Hi,

we are setting up a PtP 70Km link, using the DBII miniPCI card and nv2. The card is MiMo, I will configure it in N mode. I will using 2 29dBi dish antennas.
Have you any advice for configuration/set-up?
How much throughput can I expect?

thanks!
Davide

Hi,

we are setting up a PtP 70Km link, using the DBII miniPCI card and nv2. The card is MiMo, I will configure it in N mode. I will using 2 29dBi dish antennas.
Have you any advice for configuration/set-up?
How much throughput can I expect?

thanks!
Davide

I am assuming 5.8Ghz

make sure you dont exceed EIRP for your region. In the US it is 53dB for a point to point link so your radios would have to be 24dBm or less. MAX antenna is 29dBm. That means radios no more than 24dBi. remember, if your radio link can fall to MCS0 and you are using default power levels, it could be up to 27dB! set radio up accordingly.

Next, be aware that your fresnel zone is going to be huge. Realistically, you will need 100m towers on both sides if the ground is level between them. If you are on hilltops then you can do less, but at the 35km midpoint your LoS line will need to be at about 100m to clear 60% of the fresnel zone.

These are very hard links to make because of how the fresnel zone balloons out at distance.

If the stars align for you and you have enough tower high and clean fresnel and your SNR is good, you might get 120Mb sync rate and 60Mb/s aggregate.

If you can't get enough height, then lower your modulation rate all the way down to MCS0 on the AP. Make sure you are passing as much traffic as possible over the link and watch the signal levels and the noise floor. You need 20dB SNR. bump up the MCS level by one and repeat the test until you cant without dropping below 20dB SNR.

Also, write down the sensitivity levels of your radio for each modulation. Use the worst number between the sensitivity and the noise floor (if the noise floor is -96 but the sensitivity is -85, then use -85).

I might even consider aiming for 24dB SNR because when you get weather fade it will effect very long links more.

Good luck and please post results!

Ubiquiti has 30 and 34dBi dual polarity dishes, and Maxxwave makes a case that fits right into those using a 411 or 711 board.

Ubiquiti has 30 and 34dBi dual polarity dishes, and Maxxwave makes a case that fits right into those using a 411 or 711 board.

'UBTIK' available at baltic.

*BUT*, why not just use a rocketM5?

Anyway, assuming this is in the US, there is that 53dB EIRP limit.
53dB - 34dBm dish = 19dBi MAX radio power
53dB - 30dBm dish = 23dBi MAX radio power

end result, same power at receiver, more money out of pocket.

differences
34dB dish will see more noise. this may not be a factor with such a narrow beamwidth
34dB dish will have to have less radio tx power, turning down the radio can clean up the signal.

34dBi dish will see less noise than the 29dBi, narrower beamwidth and I'm betting less likely to interefere with the other link than dual 29dBi unless they are spaced pretty far apart. The cables for the dish are a lower grade coax, but they are also about 4-6 inches long.

I like Ubiquiti stuff, but I prefer MT for AP's and long links. With a 34dBi dish you will need less power, so having to use less power isn't a problem.

Hi,

we are setting up a PtP 70Km link, using the DBII miniPCI card and nv2. The card is MiMo, I will configure it in N mode. I will using 2 29dBi dish antennas.
Have you any advice for configuration/set-up?
How much throughput can I expect?

thanks!
Davide

I am assuming 5.8Ghz

make sure you dont exceed EIRP for your region. In the US it is 53dB for a point to point link so your radios would have to be 24dBm or less. MAX antenna is 29dBm. That means radios no more than 24dBi. remember, if your radio link can fall to MCS0 and you are using default power levels, it could be up to 27dB! set radio up accordingly.

Next, be aware that your fresnel zone is going to be huge. Realistically, you will need 100m towers on both sides if the ground is level between them. If you are on hilltops then you can do less, but at the 35km midpoint your LoS line will need to be at about 100m to clear 60% of the fresnel zone.

These are very hard links to make because of how the fresnel zone balloons out at distance.

If the stars align for you and you have enough tower high and clean fresnel and your SNR is good, you might get 120Mb sync rate and 60Mb/s aggregate.

If you can't get enough height, then lower your modulation rate all the way down to MCS0 on the AP. Make sure you are passing as much traffic as possible over the link and watch the signal levels and the noise floor. You need 20dB SNR. bump up the MCS level by one and repeat the test until you cant without dropping below 20dB SNR.

Also, write down the sensitivity levels of your radio for each modulation. Use the worst number between the sensitivity and the noise floor (if the noise floor is -96 but the sensitivity is -85, then use -85).

I might even consider aiming for 24dB SNR because when you get weather fade it will effect very long links more.

Good luck and please post results!

The thower are both on top of hills, ~500 meter above the sea. So I think in worst case there are >200 meters clear on all the path for fresnel clearence.

These are very hard links to make because of how the fresnel zone balloons out at distance.

How means "balloons" ?

34dBi dish will see less noise than the 29dBi, narrower beamwidth and I'm betting less likely to interefere with the other link than dual 29dBi unless they are spaced pretty far apart. The cables for the dish are a lower grade coax, but they are also about 4-6 inches long.

I like Ubiquiti stuff, but I prefer MT for AP's and long links. With a 34dBi dish you will need less power, so having to use less power isn't a problem.

Do you mean that you like Ubiquiti antennas and you use the Rochet Dishes with MT?
Generally speaking, do you think that using less power, and much gain on antenna, is that the better way?

Have you used nv2 for long link? is it better than nstreme?


thank all for advices, I will post the results!

The thower are both on top of hills, ~500 meter above the sea. So I think in worst case there are >200 meters clear on all the path for fresnel clearence.

That's great. This is probably the toughest thing about long distance links, getting high enough to have a clear fresnel zone

How means "balloons" ?

The widest point of the fresnel zone gets very very large at long distances. Imagine that you wanted to make a 200km link. You probably have enough power you make that link with mikrotik or ubiquiti equipment, but the fresnel zone gets so large in the middle that you would need a tower 500m high. This is compounded by the curvature of the earth which effectively makes it look like there is a big hill between the towers.

Do you mean that you like Ubiquiti antennas and you use the Rochet Dishes with MT?

This is a good method, I have a few of these. If you look at baltic networks they have a product called the UbTik which is a mikrotik rb411ah board in a metal enclosure that snaps on ubiquiti antennas like a rocket. FYI, I have the UbTiks with the R52Hn cards. I run these with an output of 17dBi, any more and they seem to get noisy and the link quality drops. I also like to run rockets at <20dBi, cleaner signal.

Generally speaking, do you think that using less power, and much gain on antenna, is that the better way?

I don't think that you could pick a method as a best-case generally speaking. I would say that running a radio at 80% power usually cleans up the signal and gives a more consistent link. More gain = more noise potentially visible. More power = more noise to other devices.

Have you used nv2 for long link? is it better than nstreme?

I don't know, I use nv2 or airmax, only tested with nstreme and found nv2 to be better for that particular link.

thank all for advices, I will post the results!

please do, thanks

generally speaking, at 5GHz, using actual radio cards, using less power, and much gain on antenna, is alway the better way!

so

we are still waiting for the results, share please =)

Sorry, but the link hasn't been build