hi,
i wonder if 802.11ac on 40mhz spectrum will give better result than 802.11an on 40mhz ?
Depends on signal. Higher modulation rates give faster speeds.
i meant same the signal gain. i want to know if 802.11ac is worth upgrading, without using 80 mhz spectrum, because it is gonna be very hard to avoid frequency interference on this wide spectrum.
400 vs 300mbps modulation rate
I have upgraded one 320m link with R11e-5HacD running on 40MHz channel, using x86 dual core Atom boards and Jirous PAR24 Extrem MIMO. With previous R52n-M I was getting slightly over 200Mbps… This link had 100% CCQ but over a time the performance dropped to 150Mbps until I reconnected and then went back to “normal”… I could not figure out why.. so decided to upgrade to mPCIe and AC as it became available.
After upgrade to AC, using same channels, same antennas, same boards I have 210Mbps stable all the time, latency avg. 2ms even under full load. Link speed 360-400Mbps, CCQ around 90-99%, SNR 59dB
Honestly I expected more but am happy with it.. upgrading to R11e-5HnD only would give me same results I guess…
I have read somewhere that AC has higher overhead than N and higher modulation doesnt automatically mean higher throughput.. But I am also testing Ubi AC gear and there is an improvement for sure. 140Mbps on 20MHz channel if you have strong enough signal for 256QAM…
Hope this helps.
Theoretical speed…
Physical layer speed.
Well… I guess that winet asked if its worth upgrading. Physical layer speed or theoretical throughput is useless here. Have you got any real throughput experience?
interesting topic
because on corporate wlan using 80 mhz channels its a fantasy.
without using dfs channels (because many many end devices do not supports them) its only 2 80mhz channel.
with 40mhz channel you have the minimum 4 non overlapping channels (without DFS channels) to take care of co-channel interference and do proper channel reuse.
I think the big deal of AC its the massive use of 5ghz on 40mhz channels because of wide support of end devices and cost reductions.
Previously 5ghz ear had a absurd premium cost
Take the percentage increase of physical layer throughput. You see the same increase in real performance if your signal is strong enough.
MT has to do some work to make that happen with nv2. With plain 802.11 you see this gain.
Take a look here.
http://www.winncom.com/images/stories/GoNet_WP_Things_they_dont_tell_about_80211ac.pdf
The most interesting part is this:
MAC INEFFICIENCY
The degradation in MAC efficiency is a small fact that is usually left unspoken. In the ‘old’ 802.11n
standard operating at 20 MHz the MAC operated with over 85% efficiency. By comparison, the
802.11ac MAC efficiency is 65%. We are ‘wasting’ 20% more time on overheads.
So you say that you can see this gain only with plain 802.11.. I have tested both 802.11 and NV2 and did not see much gain.. 802.11 is useless on P2P as with 50% load latency is too high.. NV2 latency is stable even under full load but AC is much more sensitive to interference which could also be the reason why I do not see throughput increase even with strong signal and physical rate 400Mbps. But as you say, there might me an issue with NV2… so we have to wait. Currently there is no advantage of using AC apart 80Mhz channel width.
You find good numbers doing some math with the link planner from mimosa. Their MAC-Efficiency differs with framing size 2/4/8ms. So you’ve a tradeoff latency vs. speed. 802.11ac is not more sensitive to interference with same modulations. But to get higher modulations you need a cleaner channel of course. With a dirty band .ac (wave1) will not help.
To use .ac you want higher gain antennas with good fb-ratio. (Dont buy SXT-AC wait for QRT-AC). UBNT does some additional filtering to respect the higher need of .ac.
.ac will shine with the appearance of wave2 devices.
Dude, we are talking about Mikrotik right? I am sure Mimosa and UBNT have some proprietary MAC Layer and definitely not only standard 802.11, therefore their MAC efficiency is much higher (for example Ubi AC = 140Mbps real throughput on 20MHz channel, even the latest gear can deliver more throughput on N-like modulation if you have not strong enough signal to get 256QAM). UBNT P2MP also does not currently support mixed A,N,AC but AC only. Mikrotik does and that might be the issue (I’ve read somewhere).
I am using high gain shielded antennas and can get full 400Mbps physical rate on short distance, signal -51, quite clear channel (its in a city but not so busy) but still cannot get more throughput compared to N.. So is this NV2 issue and will it be better if Mikrotik guys improve it? I would expect at least similar throughput to UBNT.. which ideally on 40Mhz wide channel around 250-280Mbps real TCP throughput.
Sorry I am not so good in theory but rather doing real tests 
In my case also. I tryed ac devices (Mikrotik) and compared them to N, using TCP tests, but N was better than AC. So we dont plan to use Mikrotik AC devices in near future.
What I heard from folks in ISP business Ubnt has much better AC devices (at least 1/2 faster than Mikrotik), so maybe we try UBNT.
But as it was said, this all is maybe because propietary chips of Ubiquity and Mikrotik is using Atheros, which are very sensitive to noise and in real enviroment useless…
Hope Mikrotik will start to use some filter on RF parts or some chips…like Ubiquity does…
We are talkin 802.11ac. Mimosa is based on a standard .ac chipset. UBNT uses the same Atheros Chipsets MT does. So results should be comparable.
Yep, they do. But for example UBNT uses also a TDMA ASIC, plus the filtering technology… Dont know about mimosa but its definitely not plain 802.11. Its based on atheros but possibly modified (lets say improved) MAC.
We are on mikrotik forum and if anyone ask if its worth upgrading to AC, I suggest he asks about Mikrotik. And for now Mikrotik ac is not worth upgrading from N. But I believe they can make it better so I am ready with AC chips waiting for SW improvement 
Mimosa uses a Quantenna chipset. The UBNT AC Lite is normal .ac. These AF5X are custom designed chips (no .ac). Yes MT has to do some work for .ac. Esp. nv2 needs adaption. The problem might be the cpus on the MT-.ac boards are not much faster than on .n boards. SW-Based TDMA needs some CPU-Power.
Well, I was looking at Mimosa some time ago, it looks great but many people doesnt like the “full automatic” features So I might give them a chance when they are available in my country.. BUT I dont like the latency though.. UBNT AC lite has the airmax ASIC, just not the airprism… so it doesnt have the filtering but still can deliver up to 140Mbps on 20MHz channel.. AF5X is Invictus custom based chip, thats clear.. Personally I cant wait for those units. I am also guessing the CPU on MT boards is weak, but.. there is no problem with CPU usage with max load. It must be some internal limitation… I am a little surprised if this is the case why MT did not design something similar to UBNT and add another chip just for TDMA… I am sure we all would be willing to pay more for those boards if they would increase performance.
You dont have to use them and they do not work now. As with other vendors the announced features are repaired later 
.
With a PHY-Rate of 2x86.7 up to 140Mbps is doable under ideal conditions. Dont expect to see the numbers of the fan-boys on UBNT Forum. All shines there even when they have to restart between tests.
MT does not design custom wireless chips. Do a lab test with 2 sxt acs. You will see better performance with plain 802.11ac. This chips are built to handle the standard protocol at full speed. If you add TDMA Protocol you’ve to tweak. Packing and scheduling packets (retries) at this rates is a hard job. And doing e.g. 2ms frametimes cpu has to hurry to get work done until sending the next frame. If too late it has to wait for the next frame. Doing standard 802.11 with this chips you throw the data in and the chip handles the protocol (aggregation, retries, …) while the wireless chip does the wireless job the cpu can do other jobs. This is why UBNT does these custom chips.