ePMP 3000 Sector Antenna F/B Concern, May Not Allow MCS 8/9 in Frequency Reuse Deployment

Products ePMP
1

This is more for the Cambium staff, but hoping the answer will benefit us all. 

Looking at the ePMP sector antenna specs, it lists a 30db front-to-back ratio. I have seen no specs for needed SNR for MCS 8/9 for the ePMP 3000, but I am suspecting SNR values will be close to the current ePMP 1000/2000 and inline with current AC gear SNR levels. 

With that said, and we use current SNR levels for most AC equipment, using a 20Mhz channel, a SNR of 29db is needed to reach MCS8 and a SNR of 31db to reach MCS 9. Increase channel width to 40Mhz and a SNR of 32db is needed to reach MCS8 and a SNR of 34db to reach MCS 9. 

With a standard radio tower installation (Rohn 25/45/55 type tower), using freq resuse, how are we to ever reach MCS 8/9 modulation levels for download using the ePMP 3000 sector antenna? I understand we will need near perfect conditions to reach MCS 9, but I would hate for the antenna to be the one factor holding us back from MCS 8/9, when all other conditions are met. 

The ePMP 2000 sector sports a 35db F/B. Mimosa just released a 43db F/B antenna (seems they learned a high F/B for AC use is extremely important). Many guys who have used the RF Element horns rave about how well the horn cuts noise down, and allow higher modulation rates in freq reuse deployments due to their high F/B ratios. 

Just curious of Cambiums thoughts on these concerns.

1 Like
2

I imagine someone more qualified than me will answer, but shouldn't there be some additional dBs of F/B when you consider the tilt of the antenna? I believe 30dB is the value if all APs and SMs are mounted on the same plane (at the same height).

SMs that are closest to the tower and would (typically) have the greatest uptilt to the AP would (typically) also enjoy the most attenuation to the sector facing away. That is they would have a greater F/B than those that are inline with no tilt. These are also the SMs that you would want most to attain MCS 8/9 as they are the closest to the tower and have the strongest receive signal levels.

1 Like
3

 Yes, it is worst case scenario. And if you take all 4 plots for Elevation and lay them on top of each other, you see most areas are 30db F/B or a little better. But unless something has changed with SNR levels, additional tilt will not allow MCS 9 since all areas are close to 30db. This is especially true if you use a 40Mhz wide channel. If the SM is 0.25 miles away and sees its AP at -50 but also see the opposite AP at -82 (32db F/B), the SM can still only demodulate a MCS 8 signal at the best in a 40Mhz channel (and just barely if at all). The SNR level needed and the F/B produced by the antenna will not allow higher modulation (unless Cambium has thrown in some more secret sauce). 

There is a reason guys love the RF Element horns, and why Mimosa designed a 43db F/B ratio antenna for AC use.  You must have high F/B ratios to reach high modulation rates. 

I am a huge fan of Cambium. I am new to being a WISP (1.5 years now) but I have been working with their product since it was Canopy. We mainly used it then for Public Safety. I am definitely not trying to knock them. Just have some concerns. I also know RF and know an antenna can make or break a product.

A few years ago a fire chief bought a $900 50 watt Motorola radio for his personal vehicle. He insisted on a hockey puck antenna, no matter how much we discouraged it. A few months later he came back raising hell about his junior fireman out talking him at fire scenes with their $75 50 watt "Amazon" radios. The difference was they were using gain antennas while he was using a negative gain antenna. A superior product was bested by the wrong choice of antenna. I do not want to see Cambium be the fire chief and Mimosa be the junior fireman due to better antennas. A superior product can only work as well as its weakest link. 

This may not be the case. My worries may be unwarranted. Cambium always seems to up the competiton. I am hoping for this release to be the same. And if the antenna will not cut the mustard, we can find one that will and still carry on with the ePMP 3000. 

Thanks for replying Seth. I appreciate any and all views on this. 

4

Don’t forget to take antenna tilt into this math. The flat ftb is of little relevance in many situations. The antennas rear signal spread on the vertical beam needs to also be considered, it to is almost greater than 30. Tilting these antennas just 2 degrees (which still provides a shot to horizon, you then are approximately-35 to your cpes who are center beam of the facing sector and no longer in center rear beam of the opposite sector. This same method is required to get max modulation from kp sectors on epmp hardware now. If you run them flat, you’ll be stuck at a max rate if mcs13 on the downlink due to the 25db ftb on the 5hv2hv sector. Tilting it 2.5 to 3 degrees now brings the snr up to 29~31 allowing for max mod in the downlink.

