ePMP 3000 AP and dual horns

Cambium makes a dual horn 60 degreeadegree that allows MU-MIMO to keep working. 

I just installed one recently in a high noise environment where there are many other WIsp nearby and it's been great so far. Granted, we currently only have 1 SM connected (about to be many more) but it's been working great so far and the interference has been reduced significantly. 

1 Like

I am aware of that antenna. In this case for us, I need 180 degrees of coverage so the thought was two 90 degree horns with no overlap.

Or how about 3 and keep Mu-Mimo?

This is a micro-pop. No room for more than one AP. 

1 Like

I believe I have read that if you put the radio in split-sector mode with two horns running back-to-back you may still be able to utilize mu-mimo.  Perhaps Fedor or another Cambium engineer could chime in?

Deleted (my info was incorrect)

I hate to add confusion. But, I'm almost positive we have 1 horn connected to 1 & 4 and the other to 2 & 3. I'll try to get one of our guys go by and confirm. IIRC, that was based on a user's post on this forum a while back. We played with a couple asymmetric horns on a 3000 pointed the same direction. It seems to work a little better than the 2000 for a couple SMs but about the same for others so we didn't go any further with it.

It would be nice for Cambium to provide some sort of answer.

1 Like

@Au Wireless wrote:

I have a 3000 AP on a MU-MIMO omni (KP Perf) that is not working as we hoped. I am pretty sure the reason is most of our clients are below the vertical beamwidth of the omni.

So, probably not normally a good idea with an omni but since you seem to only need 180° coverage below the electrical downtilt of the omni... have you considered tilting the Omni ?

So, we are looking at forgetting the MU-MIMO and connecting two RF Element 90 degree horns we have in the warehouse. We would be using these in a 180 degree pattern (non overlapping).

Couple questions:

1) What ports on the 3000 AP will these two antennas connect to?

2) While not MU-MIMO, does the AP act like two APs (one on each antenna)? 

3) How is the bandwidth split between the two non-overlapping horns with a 3000 AP?


According to Cambium Dmitry: (Note, the RF Elements Twin Horn Bracket s is two horns pointing the same direction not two horns 90° off of each other)


1. If you use a pair of horns with the official RF Elements "Twin Horn Bracket" you will not see any MU-MIMO. Although besides MU-MIMO system will work perfectly fine.

2. If you use a pair of horns looking in the different directions you will not see MU-MIMO unless you do the following:

a. Connect one horn to chain 0 and 1, connect second horn to 2 and 3.

b. Update software to 4.5-RC and pick Omni antenna option under the RF settings.

2. If you use a pair of horns looking in the different directions legacy non-11ac subscribers most probably will work with single stream rates.

3. Spectrum analyzer will work only for the chains 2 and 3.



So, I obviously do not get how MU works on the 3k AP. If you connect 0 and 1 to horn pointing North and 2 and 3 to horn pointing South and you have an AC client to the North it will work normally but a N will only see a single stream ? So no MiMO at all for N but AC client still sees 2 streams ? How does that work ?   However looking at what others have posted if you connect North to 0 and 2 and South to 1 and 3  then MiMo works for.. I assume N and AC but no MU ? 

Or am I misunderstanding what is meant by "single stream" ?

I thought about tilting the omin but since the customers are spread in 180 degrees off of it, that will screw up the alignment of the two chains for any users not directly in line with the direction I tilted it. I think it will make it worse.

I don't think MU-MIMO works at all for non AC clients anyway so I sort of understand the point Dmitry makes. We are 80% AC clients and the few N clients still on this AP are due for replacement next month anway to AC so we should be ok.

From what I am reading, I connect chain 0/1 to one horn and point it south west. The other horn connects to chain 2/3 and I point it south east. That gives me 180 degrees (with no overlap) from east to west and I get MU-MIMO on AC clients....

That's how I read it but one of us, probably me, isn't understanding "single stream rates" for N radios.  It sounds to me like you read that as no "MU MiMO" for N radios (which they can't do anyway so say say it that way)  while I read it as no "MiMo" e.g. single stream = single chain for N radios.  

I don't know exactly what he means when he says "single stream rates".

Here's some data from our antenna change. We started with a 3K AP on the KP Performance MU-MIMO omni antenna. This is mounted on a mini-pop house serving 21 clients.  We were not happy with groupable SM numbers, the MU-MIMO performance or the RSSI numbers. I suspected one of our problems was most of the customers were below the optimal vertical beam height of the omni. The AP sat up on a hill and the clients were all below.

