I'm wondering if this is a firmware thing or a just how AC is thing. When we swap out a F200 with a F300-25 we normally get pretty much the same RSSI +/- a dB but the F300-25 has less SNR. This isn't an issue when replacing F200's at -65 RSSI and 31dB SNR to a F300-25 at -65 RSSI and 28dB SNR or so but I'm finding that where a F200 would run just fine at -76 and 22dB SNR the F300-25 at -76 RSSI is only getting around 16dB SNR and the customer starts complaining about buffering and slow speed tests.
We haven't changed out a lot of F200's with F300-25's just yet but the F300-25's constantly have lower SNR at the same RSSI than the F200's when connected to N or AC access points. Same with F180's and F300-16's .Installed a new 3000AP , replaced 3 F180s with F300-16's. The 180's were around -55 and 43dB SNR while the 300-16s are -55 and 33dB SNR and all we did was slide the 180 off its bracket and slide the 300-16 on (sweet! Sure wish this worked with F200 to 300-25 swaps...). At these levels it's not a problem as all the 300-16's are maintaining MCS 9 up/down but, is this expected behavior of N vs AC ?
Yes, we noticed this when we first started trading out 180's and 200's for 300-16's and 300-25's. We noticed the same RSSI, but lower SNR. This was connecting to N AP's. We also had lower MCS values due to this lower SNR value. Once again, this was when connecting to an N AP.
Connecting AC SM's to AC AP's (300's connected to 3000's) we still see lower than expected SNR values (based on previous N devices). But,...we reach the expected MCS values, so we have not been concerned. Also, I think this is all expected as Sakid said in a webinar quite some time ago that the 3000/300 had better C/I performance than it's N predecessors (2000/1000). I think this increased C/I performance is only true when AC connects to AC (300 to 3000).
Also,Looking at the spec sheets for the 200 and 300-25, it looks as though a lower SNR is expected. 200 spec sheet shows in a perfect environment (no other signal or noise on same freq as the 200), it needs a minimum signal of -92dbm to hold MCS 0. The 300-25 shows it needs -89dbm. That is a 3db difference, or in layman terms, the 300-25 needs twice the receive power of the 200 to connect at, and hold, MCS 0.
I just logged into a 2000 AP. I have a 200 with a -64 RSSI and SNR of 33. I also have a 300-25 with the same RSSI but a SNR of 27, a 6db difference. These SM's are in the middle of nowhere with no interference...both customers have cnPilot R190 routers, so no 5Ghz self interference as well.
The F300-25 was previously a F200. The 200 was replaced with the 300-25 last summer because of lightning damage. The RSSI stayed the same, but SNR and MCS levels dropped. The 200 was reaching MCS15 over 90% of the time for DL. The 300-25 that replaced it never sees MCS15(7). Only reaches MCS14(6) 0.7% of time and MCS13(5) 84% percent of the time. This same RSSI and SNR value connected to a 3000 would most likely be at MCS8 most of the time (comparing to our other 300's connected to 3000's). Since a 3000 will be going up at some point, I am not worried about this.
We have done a fair amount of swapping since that post and it doesn’t matter if it is a N or AC AP or pure AC clients the AC radios always have significantly less SNR than N radios at the same RSSI. The less RSSI the greater the difference ( e.g. at -55 the difference might only be 3dB while at -75 the difference is -10dB) so it gets exponentially worse.
We are very rural here and have very low background noise for the most part , especially at 5Ghz. As a result we have F200’s out there at -80 RSSI that work just fine while the F300’s will not work at all at -80 and have problems even at -75 .
Also does not appear that the F300 radios get to use as high a tx power (since the client is PTP) as the N radios do making them even worse ( e.g. on 3000AP/Sector the F200’s are all tx power 29dBm while all the F300-25’s are 25dBm meaning all the N’s have significantly higher RSSI / SNR at the AP than the AC’s).
I assume the AC client radios are not allowed as much TX power as the N radios for regulatory reasons (FCC here). I do wonder , if that is true, if it’s not possible to approve them for it or Cambium simply had no motivation to get them approved.
OK. Well, my idea was that if you were seeing this on F300’s while on ePMP3000’s — that the 4x4 nature of them does mean that a -70 is not the same between 2x2 and 4x4.
A -70 in 2x2 will have each chain (theoretically) be a -73. Meaning that -73 + -73 = -70
In 4x4, each chain would be a -76… meaning that -76 + -76 + -76 + -76 = -70
Of course, most likely the signals won’t be taht perfectly balanced, so realistically, in a 4x4 scenario, that -70 might have chains of -76 -74 -84 -81 all adding up to a -70, but half those chains might be just hanging on by their fingernails.
But, this is probably NOT what you’re running into I guess… it was just a thought. It would be great to be able to see the per-chain signals on the AP, to be able to diagnose if these sorts of problems come from individual chains being low signal or not. But without that, we’re left scratching our head.