This is my first shot at a PtP link. We have several PtMP 802.11 links going. Setup is 2-RB112 with SR2 cards in them. This link is connecting a remote pop with our main pop. I have it setup with exact fit framing and polling turned on. Using a 10mhz channel. The system is reporting a -59 on both ends of the link. Best I’m seeing is 18/24 mbps. Why isn’t the card in a solid 54mbps with that type of signal strength?
Noise floor is being reported at -104 and tx/rx ccq averages out to about 50.
We are using this as a test link for a bigger link that we would link to pull a full 30 mbps full duplex through, but if we can’t get full throughput with a signal like this, I’m not sure we will be able to get it on our new link with a SR5.
Speed tests are giving me about 4 meg. Speed on this link is not important. Canopy at the other end can only push 4 meg anyway.
My question is more on the line of why can’t I get the radio to hold at a solid 54 meg or even 48 meg with a -58 on both sides? The SR2 shows that it needs -74 to do the 54 meg. So even if I drop 5 dbm for the 54 speed that should give me a -63 which should be enough margin to get the 54.
Even at -58 the system seems to be having trouble holding a 24 meg connection and falls back to 18 on a regular basis.
Is it just me or is this way off…i’m looking at a WDS setup i have between a rb112 and and wrap 2c both with 8602 cards (the wrap is the host with 2 cards, rb112 has one card).
right now i’m seeing: signal: -65 stN: 34db tx / rx CCQ: 94/89
UPDATE: i just tested my setup from above and i’m getting 5.4-6mbit. 2.4ghz..udp tests..problem is the rb 112’s CPU is maxed out. Im NOT using NSTREAM which adds cpu overhead…
I think you need to test the BW with out the BW test (ie ftp a file across the wirless link to a standalone ftp server and see how fast it goes ).
My problem isn’t the speed, I expect the lower speeds on the RB112. But with a -59 on both sides why is my CCQ so low and why is the radio not locked into 54 mbps.
As I said we are thinking about using a setup simliar to this in the 5ghz band for a major backhaul, but I’m a bit gun shy now.
You liikely have an interference issue that MT is not reporting properly. Try moving around the band(s) and see what difference you get on the different channels.
Lousy throughput is generally the most reliable indicator of interference in the area. Also manifests itself by not being able to achieve high over the air rates.
Also, try turning down the power on the SR2s. There have been some issues reported with overpowering the amplifiers on the SR2s when using narrow channels. Try setting for “card rates = 20” or lower and see if this helps. It may reduce distortion out of the PA.
Use “best fit” framing as well, although that is not the real problem of course.
I knocked the power down on both ends to 17. This seemed to help my radio speed, which now is 24/36 or 36/48. It also seemed to help my CCQ on the client, which is now in the 79-85 range.
Odd thing is even though I dropped the transmit power from 26 down to 17, the RX signal on both side only show a drop of 1-2 DB. Any ideas on that?
Interesting. I’ve had some excellent results down around 14 as well on the Engenius 8602s so it might be worth you while dropping the power even a bit more. All you really care about is over the air link rate and IP throughput…
Can’t say for sure, but I think there may be a couple of things happening here. First, your SR2s may not be capable of full output. There does seem to be some variability in most of these high power cards.
Second, the output power calibration may be off a bit in the MT. Even less sure about this one.
It does seem that the WORST thing anyone can do is try to over drive one of these high output radios. All it does is creat distortion, and the ONE THING that OFDM doesn’t like is distortion. Even regular low output radios behave the same way…
First rules seems to be with a problem link is turn down the power. Strange but true. We’ve seen it time and again.
Something else to watch for on long links particularly:-
Buy the best antenna you can afford. With dishes you are looking for a highly accurate parabolic shape that will stay nice and rigid in the wind.
Cheap is usually bad.
Reason for this is to get the signal to all arrive at the same time at the far end. You only need a few millimeters of error in the dish form to create the equivalent of a multipath problem. Then the OFDM engine has to work overtime and your throughput suffers.