Hello meshers from N6TVE-WCC-BM2HP-SE60. I just plugged in, fired-up, and after a couple of reboots to get the wifi settings right there's a new node on the mesh in Santa Barbara :-). Location is 5637 W Camino Cielo @2100' with view to the SE, approx from east end of Santa Cruz island to La Cumbre Pk.
Equipment is Ubiquiti Bullet M2-HP with 15 dBi Yagi. I can connect to K6TZ-GIB-NSM2-W60 (that's where the antenna is pointed) as well as K6FLD-HM-RKTM2-WNW. My node can see K6LUD-SB-NSM2.local.mesh (60%-90% LQ) but can not be heard by that node at all (0%NLQ), not sure what that means...
"mor to come"...
Equipment is Ubiquiti Bullet M2-HP with 15 dBi Yagi. I can connect to K6TZ-GIB-NSM2-W60 (that's where the antenna is pointed) as well as K6FLD-HM-RKTM2-WNW. My node can see K6LUD-SB-NSM2.local.mesh (60%-90% LQ) but can not be heard by that node at all (0%NLQ), not sure what that means...
"mor to come"...
73
Orv - W6BI
I thought it was a problem on my end, but I'm reaching the conclusion that the SBARC Gibraltar node has a local noise problem. The issue is that link quality from gibraltar to me is great but the other direction is noticeably worse. Well, I took a look at that node's mesh status page and all outgoing links are better than the corresponding incoming, i.e., LQ < NLQ. Here's a screen shot:
If not already, take a look at tuning 'distance' on all the clients and 'dial them in'. I see this general symptom on hub sites-- LQ of neighbors much lower. While a hub site generally has sector or omni coverage receiving noise from a wide area I'm seeing clients with very strong SNR generally way above this noise. CSMA in theory scales up to a much larger number of clients, but then I suspect most studies are not with long distance clients where the timing is all over the board.
The problem I see in Orange County is that there are about 4 'hub' 2.4Ghz coverage areas. These cells are seeing each other and it my testing seems to be demonstrating that the distance parameter needs to be set for the farthest signal it is hearing or more inefficiencies occur. I've found instances where one hub site will send out a RTS affecting another hub site.
The LQ/NLQ is based on UDP broadcast packets. Take a look at the TxMbs values on these clients to see how they compare to the TxMbps in the opposite direction at the hub node. The TxMbps factors in the measured transmitted packet success rate from the rate selection table. This will give a more accurate view of the link. While the LQ/NLQ may be a 50% to 97% comparison, the TxMbps might be 15Mbps to 18Mbps comparison. This might be telling a different story.
Joe AE6XE
The 2 extra nodes on same frequency are providing a lot of desense on site likely which hinders everyone else.
Note: it's not uncommon for the high site to show low LQ's and high NLQ's this is normal since the high site hears everyone and gets all the collisions.
Orv
W6BI
I believe the plan is to upgrade that site to Rockets and sector antennas in the future.
The distance parameters were setup for the Gibraltar peak nodes based on the farthest expected connection in each direction. West node has extra RF shielding. The other two nodes we removed the extra shields because of significant performance problems. The addition of the shields dropped peoples SNR by 6db.
The shielding would make a lot more 'hidden transmitters' and contribute to more collisions and inefficiencies. When they are all sharing the same freq-channel, it would be better that they all hear all the possible clients to better coordinate using the channel. The mesh node does a Clear Channel Assessment (CCA) before transmitting. If it can't hear a client an adjacent node is currently receiving, then it can transmit and drown out this neighbor. It might improve even more by removing the West node's shielding.
[correcting this statement before some one else does :) ...]
The data rates must have jumped up without the shielding? The specs show at higher MCS rates the txmit power is lower that might explain some of the SNR drop.
Alternatively, use 5Mhz channels, add the shielding back, and use ch -2 and ch -1, and ch1 (if part 15 allows this channel to be used...).
Joe AE6XE
Maybe the node facing west (which is the one I'm hitting best) would be better without shielding also to improve SNR?
However, on the other hand, I can see a factor that a sector or limited coverage area antenna does not receive all the interfering signals from all directions the omni will hear to decode a given signal and the directional coverage generally has higher gain or higher SNR than an omni.
I'd be speculating on which is better and may depend on a given environment and specific equipment.
Ideally, with money growing on trees and available channels, we'd all do 90deg sector coverage (maximize gain and SNR) and 4 devices each on different channels (no contention) at every tower site. We do get more channel options on 3Ghz and 5Ghz to realize this.
Joe AE6XE