5
@Chris_Bay wrote:
Tilting it 2.5 to 3 degrees now brings the snr up to 29~31 allowing for max mod in the downlink.

Downtilt is taken into account. 

The 29-31db of F/B you listed will not allow MCS 8 or 9 (need around 32-34 db F/B for those levels).  Yes, those antennas will allow max dowlnlink modulation for a N system, but will not for an AC system. 

Edit: I see what you are saying here now. You were not exactly using N antenna data for AC radios. You were stating tilt gives a little more F/B. Yes, but even with tilt, the F/B for the ePMP 3000 sector is a little too close for comfort for me with the SNR/CINR levels referenced in technical papers for MCS 8/9 modulation to occur. I have seen no stated SNR/CINR from Cambium for the ePMP3000 series. I do see the RSSI levels are very similiar to the N radios. And if SNR levels are the same for the AC radios as they are for then N radios, then moving up to MCS 8 would be a 4db increase in SNR and MCS 9 would be a 6db increase in needed SNR over N gear at MCS 7 (15). 

6

Cwb, please re read my post. I sited a kp perforamnce antenna, the 2hv5hv sector whos front to back is a measley 25 when flat. You get 4 to 6 more db with proper tilt. The cambium sector by plot will be similar. Sector tilt is EXTREMELY important for optimum propigation in all systems ptmp sector based systems. You’ll never find a carrier sector at 0 tilt.

7

Yea, I noticed I misread your post. Sorry about that. I corrected it above. I am also very aware of antenna tilt, as I have been in radio system design, installation, and repair since since I was a young child. 

We got away from KP because of their poor F/B performance at 2.4Ghz. Used a angle level to set sectors at 3 degrees at that. The Cambium/Mars sectors work much better. 

8

Maybe I should have just asked what was the SNR values needed for MCS 8 and 9 for both 20Mhz and 40Mhz channels for the ePMP 3000 line. 

Can someone with Cambium provide the SNR or CINR values needed for the ePMP 3000 line to reach MCS 8 and 9 in 20Mhz and 40Mhz channels?

Just knowing these values can help clear this all up. 

9

Mcs8 is max in 20mhz mode, judging from the spec sheets from the sensitivities reported, looks like 28db snr for mcs 9, and 25 for 8. Those are estimated based on the receiver details listed.

I don’t see them being more than 30, cambium wouldn’t spec an antenna that would hamper performance to such an important product…

10

Chris, my basic issue that I would like some clarification on is Cambium is releasing a product that needs a higher SNR/CINR over their current product for maximum modulation but is releasing an antenna with less F/B than the current antenna. Their competition just released an antenna for MU-MIMO AC systems which has 43db F/B. I do not think the competition did it just for fun. I think they did it out of necessity.

As for only MCS 8 at 20Mhz, Cambium claims, and Ninedd shows with screen shots, the 300 series peforms at MCS 9 at 20Mhz even though spec sheet shows MCS 8 as max. Maybe that is just in PTP mode. Cambium can enlighten us if MCS 9 is also available in 20Mhz PtMP as well. 

For SNR they have to at least be close to the AC standard specs which is MCS 8 29db and MCS 9 31db in a 20Mhz channel, and MCS 8 32db and MCS 9 34db in a 40Mhz channel. Unless Cambium came up with some more secret sauce for the ePMP 3000, this rules out the figures you estimated. 

Also, the current ABAB Frequency Reuse Manual specs MCS 7 (15) in a 20Mhz channel at 28db. N technology is the basis for AC. All that was really added was MCS 8 and 9 and 80Mhz channels. Going further with that, if the ePMP 3000 series is very close to SNR/CINR values of the ePMP N devices, and we carry on with the chart in the ABAB Freq Reuse Manual, then for a 20Mhz channel MCS 8 would be 32db and MCS 9 would be 34db. 40Mhz, MCS 8 34db and MCS 9 37db. This is just how the math adds up. 

If you noticed, I purposely left out 80Mhz channels because F/B values would need to be close to 37-40db to allow full MCS rates (and because I would personally never use a 80Mhz channel on a sector). This does bring us back to the antenna Mimosa designed for AC MU-MIMO use with 43db F/B. I think they realised it had to be done to achieve the full potential (MCS rates) of an AC frequency reuse deployment. 

