802.11ax

Redmor · Thu Jan 31, 2019 9:45 pm

Is MikroTik planning to make some new antennas with 802.11ax protocol?

mistry7 · Thu Jan 31, 2019 10:45 pm

Yes sure, you can become the next unpaid Beta Tester

ARM Hardware is now avaible for 13 Month, and does not work as aspected....

honzam · Thu Jan 31, 2019 11:49 pm

ARM Hardware is now avaible for 13 Month, and does not work as aspected....

I think longer...

Fri Feb 01, 2019 3:34 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?

blingblouw · Fri Feb 01, 2019 3:44 pm

Guess a lot changes in 7 years

"you can go with either of them, nstreme and nv2 is going to be faster than vanilla 802.11, and if you are using only RouterOS wifi clients, that is good option to choose.

the fastest of all three would be NV2 in normal conditions."

viewtopic.php?f=7&t=66740&p=339727#p339727

Fri Feb 01, 2019 3:45 pm

yes, this is what the word "legacy" means. it means we added nv2 to improve 802.11n. It was not designed with ARM or 802.11ac in mind. These new systems have no benefits, as they are already superior.

pcunite · Fri Feb 01, 2019 3:58 pm

Guess a lot changes in 7 years

It is fair for MikroTik to fix an issue with a standard that later gets updated and moves on past the fix.

blingblouw · Fri Feb 01, 2019 4:33 pm

Guess a lot changes in 7 years

It is fair for MikroTik to fix an issue with a standard that later gets updated and moves on past the fix.

Totally agree. I just think that documentation needs to be updated to reflect that

2jarek · Fri Feb 01, 2019 5:35 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?

Test 2x ARM Bridge. Not stable & slower than MIPSBE.

mistry7 · Fri Feb 01, 2019 6:30 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?
Test 2x ARM Bridge. Not stable & slower than MIPSBE.

For us the same, ping lags on arm Based CPE, VoIP not possible

honzam · Fri Feb 01, 2019 8:41 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?
Test 2x ARM Bridge. Not stable & slower than MIPSBE.

The same problem. NV2 is slower and have other issues. Nstreme is also slow. Best is 802.11, but this is not suitable for p2mp

high ping, jitter and lossability

mistry7 · Fri Feb 01, 2019 10:25 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?
Test 2x ARM Bridge. Not stable & slower than MIPSBE.
The same problem. NV2 is slower and have other issues. Nstreme is also slow. Best is 802.11, but this is not suitable for p2mp high ping, jitter and lossability

But it looks like it is the "new" Mikrotik Quality....

bastys · Fri Feb 01, 2019 10:27 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?
Test 2x ARM Bridge. Not stable & slower than MIPSBE.
The same problem. NV2 is slower and have other issues. Nstreme is also slow. Best is 802.11, but this is not suitable for p2mp high ping, jitter and lossability

Yes, I have same problem with 802.11. Always same (ptp and ptmp)

Sat Feb 02, 2019 2:31 pm

Mikrotik should not allow nv2 to be selected on 802.11ac hardware.

It is clearly broken, why even give the customer the option of being disappointed

pcunite · Sat Feb 02, 2019 4:58 pm

MikroTik should not allow nv2 to be selected on 802.11ac hardware. It is clearly broken, why even give the customer the option of being disappointed?

They need a user experience manager who thinks about these things. Hard to find super techie and yet a client advocate in the same person.

r00t · Sun Feb 03, 2019 1:53 am

If nv2 compatibility vs. older hw is the issue, why not just call it nv3, change the timings and whatever is needed for it to work on ARM and clearly state it only works on the new hardware and it's incompatible? If the compatibility is in the way, fine... but at least have a new protocol that's stable between ARM devices.
And if designing new protocol is too hard for P2MP, at least have one for P2P links that works...

Sun Feb 03, 2019 5:02 pm

There is a (big) arm/nv2 thread already.
This thread is about 802.11ax. Would like to hear about it, too.

R1CH · Mon Feb 04, 2019 7:16 pm

You can buy 802.11ax routers on the shelf in retail stores already, yet zero communication from Mikrotik about their timeline... this is rather worrying.

Tue Feb 05, 2019 11:51 am

You can buy 802.11ax routers on the shelf in retail stores already, yet zero communication from Mikrotik about their timeline... this is rather worrying.

It's only one month until the EU MUM. Maybe they will announce some 802.11ax gear.

mkx · Tue Feb 05, 2019 12:02 pm

It's only one month until the EU MUM. Maybe they will announce some 802.11ax gear.

