Trying to get a new tower going with our first Force 300 PtP link. The link is up and running -55 RSSI and 36 SNR so seems to have good numbers. Doing a speed test from tower base (speedtest.net) gives the numbers I'd expect given the time of day and a 20MHz channel. I have two epmp 1000 AP's on the tower on different 20MHz channels than the force 300. All three radios were about at same height on top of 120' tower. Today I climbed and put a second 'client' force 300 about 10-12 feet lower on the tower than the AP's to see if I have a bad radio. Couple issues:
1. I can't manage the subscriber side of the link through the web page however, the page starts to load then just spins and spins. I can SSH to it wth no problem.
2. (NOTE: I haven't tested this again with new radio located a bit lower on tower) Although I can get proper speeds connected to the switch at tower base through the link as soon as I try a speed test through a subscriber radio connected to one of the epmp 1000 APs the speeds drop to a fraction of the link capacity. Is it possible that even on different frequencies the epmp 1000 AP's can affect the force 300 link? Frequencies are not adjacent channels either. Force 300 at 5805MHz and AP's in lower 5700 range.
Any insight appreciated. I can post more relevant config info if it helps.
Update on this. Looks like the web page issue loading issue might be a bug getting fixed in next release. According to the one customer off the new tower things were working good after the radio position change. I'll try to report back with a more specific update, planning to get out there to do some signal tests later today. Hoping I just misjudged the physical separation required between AP and PtP radios.
Just remember that radio waves work on the inverse square law. A little movement on a tower can make a big difference with interference. If your PTP was 2' away from an AP and you moved it to 4' ( 2 times the distance; or 2 squared) away, the AP is then seeing the PTP's power 4 times less than it was. If you moved it 10 feet away (5 times increase in distance) then the AP is seeing the PTP's power 25 times less than before.
There are too many missing variables in your posts to know if this is the case. Just throwing this out to keep in mind for future reference.
Thanks for replying, I am thinking this explains it from a technical perspective. I did a wireless link test between SU and AP at same time as doing same test between both ends of PtP link yesterday and speeds were perfect so from an RF perspective I'd say looking way better, AP link and PtP link seem to not be drowning one another out anymore. Actual speedtest.net results better but not great but that is unrelated from looking at the NMS graphs, waiting for a 24GHz link to arrive to alleviate a network bottleneck.
Well still issues. I can do the wireless link tests with no problem in the web interface but not getting much for real throughput across the link. 10-15Mbps max. Signal and SNR great. I have tried setting entire link at 100Mbps as we have epmp1000 AP's at far end - just in case of a flow control issue or buffer issue or something going from a gig port to a fast ethernet port on the switching side (toughswitch 5). It's only a 20MHz channel across the ptp link though so barely (if at all) pushing a fast ethernet port. I guess next steps are to bypass and/or change out switch, try cable changes, bypass ligntning arrestors, etc. Hoping am not missing something obvious with the force 300 config. I don't think this is any different than our other towers that are all fed by cambium gear. I'll try to post back if I figure things out, just seems strange.
For testing, make sure you're running 4.3.1 on your radios. On the Mikrotik make sure you're running ROS 6.44 or greater. Make sure you're using 5+ TCP streams for your bandwidth tests. ALSO... keep in mind that these are TDD radios and have fixed DL/UL ratios.
I have faced the same issues on sites with a lot of "noise" (not only 5GHz) from various radios and sources (excellent link test but low throughput) the workaround (depending on the case) was,
- If you have proper grounding then connect everything to the ground (not only surge arrestors but switches etc) carefull of course no to have grounding loops
- Remove surge arrestors. If the grounding is not good then it will "add" more noise on the cable. The bizarre thing is that in some cases you will not see errors on the ethernet (monitor/performance tab) but still the throughput will be low. After removing the surge arrestors you may also have to play with the Ethernet port speed.
- Ethernet port speed. In noisy sites, we have encountered issues with the combination of ePMP and UBNT Toughswitch. Setting both Ethernet ports on 100M instead of auto or 1G usually works.
- Setting the Radio Ethernet port in Auto and fixed 100M on the switch. I know it does not sounds logical but it have worked.
Also we have tested expensive UTP Cable with grounding points but if the grounding of the tower is not proper then the result is even worse (ports flaping).