I am hopeful Cambium would not release an antenna that will hold its product back. But, before I go dropping money on 12-18 of these with sectors and smart antennas, I would like to make sure that is the case. In my time in communications (most of my life) I have seen some top notch manufactures make some wrong decisions in the name of saving money or to stay competitive price wise. 

11

Let's look at required F/B for ePMP 3000  antenna from other side.

We do not see required for 256QAM5/6  ( MCS9)  SNR/Sensitivity level in ePMP 3000 spec. But we see that in Cambium  PTP550 spec. I suppose ,  these parameters are equal or  similar for ePMP 3000.

So  Cambium  PTP 550 802.11 ac wave 2

MCS\Rx Sensitivity

20 MHz 40 MHz 80 MHz

Lowest MCS -90 dBm -87 dBm -83 dBm

Highest MCS -66 dBm -62 dBm -59 dBm.

 Let's  look  how  system works  at   most popular 20/40 MHz channel bandwidth.

If we set in ePMP 3000 Target RSSI level -60 dBm,  it will be enough to have Uplink  256QAM5/6 ( MCS9)  in 20/40 MHz ( without interference).  Antenna F/B =30 dB reduces in ABAB Frequency Reuse scheme secondary interference from opposite sector CPEs from  -60 dBm  to  -90 dBm.  So this interference  does not not impact to system  and  antenna F/B  30 dB is enought to work in ABAB  with FR without problem.

Iinteresting   question , how it is possible for Cambium 802.11 ac wave 2 system to have,  for example, -66dBm Sensitivity  for 256QAM5/6  at  20 MHz channnel bandwidth? I suppose,  required SNR for 256QAM 5/6 in 20 MHz not less then 32 dB.

Suppose ,  that Noise Figure= 3 dB, and  termal noise  Power Spectral Density (PSD)  in receiver   is  -111   dBm(W/MHz), In 20 MHz it is -98 dBm. This value  is also frequently  (  is considered that it is  wrongly) called Noise Floor (NF).

So,  Sensitivity at 256QAM 5/6 at 20 MHz= Noise Floor  ( -98 dBm) - SNR ( 256QAM 5/6 =32 dBm)= -66 dBm. 

For  QPSK 1/2 (Cambium  lowest MCS )   Sensitivity at 20 MHz  is -98dBm  - 8 dB = -90 dBm.

So  Cambium 802.11 ac wave2  has lower termal Noise  floor then  Cambium 802.11n ( due to less  value of parameter, that is called Implementation Loss).  It allows Cambium ePMP 3000 to work  ( theoretically ) at 256QAM 5/6 at the simular RSSI level as ePMP 1000/2000 at 64QAM5/6 ( with different required  SNR/CINR).  And antenna F/B= 30 dB  for ABAB Frequency Reuse is enought for both system ePMP 3000 and ePMP 1000/2000.

1 Like
12

I left this post alone until I could get input from a Cambium rep since it seemed there were misunderstandings about my point. The post above proves this. It spun off into the AP, which is 4x4 when I was talking about the SM's which are 2x2. The SM's are also not mounted hundreds of feet in the air, hearing noise from all over. Apples were being compared to oranges. 

In any event, a Cambium distributor/dealer sponsored a webinar on the new ePMP 3000. I asked what SNR/CINR was needed to acquire MCS 8/9 in a 20Mhz channel. The answer was, "in the neighborhood of 34-36db". I personally feel the presenter was shooting a little high to be on the safe side. Even if it is lower, it is not by a long shot. The question and answer portion ended before I could ask how we can reach MCS 8/9 in a frequency reuse scenario with a sector antenna that sports a 30db F/B. This also makes the fact that Mimosa came out with a sector antenna for 802.11 AC frequency reuse that specs a 43db F/B more significant. 

The Cambium rep also displayed real world test data where 20Mhz sectors were passing around 200Mbps in MU-MIMO. Pretty  awesome!!! I was unable to ask if these tests were in a frequency reuse scenario or what the RF enviroment around the SM's was like. 

But.....since the AP was brought up earlier, we can see that using the SNR/CINR given by the presenter, a target RSSI of -60 will not allow you to reach upload MCS 8/9 in almost any installation. This is why I did not worry about F/B when dealing with the AP, the surrounding noise makes it a moot point. You can not reach a high enough SNR/CINR. If noise floor is at -90, you will need a target RSSI around -56 to reach MCS 8/9 for upload. A noise floor of -85 will put you at a -51 target RSSI. How many of us have noise floors that low at our AP's?