Suburban Internet - Is It Really Feasible


About roof top AP mounts:

This is done here with us for many years already. Usually the house owner gets free internet in return. Usually when the owner sells the property the new owner is all to happy to continue the deal…

We’ve had once an instance with a top floor appartment owner that fell out on us and disconnected the power to the tower. Panic! But within 2 days 99% of the clients were realigned to another tower and a week or two later we found ourselves a new building. The falling out has all to do with an ex employee that started a business on his own… so I’ll guess you’ll have to watch for that…

The occasional hassle with owners compared to the usuall huge renting costs (and in-availablity) of towers made us strong believers in the ‘house roof mount’ fit for years already. I’ll guess we’ve set a trend in this region. Since then we now see installations everywhere from many providers on many different houses of appartment blocks.


@Rudy We get a lot of wind. Im located in Nebraska. The wind blows all of time. There are a lot of times throughout the year where we get 40-50mph or more wind gusts.

I haven’t had any issues with any of these sites on windy days. Its mounted low enough there is no movement. Its on the trunk. There is hardly any wind load to the antennas Im using.



Im not sure why your speeds drop off. My guess is this could be a issue wit a router or something else on your network. (Not the A5 or C5) I would check ethernet and your PHY rates. If PHY rates drop while the speedtest is performed it could be interference.

We compete with two other WISPs in my area. They are on two different towers in the same town. The difference is my customers are all VERY close to the AP. So I don’t have issues… Im able to run in 80MHz channels because of all the trees in town. They basically block all of the A5s from seeing each other. The A5s are also mounted low (less than 50ft) so they are well below the tree line protecting them from interference.

I dumped UBNT because we were only getting about 10Mbps or less per client. Im actively dumping UBNT throughout my entire network and replacing with Mimosa.

I don’t let my installers install any C5 with a signal worse than a -60 in town without my approval.


Hello @Kent Kent,

can you depict a precise list of buyed items ?

I have done following list:

  1. A pole 50mm diameter 3 meters high 3mm thick
  2. A pole 42mm diameter 3 meters high 3mm thick impiled on first one
  3. A plastic rain cover for top pole
  4. A5-14
  5. B5-lite or B5 + a small C pole segment to fix onto
  6. A pole case to put a PoE 4 port switch, 1 routerboard CCR1009-7G-1C-PC, 1 UPS

actually the cost is under 2.000 €.


I have seen you are using IgniNet 60 Ghz backhaul. What is max distance on that PtP ?

I’ve read on distributor that the distance couldn’t more than 1,5 Km. It’s real ?


The price I gave was a MicroPOP with Metrolinq for backhaul. They are very short range. The longest link I have is 1/4 mile.


How is your spectrum ? Can you share it ?

I had in mind to use a 40 Mhz spacing but if are there too much
interference ? And if tomorrow many and many others will do the same thing
in same area what happen then ?

This is the only my doubt.

Massimiliano Iavazzo
Responsabile tecnico
Telecomunicazioni srl con socio unico
sede legale: Centro Direzionale di Napoli is A7 - 80143 - Napoli NA
sede operativa: via G. Verdi 48/A - 81031 - Aversa CE
P.IVA: 07136161218
t: +39.081.19309465
f: +39.081.19758072
Rispetta l’ambiente. Stampa questa email soltanto se è estremamente



We run a similar setup. What routers are you using? I’d like to compare notes and understand exactly how you have things set up.



Sure. There is a contact form on our website That will come to me and we can talk off list.


Kent would you be willing to discuss off list on some questions regarding the environment?


I think it would be better discuss all details here.


  1. how many C5 do you linked to a single A5 ?
  2. what are your delivered bandwidth profile and MCR ?
  3. can you share your spectrum to permit everyone to understand your actual condition to reach that profiles?
  4. You tell us you used Metrolinq to connect togheter microPOPs. It’s ok. But where you get mainly connection ? How many hop you will pass to reach your internet gateway ?

