EPMP - UBNT Change Over

@pgator17 wrote:

Is the 2.6 in beta test form yet with the 10mhz wifi? 


2.6 is not beta. It is an official release that you can download here: https://support.cambiumnetworks.com/files/epmp/



I'm still waiting for our Force 200s before I do anything with 2.4 (which should be here soon now)... now that 10mhz channels are supported in WiFi mode, it should be easy to do some direct comparisons to UBNT.

I've tried to put all customer routers on a set channel away from the AP but there have been cases where i've had to change the AP's channel months down the road. At that point I don't have access to the customers router remotely anymore, so I can't change the channel unless I do a service call.

I think the force 200 units will help tremendously with this one problem i've had, I just haven't been able to test that yet.


we've changed hundreds of ubnt 2.4s to cambium and its wonderful.    the EPMP have been replaced using the same channel space most of the time durring the change over process. Both systems in 10mhz mode

I'll try and keep this somewhat short:

We’ve done most of our change over’s 1 AP at a time, and making it a cold change over until 2.6, we will be doing a slow and soft change over in a few weeks to see how it goes. 

We’ve been pulling 1 UBNT AP and replacing it with a cambium AP with the cambium MAC active. 

While swapping we've talked with the customers ahead of time and told them to expect a few hours outage, changed the AP on the tower and then changed all the connected subs.  2 guys have been changing 1 AP and 10 to 15 subscribers in a half a day usually. of course that fluctuates with the customers distance from the AP ect. ect. 

Doing 1 AP at a time has worked pretty well for us. We’ve changed the UBNT APs were customers have been having a lot of problems first, and 2nd the bulkier customer loaded panels. after the first AP is changed on a tower, some of the neighboring subscribers can be moved in advanced at a reduced signal to lesson the downtime experienced by a lot of subscribers.

with the UBNT gear, we've ran into the same problems everyone else has, trying to use 4 sectors on one tower is a massive challenge in itself to keep it running smoothly. having to keep the powers very loud,  low 60s have been our cut off point on connecting customers with UBNT.     our single biggest problem with the UBNT gear has been keeping the uplink togher, the lack of snyc cupled with our self interfence has been a major problem. even with shields and good sector tilt.  GPS has ended all of those problems for us. 

the next issue we've had is customer count, once we cross 12 or 15 it falls apart. 

Before changing over, about half of our UBNT sectors would struggle to get past 12 to 15 mbps at night, and most would have erratic performance and we have been getting more and more complaints. 

After changing the load on a lot of these same sectors have improved from 12 to 15 during peak to 20 to 25 mbps WITHOUT wild swings in latency and customer complaints have stopped.      the 15 MS ping from the cambium mac is generally stable, and has provided a much better experience for our customers, during peak, the games have reported a drop in ping time, and good game playability.     You will get maximum benefit after changing all of the APs on a tower and enabling snyc.   Flex works well for some, but we’ve found the general network capacity sync brings us is very much with that small latency increase.

our busiest 10mhz AP has 31 subscribers on it currently but its just an Omni and we are going to sectorize it soon.

the first large area we replaced all of the UBNT 2.4 gear with cambium, at peak, the backhaul feeding in was an AF5, showing about 100 meg used at night, once the changeover was finished and all we had done was changed the subscribers and tower equipment, the same group of customers was reaching 190 meg durring peak.  and the best part for us, complaining customers turned into happy customers.   we've been pushing sales on the same network segment as are seeing peak flows to 260 mbps.       this area is served with 22 GPS snyced APs and it was served by 22 M2 rockets prior.    

Another big perk of the EPMP, once we have finished large areas, we have used the GPS sync to use only channels 11, 8 and 5 for the few 3 sector setups we have, and the four sector setups have been using 11 and 8. we have been always setting customers to channel 1 on their router.   This has been working great. 

if you plan to GPS sync, which i highly recommend.   read up as much as you can about it, it works great, however its deployment and effective use depends on a lot of factors like coverage overlap,  and proper antenna tilt.    all parts put together, you can expect 3 to 4 times the capacity you see without GPS sync in a  dense network environment. 

 we have field depoyed 33 snyc 2.4 APs and last count was 350 CPEs.    We’ve got plenty more rockets to still get down and I can't wait. 

Others have commented about the lack of higher gain CPEs, we've been using the passive reflectors with the basic units and getting better modulations than the nanobeams.   actually most of the time, the integrated units do better than the beam for us since the uplink channels are so much cleaner thanks to GPS. but who doesn't like to park every radio at max modulation?  

in short,   our customers are happy with camibum, we've got extra bandwidth to sell and we are not sending trucks out to constantly to take care of petty signal problems.    UBNT gear is back down to less than half of our network and account for 80% of our truck rolls for deliverly related issues.  

spending $340 more on the APs have ended up saving us a lot of money in labor, and significate network performance improvements. 


