little interference) the free space loss is only 143 dB, and with the same radios and even 30 dB grids, your signal could be around -65, which would be very solid.

Of course, going up to the 400mw cards should add some margin, but that's assuming they're as sensitive as the CM9's. As well, the 400Mw cards don't do 400Mw at 54 Mbit, so there's a trade-off between power and link rate as well.

I'd recommend using the calculations to try various options. The longest link we do is 45Km, and StarOS came in exactly what the calculators told us to expect, and our CM9 to CM9 link ($55 radios) with 24 dB grids ($50 antennas) has been rocking for over a year now.

[gerencia@e-digitales.com]

very clear, thankyou

[George]

The only thing you are really going to have to watch out for is required fade margin.

To run a 50+km north/south link in my area we need more than 50dB of fade margin at dawn and dusk in late Spring and early Autumn. The rest of the year everything runs just fine.

To get rock solid performance we need space, polarity and time diversity and the only license exempt radios that can provide that today are Orthogon Geminis and Spectas. Orthogon also happens to have a link calculator that takes into account your geographical location and the relevant fade characteristics of that location.

I shudder to think what we would need to keep a 130km link running during these times of year, in fact I simply wouldn't try it.

For peace of mind, seek out one of the licensed microwave guys that run long-haul 6GHz links in your area. They will know all the ins and out of the situation and be able to give you some really good advice on what is required.

I have hopes that MIMO will improve our situation moving forward, but that remains to be demonstrated in the field.

All the best,
George

[ninedd]

To run a 50+km north/south link in my area we need more than 50dB of fade margin at dawn and dusk in late Spring and early Autumn.Hi George. What causes this for you? Are you saying you need to have 50 dB extra SNR because it'll fade that badly at certain times?

In our case, we have a link nearly that long - a 45Km link - and we use standard CM9's and standard cheap 24 dBi grids. This is a remote area, so we have the channel space to use 2.4Ghz, which may not hold up forever, but hasn't been an issue for the last 14 months or so. The math says we should be around -72 RSSI and that's exactly what we see, day and night, summer and winter, spring and fall, rain or shine, snow or sleet. According to the calculators, 50Km should only be about 1 dB difference. What causes your fade Spring and Fall?

[George]

Yup, even with around 50dB we will lose three of these links in late May or early June and again during Indian Summer in the fall. This area is in the textbooks for some of the most extreme fade conditions found in North America.

What happens is really severe ducting caused by cold air/warm water/hot exposed rock or vice versa depending on the season. Either way, it sets up some really nasty thermal effects.

This does two things. First, it ducts the signal away from the antennas its intended for. Second, it ducts 5.8 noise north from Toronto to dramatically raise our normally silent background noise levels.

We can watch the effect in real time on the Orthogon spectrum analyzers.

We see see the radios struggling to maintain adequate signal to noise by ramping up power to the 25dB max while receive signal levels drop below -95. (For comparison, normal power levels while everything is quiet are only +2dB).

At the same time the noise floor rises from -100 to -75 across the band. This varies on both a slow cycle (five minute or so quasi sine wave) and fast cycle (several times per second).

Enough to cause problems for about four hours from around 5am to 9am on really bad days. Sunny and calm with cold nights and hot days are the worst conditions. The wind is our friend as it breaks up the ducting.

The Orthogons can cope with a 50km link if you use a four foot dual polarity dish on one end talking to a pair of single polarity four footers in a space/polarity diversity configuration on the other end, utilizing around 10M vertical between them to create a truly diverse path profile.

To give you an idea how bad this can get, the 6GHz guys I work with on the same towers use waveguide, 12' dishes and 32dB radios with 20M of space diversity. They still can't maintain the 99.9997% required on the southern-most pair of hops even with this much horsepower at their disposal. And they don't have the same problem of a rising noise floor as their licensed band is quiet even with the ducting from TO.

We're now up to SR5s running over waveguide on the same 12'ers and will still lose our shirts each June. We simply cannot create enough fade margin to keep the link up without space diversity.

This is an extreme case of course, but I use it to suggest caution and more local research when assessing the potential problems of very long links. You CAN get bit. We're now gearing up for our fourth season of this particular nightmare.

George

[tog]

Wow, is there any way you could maybe make your links shorter? Put one or two hops in between so you don't have such a long link to contend with?

[George]

We could break the links up, but then we start to deal with a lot of hops. We already have six 30mi (50km) hops in our backbone, breaking those would add maintenance and latency. We're 146 miles from fibre to my home and I get about 9-11mbs to my desktop currently.

We would also pay more for towers, and quite a bit of the space we pass over has very low potential customer density even by our standards.

The benefit would be cheap radios and higher throughput which would be very nice.

I'm seriously thinking about giving the main backhaul to the licensed guys to worry over if we can get a decent price.

We can live with losing the occasional packet, in fact would never even notice it. Then I can sleep at night in the spring!! The big downside is they are telco interfaces, so we would have to go up and down from DS3 to Ethernet.

George

[tog]

I never thought I'd say it, but in your messed up situation I think something licensed might be a good idea!