If someone could add other question, I think there’s no best place to do.



@Rudy do you have any update regarding your performance collapsing with Mikrotik CPE ?


Where are we now?
We have 55 Mikrotik CPE’s (almost all SXT’s) allowed to associate to the A5-14 and one C5. On average we see around 49-53 connected units. (Some switch their antennas of at times)
Assigned speeds are 50burst/25Mb download and 5burst/1Mb upload for most of these clients.
All clients are within 400meter range and almost all in a 270 degree sector of the AP (Thus one ‘panel’ of the A5 hardly gets any data to it… )
We’d use queues for the speed limits in a Mikrotik CCR router at distance. But for test purposed we also set speed profiles for each client in the access list of the A5-14 at much higher level then our main limits in the Mikrotik.
Usage on the AP we see only peaks at around 50Mbps which is easy to produce by the A5-14

Lately we didn’t run a lot of max. throughput tests because in fact the network runs fine. Clients we’d speak are happy and some of them use streaming content and they have no problems or so little they don’t bother to mention.

We found that Mikrotik in the last year or so change the rts/cts threshold settings (we didn’t look at that for years…) so my standard perception (still embedded in a standard config script we’d use) was “256” where we now learned it has to be “0”.
So we changed it on all clients and we add several more SXT-ac units capable of the same 80Mhz wide channel as the C5.

Overall we are very happy with the ‘interop’ setup so far.
In the past we’d tested 802.11 legacy with Mikrotik AP’s and that was a disaster. And in tdma 30 units is sort of the maximum where now with the A5-14 50+ units don’t seem to give issues. If fact it even works better then the MT’s…

The ‘collapse’ of the network under normal conditions we don’t have and as said, we didn’t do a lot of testing on max speeds lately.



  1. 5-40 clients per A5.
  2. Most of my A5s have the clients set to unlimited now (for testing purposes)
  3. noise is bad… but thats why you build your A5s to be close to your clients less than 400meters. Mount low to help shield the A5 from noise.
  4. 3 hops.


Hey Kent, regarding your point 3 - I’m actively looking at some potential houses for MicroPOP hubs, and I’ve been mostly looking at houses which have clearance above the tree canopy. Looking at your pictures from earlier, your environment looks typical of mine. Do you find that mounting below the tree canopy is working well - ie, you’re getting those high modulation rates at 400m despite a “few” trees (not just 1 tree)? And that the tree “shielding” helps more than it hinders?


Yeah… About half my sites are actually below the tree canopy. The trees “shield” the A5s from competitors noise.


3). Bad noise. How bad? What are the levels you see? And what are the signals between AP and CPE’s? As long as the S/N is ok too much noise ain’t bad. And probably in any urban environment you’ll find lots of (5Ghz) noise…

  1. 3 hops? What you mean with that? The A5 has 3 hops to go to your central? Irrelevant info for specific A5 performances.


We have regurlarly 59 customers connect now. Still no probs. Peak usage we see now is 60-70mbps but average we are still around 25-35 only…

We started to tests with streaming IPTV (no QoS in place1) over the A5 to two C5’s, one even with two simultanious streams, and good, very good.
We are now going to seek to do tests with more IPTV users soon.

Streaming IPTV is very demanding. It needs a stable, low latency and low jitter network. 59 clients (not all using data) is demanding and if 2 of these are demanding continious 4-6Mbps (HD TV!) data streams where other users still do their ordinairy stuff (40-60Mbps average combined user traffic over the A5) then I am delighted to see no buffering or cuts take place on the streams…

We tried to do the same on a 30 clients Mikrotik Netmetal running NV2 (tdma) and during rush hour the client noticed buffering.

All in all I am still very happy. I’d probably first proceed in replacing all my AP’s into mimosa before replacing all CPE’s of this first A5 into C5’s.
The latter cost me a leg and the extra performance probably isn’t needed, and is a still a promise anyway since GPS sync/tdma is still not available…