Good vs Bad constellations (yes, the graphs are useful)
Posted: Sun May 31, 2015 7:39 pm
Thought I'd show some UBNT AC gear in action and what the constellation patterns are actually useful for.
This is a PTP link with NBE-AC-19
As you can see, one side has tight little patterns while the other side is scattered MUCH more. One unit is tucked down between trees (incursion into second fresnel) and so there is little interference but some typical fresnel incursion issues. The other is up on a hill exposed to more interference.
This is in a 10Mhz channel. here are speed tests showing the difference between the good constellations and the bad.
The good pattern:
The messy pattern:
duplex pattern:
A couple things of note.
~65Mbps on 10Mhz very nice!
~25Mbps on the 'bad' side in 10Mhz is not that bad.
This link scales almost perfectly with channel width. 125Mbps*45mbps on 20Mhz, ~235Mbps*85Mbos on 40Mhz. The biggest issue with the wide channels is that it's hard to keep the CINR up with the 4x more footprint (40Mhz vs 10Mhz) to get interference.
Notice the 35 vs 30 CINR. This is a VERY good number to use to determine the quality of the link and if you should bump up from the NBE-AC-19 to the PBE-AC-500. Also notice that the higher signal actually has the lower CINR.
This is a PTP link with NBE-AC-19
As you can see, one side has tight little patterns while the other side is scattered MUCH more. One unit is tucked down between trees (incursion into second fresnel) and so there is little interference but some typical fresnel incursion issues. The other is up on a hill exposed to more interference.
This is in a 10Mhz channel. here are speed tests showing the difference between the good constellations and the bad.
The good pattern:
The messy pattern:
duplex pattern:
A couple things of note.
~65Mbps on 10Mhz very nice!
~25Mbps on the 'bad' side in 10Mhz is not that bad.
This link scales almost perfectly with channel width. 125Mbps*45mbps on 20Mhz, ~235Mbps*85Mbos on 40Mhz. The biggest issue with the wide channels is that it's hard to keep the CINR up with the 4x more footprint (40Mhz vs 10Mhz) to get interference.
Notice the 35 vs 30 CINR. This is a VERY good number to use to determine the quality of the link and if you should bump up from the NBE-AC-19 to the PBE-AC-500. Also notice that the higher signal actually has the lower CINR.