@Mathew Howard wrote:

I'm still waiting for our Force 200s before I do anything with 2.4 (which should be here soon now)... now that 10mhz channels are supported in WiFi mode, it should be easy to do some direct comparisons to UBNT.

I doubt the cambium gear will run any differently than everyone else in wifi mode, its a standard compliant wifi mode of operation.  meaning everyone is doing the same way. 

the benefits of cambium EPMP comes from the proprietary changes. 

cambium stripped most of the "wifi" stuff out and made a lot of heavy changes to the PHY and MAC access layers.    I'm sure you'll see some differences between each brand in wifi mode,  but the wifi mode in 10mhz and 5mhz channel support is there to make changing off of the unsnyced devices easier for providers. the performance improvements come from the TDD mode. 

1 Like

@wavenetwireless wrote:

I didn't do a direct equipment swap from Ubiquity to Cambium but i've installed both and definitly prefer Cambium 2.4. My main issue was the fact they only had the integrated unit, which now isn't the case with the foce 200.

The problem I usually run across is NOT from interference between the CPE and AP, it's usually because of the customer router. Since 2.4 is still used 95% of the time in the home it seems like more often than not the customer router will do a quick scan and not see that the CPE is on a conflicting channel because the channel size is only at 10Mhz. Then once they get on the same channel the CPE powers through any interference, while the customer router gets overloaded and just flat out doesn't work.

Hopefully having the force 200 with a more direct beam there will be less side/back interference, but i've yet to test them yet. Has anyone else come across this problem?

we've delt with this by only using a specific channel for customers, and never changing it for them.   also restricting it to 20 mhz channel.  the 2nd thing you can do to aid this problem is add a passive reflector  ( KP performance makes some great dishes)   and I've been very clear to my installers that the router should be installed some distance away from exteror walls and the outdoor modem.       placing the router center in the home, gives them good coverage, moves it further from the CPE and allows a little less to bleed outside. ofcourse this won't work in evey situation, but keeping router placement in mind to dodge issues from the start can go a long way from stopping calls down the road

1 Like

We had started swapping all of our Ubiquiti (customers and backhauls) to ePMP. Anyone one we could move to 5Ghz ePMP has been done and we moved about 30 customers over to 2.4Ghz ePMP and put that on hold until the 2.4Ghz Force is released. Hopefully the 2.4ghz force will work as well as the 5Ghz force, if it under performs as badly as the 2.4Ghz integrated ePMPs do then we are going to have to rethink our plans.
Across the board the ePMP "integrated radios"  get a lot less signal strength than the Ubiquiti nanostations they are replacing. On our first 2.4Ghz ePMP AP where the customer had -65 on Ubiquiti the ePMP sees -74 (something we consider uninstallable in 2.4Ghz) . The integrated ePMP radios are just horrible in our experience. Where a bare Ubiquiti Loco gets -58 LOS  1 mile from the tower an integrated ePMP will get -64 , they're awful. Now, that said, if we swap the -58 ubiquiti with the -64 ePMP the ePMP will still outrun/outperform the -58 Loco.  
There is no comparison between the Ubiquiti Nanostations and the integrated ePMPs as far as signal strength goes.  If we have a bare nanosation installed with a -65 then on 2.4 we won't be swapping that customer until the 2.4Ghz forces come out because we will never get a workable signal with the ePMP integrated radio.  Those very few ubiquiti customers we have installed with -70  the ePMP integrated radio won't even link up.

At 10 miles we had a couple of customers on 2.4Ghz rockets with rocket dishes. When we swapped the rockets out and put ePMPs on those rocket dishes the signal strengh was almost exactly the same but the ePMPs just runs so much better and the customers are just so much happier.
So far every customer we have moved over to ePMP (about 150 so far), well there is one exception but then there always is, has told us (we ask all our customers every time the call or come in or we run into them) their Internet seems to run so much better now on the new stuff.  The gamers are especially happy "I almost never see lag spikes any more!".  So even though the latency is higher on the ePMP it is consistent while on the Ubiquiti was all over the place.
The ePMPs just handle interference a lot better than the Ubiquitis .
As far as speed tests go we speed test every install without rate limits so we can see what the link can deliver. On 20Mhz channels we consistently see 80-90Mbps  while on Ubiquiti I don't think I ever seen a customer speed test over 40Mbps and most were in the 20Mbps range. At 10Mzh channels the ubiquiti  will burst to maybe 20Mbps while the ePMP consistently delivers that in a 5Mhz channel.
On another note we are finding, at least on the 5Ghz we are going to try 2.4Ghz ubiquiti sector soon, the Ubiquiti sectors perform better than the Cambium... a lot better.


