I am doing a comparison of the ePMP3000 system against various competitor products. One thing that is alarming is the difference in sensitivity of the CPE (SM) receivers. Comparing the claimed performance for 20MHz channel width, BPSK (MCS0) one sees the following:
- Force 300: -87dBm
- Mimosa C5: -93dBm
- Mikrotik LHG HP5: -96dBm (hard to believe since this would imply around 2.5dB receiver noise figure)
- UBNT Powerbeam M5 AC: -96 +/-2 dBm so lets assume -94dBm
If I assume SNR for MCS0 is 6dB, then it means the Force300 has Noise Figure of around 8dB whereas the competitors are below 4dB.
A 6dB difference in sensitivity is pretty significant in terms of link budget as it implies competitor CPE can operate at the same MCS as Force300 when at double the range.
Am I overlooking something or is Cambium just have a fairly poor NF receiver in the CPE device? (ePMP 3000 APs look better as they have -92dBm sensitivity for BPSK, 20MHz).
I’d suggest a side by side live comparison to find your answer.
I’ve used on of the other brands you mentioned and we intentionally run the very weak in the lab to see how they handle weak powers, right down to claimed sensitivity. It didn’t make it. Cambium did. Both radios stopped at the same point.
I wouldn’t attest it to a noisy rx, I’d say more like realistically achieved. The other device we tested could see the otherside at 3db above marked sensitivity but was unable to pass the modulations expected data rate and maintain a link at that level.
Thanks Chris. Yes indeed another explanation is that all the other vendors over-state their actual sensitivity numbers. This would tend to be supported by your comparitive tests.
However, the ePMP3000 AP sensitivity is more in line with other competitors so I assume Cambium are using a better LNA in the AP than the SM.
Would be great if one could take published specs at face value but they are all questionable, including max Tx power, F/B ratio, etc. Now everyone has to do their own comparitive tests and find out the hard way in the field when issues arise eg. due to poor antenna F/B ratios or higher sidelobes than indicated ie. stuff that is difficult to measure without a full antenna test range.
Couldn’t agree more. Also creates doubt by default in just about anything new in wireless technology anymore.
I suspect the other manufactures list specs for tests conducted that do not include MRC (Maximum Ratio Combining) and DSP (Digital Signal Processing). This will give a 3-6db lower spec. Basically, they are listing values for a single chain input power at the antenna port prior to its power combing with the other chain's power. Once again, this is just my suspicion based on the RSSI comparisons being close to 6db in difference. List specs this way also looks better on paper.
Here is my opinion on the Cambium AC product, and Cambium in general. They do not fudge specs. They have definitely put a lot of time and effort into this product to make it top notch. Cambium has done something that as far as I know, no one else has done. They were able to get 256qam 5/6 (MCS 9) working in a 20Mhz channel. That was not even part of the AC specs because one, it is hard to accurately produce and transmit such a tight cluster in that bandwidth, and two, hard to decode it as well. Cambium pulled the rabbit out the hat.
Some may say this in conflict with an earlier post I made about the ePMP 3000 sector antenna. It is not. I do have concerns it will not allow enough F/B ratio (listed at 30db) for 256qam rates. Especially for 40Mhz channels. I am basing this on known SNR rates for 256qam (MCS 8 and 9), which is over 30db. The PMP 450 line specs a SNR of 32db for 256qam, the PMP 450 sector antenna is rated at >35db F/B. I am just concerned the 3000 sector antenna could -possibly- hold back the full potential of a product they have put so much effort into making. At the same time, I appreciate the data Cambium releases on their antennas. It allows me to make informed decisions. They give more information on their plots than their competitors. One of their competitors does not even release F/B information on some of their antennas. That is very telling to me. Tells me to keep away from it.
At any rate, I do not think you will be dissapointed if you go with the Cambium product.
....The other device we tested could see the otherside at 3db above marked sensitivity but was unable to pass the modulations expected data rate and maintain a link at that level.
Here is something I missed earlier, which may support my theory. In this test both chains were picking up signal (hopefully evenly) at the antenna port and then combining giving a 3db (2 times) increase over a single chain. What is interesting is it seems a link was established but could not hold or pass modulation. Seems tests (not the test performed by Chris) were conducted to say, "There we go...MCS 0 was established (which it was if a link was established at all), mark it and subtract any values added by MCR". This would leave you with with approximately a 3db lower value, which is the same as the coax input port single chain dbm level.
Similar to eAlign. You have chains 0 and 1. If both show -63 (chain 0 = -63, chain 1 = -63), then the combined RSSI equals -60.
Cambium gives Sensitivity ( and Tx power also) for dual chains. So -87 dBm for dual chains means -90 dBm for single chain. UBNT and Mikrotik give these parameters for single chain.
SNR/Sensitivity depends on BER. Cambium may give Sensitivity ( I suppose) for BER =1.0 E-6, or -E5.
Other vendors may give this parameter for BER 1.0E-3. Difference may be more then 3 dB,