SNR / Sensitivity per MCS vs reality


#1

Hi all,

I’m looking at the SNR and RX sensitivity specifications for the A5 products http://ap.help.mimosa.co/access-point-faq-snr-mcs and http://ap.help.mimosa.co/access-point-faq-sensitivity-for-each-mcs-index and the figures seem a little low when compared to other vendors. Now that could be great news, but I’m wondering if anyone has any real world data with A5/C5’s to confirm the numbers in the FAQ.

For example, Cambium reckons on their ePMP you need a SNR of 28 and RX of -70 for MCS15. Mimosa specs show SNR of 22.5 and RX of -75.5 (for MCS7).

UBNT reckons for MCS9 on a Rocket AC you need RX of -65 (can’t find SNR specs). Mimosa specs show RX of -70.5 for the same.

Rich


#2

That thunder you hear is the fanboys running to stick up for Mimosa…

Mostly kidding, but for what little it’s worth, a vendor-sponsored chat room is unlikely to yield objective information.

Please prove me wrong, out there!


#3

By definition, you need -28dBm for MCS9. The question is where your base noise level is at to reach that. The Rocket AC will have a higher base noise figure, probably -93dBm. The Mimosa has a better chipset to start with. My guess is that the amplifier circuit is also probably less noisy. That would account for the difference.


#4

Thanks Rory. I know you’ve documented a lot of the results you’ve had with the A5/C5 gear. Are these numbers typical of what you are seeing in production - i.e., a 28db SNR will give you MCS9; 22db will give you MCS7 etc?


#5

I would suggest that you look at this from a different point of view… when comparing to other mfg. Mimosa A5’s have a couple of very unique things going for them…

  1. This was discussed / and pointed out on a facebook post… Having circular polarity, the FCC allows for the math for rf power allowed calcs to be advantageous for such a configuration vs traditional linear polarity chains.

  2. The C5’s are able to deal with hot signals, at levels which will bring down most of the other part 15 radios.

In view of these two technical facts, one would expect them to have better performance (SNR & RX/TX rates).


#6

Hi Faisal. Thanks for the reply.

I’m not in the US and I’m unsure if circular has any impact on the EIRP calculation here, but I’ll check.

As for the C5 handling hot signals, this is at the other extreme of what I’m looking for. I’m interested in whether the SNR values in the data sheet match users experience of achieving a given MCS - i.e. When the signal is opposite of “hot”.

Rich


#7

I understand the question you are asking… The Mimosa folks have a tendency to publish accurate and conservative specs… Your question is valid an appropriate, however not very relevant in the real world… where there is interference. We find that at lower signal levels the affect of interference starts to show on achieving a given MCS, as such we don’t try to make low SNR links.


#8

FWIW,
I am currently working on a site migration from Ubiquiti M clients to an A5-18… While this is not a A5 to C5 I can tell you from the initial set of client that we moved over, the Mimosa MCS / SNR / Signal table appears to be valid and possibly a bit understated.


#9

Thank you Faisal. That’s exactly what I was interested to know.

Rich


#10

“By definition you need -28dBm for MCS9”? What definition? For single or dual chain? For Mimosa or in general?
And you mean signal level (-xxdBm) or did you mean -28dB (as for s/n ratio we’d use dB’s)

I found some tables online and the best so far I saw is 31dB for MCS9 in 20Mhz ac channel. With a background noise level of -105dB for instance this would give a needed signal of -74 to get MCS9.

Well, don’t know but my Mikrotik links need at least -50dBm to get MCS9

So in fact, a lot of confusion or at least each vendor its own requirements?
(UBNT, Mikrotik, Mimosa, eCambium, they all say they are the best…)


#11

You better read here:
http://ap.help.mimosa.co/access-point-faq-snr-mcs