For my 2c - I have a 2.4 ePMP SM at my house.  I'm right in town, and shooting about 4km over the city, and I can hear a bazzilion routers.

With our previous gear, I basicaly had given up on 2.4Ghz in town (of course) since (obviously) no one can possibly use 2.4Ghz in a city (obviously)!  My previous AP/SM combination could sometimes get 4-5mbit in the middle of the night, but at 7-11PM we would be pretty lucky to get 1-2 Mbit.  BUT, it is 2.4Ghz and it's right across town, so what else should one expect, right?  SO - we decided that my place would be a good test for the Cambium 2.4Ghz. I mean, it's a horrible environment, but we had good long term tests with non-Cambium gear, so if it even links up from there, then that'd be a good indication of how well it would work in the Rural areas for us.

SO - I replace the AP side with a Cambium ePMP-1000 GPS Synced AP (using the same KP Sector that was there) and I replace the SM side with a Cambium ePMP-1000 Non-Synced Connectorized SM (connecting it to the same 22dB DualPol grid that was there).  So, no adjustment to antennas or aiming. Signals are in the  mid 50's, 20Mhz wide, Channel 9 / 2452Mhz...

...and I get about 100 Mbit agregate - about 70 Mbit down and 30 Mbit up. 

Downlink  71.286 Mbps
Uplink        29.484 Mbps

As I mentioned, the previous non-Cambium 2.4Ghz gear would barely work at that.

By the way, the SM I'm using is an ePMP 1000 with connectors, and it's attached to a dual-pol 22dBm Lanbowan wire grid. (http://www.lanbowan.com/ANT2327D22PG-DP.htm)  It's front/back wouldn't be great, which means the SM is dealing with un-necessary noise behind it too. I have some 2.4Ghz Force 200's here now (Yay!) and I expect that should improve interference even more.


Yes, I know... I'm not expecting it to actually perform better, or even well in WiFi mode, what I meant is that it will allow for a direct comparison of RF performance, as I have heard some reports of that they don't perform as well in nLOS environments as UBNT 2.4ghz... but that doesn't make sense to me, as the signal levels should be more or less the same, which leads me to believe that the antennas being used were likely to blame.

@Mathew Howard wrote:

Yes, I know... I'm not expecting it to actually perform better, or even well in WiFi mode, what I meant is that it will allow for a direct comparison of RF performance, as I have heard some reports of that they don't perform as well in nLOS environments as UBNT 2.4ghz... but that doesn't make sense to me, as the signal levels should be more or less the same, which leads me to believe that the antennas being used were likely to blame.

I think what you might be seeing is that UBNT M series 2.4 radios can have their power turned up to 28dBm, and UBNT has had higher gain client antennas for a long time now.

Now that the Force 200 radio exists for 2.4, I would wager money that given an apples to apples test... same TX power on AP and SM and using the same gain antennas or even the same antennas, you'd see similar near/non-LOS penetration from both systems. What would be really interesting to see would be the throughput numbers each system got given this scenario.

We noticed a point or two drop from the ubnt nsm2 overall. However our modulations have generally been better and when we’ve changed from liner antennas the xpol our chain matches have been slightly better and noticed some units that where sitting at mcs 12, moved to 13. Compared to ubnts 2.4 ghz 90 sector, the Cambium sector has smoother edges and we noticed a gain at our lobe edges over the ubnt antennas, ubnt has them for -6 db at the edge, Cambium is -3. I’m sure there will be times one system reaches further than the next on power so testing of several link conditions would be useful. For us, the overall performance has been there the entire time. The passive reflect brings the gain to 18/19 db says kp performance and it shows. The reflectors in nlos conditions have be significantly out performing nano beams in an RF stand point for us. The beams didn’t match pols well and didn’t really seem to uplink any better than nsm2s for us. The 5ghz beams in my opinion were worth buying, the 2ghz beams are junk.


Awesome. There is some good reading here...I have a single Ubnt 2.4 sector I plan to change over soon. Has about 15 subs on it on a 10mhz channel. Yes, ping times are all over the place, no consistancy at all. I have the newest beta software on it so I can see the mem and CPU counters. They are always within what I consider normal numbers, even when all registered radios pings go over 100ms...

Do yall recommend changing the Ubiquiti sector antenna out for the Cambium one? I have RF-Armor on it.

Also, will the Ubiquiti power supply work? Would be nice to not have to go in the house each time...