Here is our Monitor screen with the Omni (sorted by IP):

After changing to two 90 degree (symmetrical) RF Element horns pointed off axis (to get 180 degrees of coverage), here is the Monitor screen:

Note the increase in groupable SMs and an across the board increase in RSSI and SNR for all SMs.There are still 5 old N clients on this AP (the last 5 in the list) that we have not gotten to changing out yet.

I found the two SMs that had 13 groupable interesting...  Here is a map of the AP (top) and all the SMs. The two with 13 groupable are highlighted.

Interesting that those are on the far east of the pattern. The 5 N clients are grouped in the middle.

For visual reference, here is a shot from the AP to get an idea of height over SMs:

All devices are running 4.5 release and the AP is set to "split sector / Omni" for the antenna.  I'd like to think after upgrading those last 5 SMs, we can work on getting some better MU-MIMO groupings. Interested to hear some feedback from Cambium on those numbers, especially the two showing 13. 


That is great !  Just the kind of thing I wanted to see !

So did you end up connecting the horns according to Cambium Dmitry "one horn to chain 0 and 1, connect second horn to 2 and 3"  or the way a couple of others have said works "one horn to chain 0 and 2, connect second horn to 1 and 3."  ?

If you did it 0,1 and 2,3 per Dmitry what are you showing on the chains for the N radios (both ends) ? It looks like you're getting full modulation on the N radios (except for Uplink which also has lower SNR ? Using DFS band ?) so yeah. Sorry to pester you on this but, like you, we have areas we need 120° - 180° coverage and want to use 3k AP but we still have mostly N with a few AC clients (we are slowly changing over to AC).  I was all set to roll out something just like you have set up but the whole thing about N clients only being "single stream" has put that idea on hold until I can figure out what "single stream" means.

I followed the Dmitry where 0/1 go to horn 1 and 2/3 go to horn 2. I kept 0 and 2 on vertical and 1 and 3 on horizontal. Not sure it matters but I wanted them to at least be the same on each.

Those N radios are at full modulation. We artificially lower the max MCS on the N radio uplink. I am not using DFS but we are going to move to DFS now that I have a much cleaner signal. I could not use DFS before with the omni - but I do believe that is because we were under the beam on some of our SMs and the reception was off as a result.

When I speed test an N radio, it is getting nearly the same throughput (AP to SM) as the AC radios are.

AC Radio:

Screen Shot 2020-08-10 at 2.14.14 PM.png

N Radio:

Screen Shot 2020-08-10 at 2.14.45 PM.png


Great info !  Thanks for taking the time to answer so thoroughly.

1 Like

Great Post!  I have a small deployment situation where this may be very applicable... could I get away with using a 3000L?  Also what kind of range/distance do the 90 degree horns push?

The 3000L only has two chains, so you can only use one antenna with it.  

The 90* symmetrical horn only has 9.6db gain at the high end, so that will pretty dramatically limit range from an AP RX signal perspective.

1 Like

In DFS, we limit our range to about 3/4 mile with the horns. On full power channels, we have clients 3 miles out with no issues. We don't go further than that due to terrain. While the horns have lower gain, I believe they more than make up for it with a cleaner signal, no lobes and a better pattern.

1 Like

Connecting a 3000 to two horns pointed in different directions so you can do MU-MIMO also only has 2 chains since no client radio can see both horns at once. I would think that normally (if the 3000L wan’t such a piece of junk) that using 2 3000L’s and 2 horns would be better than 1 3000 doing MU with 2 horns.

For more gain the RF Elements 90° HG3-TP-A90 5GHz Asymmetrical Horn is 16dBi but quite a bit bigger/heavier/pricier than the Symmetrical horn.

1 Like

An antenna configuration based on non-overlapping sectors is called Split Sector.
One horn should be connected to ePMP 3000 chains 0/1 while another to chains 2/3. Polarization order is irrelevant,
As described horns are connected correctly. But MU grouping numbers are inconsistent with expectations. MU groups F300s serviced by different horns. Accordingly it is expected that F300s served by the same horn have the same grouping number…
Please ensure that AP Antenna is configured as Omni/Split Sector.


Agree with you. the 3000L is a very bad radio. Incredible but the ePMP2000 is better and I mean the quality of its hardware. The 3000L is very easily damaged, after replacing 3000L with ePMP2000 we didn’t go back to the towers :slight_smile: