ePMP 1000 -> 2000 test results

Yesterday we swapped out an ePMP 1000 AP for a new ePMP 2000 AP - mostly due to co-location noise on the tower.  In both cases, we are using a RF Elements 60 degree horn antenna.  For the 1000 AP, this antenna was part of the TwistPort product line. For the 2000 AP, we switched to a Carrier Class horn since they don't make a TwistPort. The specs on both are about the same - with the Carrier Class having 1 less dB of gain and there will be a little cable loss since it uses 12" RF cables and the TwistPort has no cables. 

Here are before and after screen shots from the Monitor -> Wireless page of the APs. A couple of the uplink MCS numbers are probably inaccurately low since I understand they drop down when there is no traffic passing across the link.

1000 AP with TwistPort:

2000 AP with Carrier Class horn:

Since the antenna specs are nearly identical, you can see a pretty decent improvement in signal quality moving to the 2000 from the 1000.  We are not yet using a beam forming antenna on this AP but may add one to improve upload speeds in the future. I have not used a beam forming with an antenna different from the Cambium 120 sector so I'd be interested to hear stories of those that have. Should not matter, I suppose.  We have been very impressed with the RF Element horns and are pretty happy to see how well they work with the 2000 AP - since this is our first deployment with a 2000.

1 Like

I couldn't help noticing that the ePMP2000 identifies clients as being about 10% closer that the ePMP1000 did.

I also notice that while downlink and uplink RSSI show improved, downlink SNR seems relatively unchanged and uplink SNR is lower.  

Are both screenshots of the same frequency in use?

j

Both screen shots are the same frequency and same output power from the AP.  That RSSI / SNR issue is interesting...  Perhaps Cambium can weigh in on that.  I have also noticed the distance issue between 1000 and 2000 APs. When we move clients between the different APs, their distance does change. 

We are getting slightly better throughput with the 2000 AP. It is mounted in the same exact location as the old 1000 was - everything else is identical.

We're using a Ubiquiti AM-5AC22-45 45deg sector antenna with our ePMP 2000 AP, we also tried using a BSA with it. The BSA is designed to mechanically and beam width-wise work best with the new Cambium OEM 90/120deg sector. There have been a number of updates to the ePMP firmware lately that have further optimized how the BSA is used with 3rd party antennas. As of the most recent stable firmware update 3.0.1, the algorithm that is used to determine if there's an advantage over using the BSA over the sector antenna, we've found the BSA is rarely used anymore. Of the 30 SM's on this sector, only 4-5 of them are using the BSA. I believe this is due to the tight beamwidth of the sector and the additional gain of the sector. I might even go so far as to say, if you plan on using tight, 3rd party sectors, do not buy a BSA.

There was some discussion (http://community.cambiumnetworks.com/t5/ePMP-2000-and-1000/RSSI-lower-level-starting-from-firmware-3-0/m-p/60260#U60260) that the newer 3.x software calculates RSSI differently (more accurately?) than previous versions, which could explain the differing RSSI but same SNR between the 2000 and 1000.  What version was the 1000 running?

Rich

Both of them are running 3.0.1 as are all SMs.


@Eric Ozrelic wrote:

We're using a Ubiquiti AM-5AC22-45 45deg sector antenna with our ePMP 2000 AP, we also tried using a BSA with it. The BSA is designed to mechanically and beam width-wise work best with the new Cambium OEM 90/120deg sector. There have been a number of updates to the ePMP firmware lately that have further optimized how the BSA is used with 3rd party antennas. As of the most recent stable firmware update 3.0.1, the algorithm that is used to determine if there's an advantage over using the BSA over the sector antenna, we've found the BSA is rarely used anymore. Of the 30 SM's on this sector, only 4-5 of them are using the BSA. I believe this is due to the tight beamwidth of the sector and the additional gain of the sector. I might even go so far as to say, if you plan on using tight, 3rd party sectors, do not buy a BSA.


If you are not using the BSA, what advantage does the epmp 2000 have over the epmp1000?


@Eric Ozrelic wrote:

 There have been a number of updates to the ePMP firmware lately that have further optimized how the BSA is used with 3rd party antennas. As of the most recent stable firmware update 3.0.1, the algorithm that is used to determine if there's an advantage over using the BSA over the sector antenna, we've found the BSA is rarely used anymore. Of the 30 SM's on this sector, only 4-5 of them are using the BSA. 


When we first deployed an ePMP2000 a couple weeks ago,  20-some clients connected (splitting load with an existing sector) and 80%-90% were using the SmartAntenna at first.  Then it dropped to about 50%-60%.  Today I logged in and looked, and ZERO are using the Smart Antenna.  On 3.0.1 firmware, it tells me the antenna is up and running fine but NOBODY out of 20 clients was using it.  5 mins later and two of them are back on the SA.  An hour later we have 5.  Very strange.

j

I can confirm similar behavior on my 2000 AP with the BSA. Running 3.0.1, 3 of my 15 SMs are using the BSA.  Strangely, those 3 are on the upper and lower limits of the vertical beam width.  Everything right in the sweet spot of the beam is on the sector.

Just wondering if you prefer the  RF Elements over the standard antennas for 1000 deployments?


@Chris_2 wrote:

If you are not using the BSA, what advantage does the epmp 2000 have over the epmp1000?


The ePMP 2000 has special 'intelligent' filtering called 'HyPure' on both the TX and RX paths which helps improve performance in environments where the AP is subject to high levels of out adjacent channel interference. The BSA helps with co-channel interference on the uplink.


@Chris_2 wrote:

@Eric Ozrelic wrote:

We're using a Ubiquiti AM-5AC22-45 45deg sector antenna with our ePMP 2000 AP, we also tried using a BSA with it. The BSA is designed to mechanically and beam width-wise work best with the new Cambium OEM 90/120deg sector. There have been a number of updates to the ePMP firmware lately that have further optimized how the BSA is used with 3rd party antennas. As of the most recent stable firmware update 3.0.1, the algorithm that is used to determine if there's an advantage over using the BSA over the sector antenna, we've found the BSA is rarely used anymore. Of the 30 SM's on this sector, only 4-5 of them are using the BSA. I believe this is due to the tight beamwidth of the sector and the additional gain of the sector. I might even go so far as to say, if you plan on using tight, 3rd party sectors, do not buy a BSA.


If you are not using the BSA, what advantage does the epmp 2000 have over the epmp1000?


An important Cambium note on this topic. We just released SW version 3.1 on our beta site. As we mentioned previously we have been making some algorithm changes on our decision to choose sector vs BSA. The latest 3.1 captures changes where we are looking at per packet interference as well as tput to determine when to pick BSA vs Sector. Overall, you will see more BSA choices starting in 3.1. Please give this a try and post your feeedback.


@newkirk wrote:

@Eric Ozrelic wrote:

 There have been a number of updates to the ePMP firmware lately that have further optimized how the BSA is used with 3rd party antennas. As of the most recent stable firmware update 3.0.1, the algorithm that is used to determine if there's an advantage over using the BSA over the sector antenna, we've found the BSA is rarely used anymore. Of the 30 SM's on this sector, only 4-5 of them are using the BSA. 


When we first deployed an ePMP2000 a couple weeks ago,  20-some clients connected (splitting load with an existing sector) and 80%-90% were using the SmartAntenna at first.  Then it dropped to about 50%-60%.  Today I logged in and looked, and ZERO are using the Smart Antenna.  On 3.0.1 firmware, it tells me the antenna is up and running fine but NOBODY out of 20 clients was using it.  5 mins later and two of them are back on the SA.  An hour later we have 5.  Very strange.

j


Joel,

An important Cambium note on this topic. We just released SW version 3.1 on our beta site. As we mentioned previously we have been making some algorithm changes on our decision to choose sector vs BSA. The latest 3.1 captures changes where we are looking at per packet interference as well as tput to determine when to pick BSA vs Sector. Overall, you will see more BSA choices starting in 3.1. Please give this a try and post your feeedback.

Has anyone done any testing to determine if the radios perform better when using the BSA vs Not using the BSA ? 

I have downgraded all my EPMP2000 full systems to 3.0 since it is the better working  BSA.

All the CPE are are on 3.1 final.

With 3.1 and 3.1RC the AP have some problems.

So guys..generaly..is good to pay for AP2000  ? or we can use 1000?

its better than UBNT?

Little off-topic, but as far as my experience is concerned and otherusers have suggested:

  • ePMP 2000 is best used in noisy environments. In clean environments an ePMP 1000 will perform almost the same. So spend the big bucks where necessary. Keep in mind a 2000 will have a higher power budget; this is important when going on solar/eolic power.
  • IMHO, if 'better' refers to stable, predictable performance then yes, ePMP is better than UBNT counterparts (M and AC lines). If 'better' refers to the use of lastest cutting edge hardware then no, ePMP isn't any better than UBNT. But once again, typical use of these devices is related to business deployments, where stable and predictable behaviour is much more wanted than higher theoric bandwidth that is more often than not lost in erratic, confusing, non-deterministic results of the not-so-much-tried-forever-beta devices.

Its gives me the sense....its gets me angry for long time that MK and UBNT making non-usable beta products..and we all here are betatester in live traffic!!!!

I was thinking between Rocket M5 Air-Prism with Prism sector and AC clients .... or cambium 2000 with smart antenna.

cambium Acces point will cost little higher, over 1000€ (unit + sector+ smard antenna) but... mabye I will give a chance to cambium :D :D

1 Like

That is the same path I took not so far ago, and so far I'm very happy with the change.

Synced APs are a must if you do care about your user's experience, and it seems that although their new hardware is very capable of TDD syncing, it's not expected a working version before achieving stability on current firmware. So maybe Q3 2017? Q4? Who knows.

I'm not bulling UBNT in any way. In fact I do like their routers and their airfibers X, but IMO their ptmp product lines aren't up to the task: the farther the SMs are from the AP the worse. It's quite literally a 'last mile' product.

Regards!

I ordered AP 2000 with smart antenna and some clients modules. Will see.....