Very anxious to get this switched, but have to get almost all Force 200 units when they get available.

For the 2ghz cpes, the ubnt 24v supplies work. They will not work for 5ghz. The sector antenna will work also, if you look arround you can find a radio mount adapter online, a hose clamp works too.

"Do yall recommend changing the Ubiquiti sector antenna out for the Cambium one?"

For what it is worth we have a 5Ghz ubiquiti sector (no shield) that appears to be out performing our Cambium sectors right now. It's just a single sector so this is pretty anecdotal atm.   Yesterday we went up and swapped out a 2.4Ghz Cambium sector (last I knew the 2.4Ghz cambium sectors are 90degree)  with a 120 degree 2.4 Ubiquiti sector and really don't  see any difference in signal strength but all the connections on this tower/sector are less than 1.5 miles so we have yet to really test it vs range and edges .  

These will let you mount the ePMP to the rocket antenna.


As for what will/won't work with Ubiquiti power supplies. I believe that right now all but one model of ePMP will work with Ubiquiti power supplies. The one exception is the 5Ghz ePMP 1000 subscriber radios (no GPS sync connectorized and/or integrated). The new 5Ghz force 180 radios will work with the ubiquiti power supplies and people are saying the new force200 radios will also work with ubiquiti radios. 

Something to keep in mind powering t he ePMP access point  (GPS Sync radios). The GPS Sync radios  appear to have resistors in them that act as heaters when the radio is starting up and it's cold. These resistors pull extra power when it is cold and you may find that 24v power sources like ubiquiti power supplies do not work where the 30v ePMP power supplies will.  I don't know if the non-gps radios have heaters, we haven't had any problems with them but almost all our customer installs have Cambium power supplies.

The problem with Ubnt sector was not the FTB ratio ? I think it's not enough to acheive MCS15 with back to back frequency reuse . Maybe with shield, it would be fine.

all of the epmp radios have the "heaters"     I'll try freezing a 2.4 unit today and see if a ubnt supply can warm it up to operating temp.

the 120 degree ubnt sector has a similar forward arrangement as the cambiums 90 degree, ubnt likes to mark the sector width at -6db and most other manufactures use the -3 as the edge.      looking at the graph between the two, the forward radiation is similar, the 90 degree ubnt sector looks more like a 65 degree on its pattern. 

the biggest difference in quality from these two brands in the vertical spread of the sector, the ubnt ones are ok, but have more sharp contrasting changes to them.     they also have poorer FTB ratio. if you've got the RF armor shields on the 2.4 they will work out pretty good, we've got a couple running like that.    the ubnt 2.4 can work for the AP without the shield but you'll likely run into a small degrease in uplink modulation when re-using channels.   of course each experiences is going to vary with that as the back lobe pattern isn't constants on the rear of the antenna.

generally we haven't been changing the UBNT antenna with the cambium antenna except new depoloyments.   all of our UBNT antennas are shielded though.


My experiance has been that the F/B ratio is very poor on the UBNT sectors... I wouldn't try to do frequency re-use with those sectors unless they have extra shielding on them. The UBNT Titanium sectors are probably fine, but the standard ones leak way too much out the back.

Hi everyone, I am new to the forum, lots of interesting info. Has anyone used the UBNT Omni with the EPMP, If anyone has info would like to know pros and cons. Thanks

we've had a couple epmp 2.4s on the UBNT omni.     we've secotrized some for load, but omnis in general are subject to the same issues

increased risk of noise, regardless of manufacture. the UBNT omnis have done pretty well for us as far as omnis go, i don't like to use them at all, but sometimes they are the idea situation match for the $$  

lower gain than sectors, generally.  

poorer SNR than sectors generally. 

no ablitly to control tilt, the UBNT omnis have some electrical down tilt, but its still a horizon shot within the antennas spread (back to the noise concern)

2ghz UBNT radio on the UBNT omni compared to

cambium radio on the UBNT omni     we still saw an improvement on overall performance, but it wasn't sharp and changin out 2ghz ubnt radios to 2ghz cambium radios.  

5ghz area, we use them in areas we don't expect much LOS traffic, 20 to 40 subs on a 20mhz channel, no issues. of course the forward power is a lot weaker than a panel.  

1 Like

This project of mine has been delayed until now....got the EPMP AP plastic clip to adapt it to fit the Ubnt sector antenna. Proceeding to put RF-Armor on it. Well, the radio sticking down lower and is larger than the hole cut out in the RF-Armor shielding. Looks like I will have to cut it out to allow for the EPMP case...Not that big of a deal, I guess...Just wanting to reuse this antenna/shield instead of purchasing new.