Replacing Ubiquiti 900Mhz with 450, Requesting thoughts on what to use.

Hello all,

Looking to replace an existing Ubiquiti 900Mhz link (7 miles, nLOS/NLOS) to get higher throughput. I'm torn as to how to improve this link... on paper it should never work with multiple 50% fresnel zone incursions (trees) and linkplanner shows very abysmal projections, but, all real world evidence to the contrary.

Been working for over 3 years with about the same data shown below, day in and day out.

Existing link Details (RocketM900's)

-69 Vertical / -71 Horizontal, combined -68

Noise Floor -99dbm

5Mhz channel

Running Airmax

28dbm TX power

16dbi Yagi's on each side

This is out of state from where I am and was installed by user, accuracy of alignment is unknown, but it's definately working.

I'm not sure if I should go with 900Mhz 450 with KP Performance 17.5dbm Yagi's, to try and squeeze out as much as possible of the 900Mhz spectrum and not play with the different variables of trying to have the user install 2ft+ parabolic antennas and the differences in higher frequencies, interferance, etc...

*Side note* there is another 900Mhz ptp installed 90 degrees off of this link by the user, and there has been some known local interferance, but keeping the channel width low has kept that from being an issue.

The problem is, I have no doubt that the Cambium radios will outperform the Ubiquit ones in every way, but will the throughput be where the user needs it to be? Obviously I can't rely on linkplanners calculations, even with a custom path profile the best case scenario ("on paper") that I'm seeing is 46Mbps at 91.8966% reliability in a 15Mhz channel, 85/15 duty cycle... but because the real world numbers are so much better I should expect to see better results here as well?

This is a "residential" use case, would a sub 1Ghz licensed link be capable of doing those throughputs?
Should I go with 2', 3' parabolic antennas and switch to 2.4Ghz or 5.8Ghz with those kinds of signals in the 900Mhz band? 

Approx 12Mbps of throughput in one direction on average on this link as it exists now, peaks of 16ish, as low as 9Mbps. New link would ideally average 30Mbps+ to be a viable option/expense, with 60Mbps being the target, anything above that is gravy. Bonding would be worth considering if it made sense.

Any thoughts are appriceated!! Let me know if you need more data from me.

If the UBNT 900 is working now, the Cambium 900 will work better. I have no doubt whatsoever on this. I've done exactly what you're doing, taking a UBNT 900 PtP link and swapping it out for a PMP450 900 and the link improved dramatically. It basically went from nearly unusable 5MHz channel width to being able to use a 20MHz channel width and getting around 75mbps out of it. ALSO, I'd spring for the KP 17.5dBi yagi's, they're very high quality and work great.

One more thing... LinkPlanner tends to be very conservative with it's estimations, especially when it comes to calculating nNLOS paths.


Did you proceed with the 450 900MHz PTP radios?

If so what was the outcome?