Holding your breath will only help you to suffocate

pe1chl · Tue Feb 05, 2019 12:11 pm

If nv2 compatibility vs. older hw is the issue, why not just call it nv3, change the timings and whatever is needed for it to work on ARM and clearly state it only works on the new hardware and it's incompatible? If the compatibility is in the way, fine... but at least have a new protocol that's stable between ARM devices.
And if designing new protocol is too hard for P2MP, at least have one for P2P links that works...

What we need is NOT yet another incompatible P2MP link protocol, but implementation of some standard or at least interworking with what other manufacturers (especially UBNT) do.

vecernik87 · Tue Feb 05, 2019 12:16 pm

I heard that they are releasing new version of RoS with airmax compatibility on 1 Apr 2019

Tue Feb 05, 2019 12:21 pm

I heard that they are releasing new version of RoS with airmax compatibility on 1 Apr 2019

+1

pe1chl · Tue Feb 05, 2019 4:10 pm

I heard that they are releasing new version of RoS with airmax compatibility on 1 Apr 2019

That must be RouterOS v7 !!!

Tue Feb 05, 2019 4:43 pm

I heard that they are releasing new version of RoS with airmax compatibility on 1 Apr 2019
That must be RouterOS v7 !!!

Codename: Unicorn

For sure .ax will be earlier. Think back: MT was the first with .ac.

pe1chl · Tue Feb 05, 2019 4:48 pm

To do anything with P2MP protocols we need compatibility between manufacturers, at least between UBNT and MikroTik.
Now we can do only standard 802.11 because there always is at least a MikroTik client on UBNT AP and vice-versa.

r00t · Wed Feb 06, 2019 3:01 pm

LOL... Good luck with that, I'm sure manufacturers are really thrilled to publish all their secrets so they can make less money... this is simply not going to happen.
Only possibility is if TDMA is part of the future official 802.11something specification.
And it's a good question why this wasn't done long time ago... it's not like TDMA protocols like nstreme, NV2, AirMax are new.

Wed Feb 06, 2019 3:20 pm

LOL... Good luck with that, I'm sure manufacturers are really thrilled to publish all their secrets so they can make less money... this is simply not going to happen.
Only possibility is if TDMA is part of the future official 802.11something specification.
And it's a good question why this wasn't done long time ago... it's not like TDMA protocols like nstreme, NV2, AirMax are new.

802.11ax claims to do the job:
https://www.qualcomm.com/solutions/netw ... es/80211ax

ofdma gives frequency slices to cpes
Uplink resource scheduler handles media access (So what is different to TDMA).

So the TDMA/compatibility discussion might be gone ...

pe1chl · Wed Feb 06, 2019 3:21 pm

We had the same issue in amateur packet radio and it was solved in the nineties. The solution was called DAMA and I think it is much like NV2 and AirMax, maybe less a few of the special optimizations.
It is unbelievable that 25 years later there still is no 802.11xx standard that covers this and can be implemented by every manufacturer.
Then at least you can have rudimentary collision avoidance on a P2MP link, even when the manufacturer-specific protocol works a bit better.

Does 802.11AX still not include such a system? I would think that an "all new High-Efficiency Wireless protocol" would... the "collisions due to hidden terminal" problem should be present even in indoor networks at increasingly higher frequencies...

Wed Feb 06, 2019 3:25 pm

We had the same issue in amateur packet radio and it was solved in the nineties. The solution was called DAMA and I think it is much like NV2 and AirMax, maybe less a few of the special optimizations.
It is unbelievable that 25 years later there still is no 802.11xx standard that covers this and can be implemented by every manufacturer.
Then at least you can have rudimentary collision avoidance on a P2MP link, even when the manufacturer-specific protocol works a bit better.

Does 802.11AX still not include such a system? I would think that an "all new High-Efficiency Wireless protocol" would... the "collisions due to hidden terminal" problem should be present even in indoor networks at increasingly higher frequencies...

It does. Look my post one minute before your's.

mkx · Wed Feb 06, 2019 3:28 pm

the "collisions due to hidden terminal" problem should be present even in indoor networks at increasingly higher frequencies...

Those networks might still be wireless, but will become more and more NFC to a point where it won't be possible to have any obstacle between radio partners (except for their housings) and thus there won't be any "hidden terminal" due to lack of signal coverage.

pe1chl · Wed Feb 06, 2019 3:33 pm

It does. Look my post one minute before your's.

DAMA does not require any special arrangements at the radio layer. It just polls the clients (at the link layer protocol) in a ring to ask them if they want to send data.
It is my suspicion that both NV2 and AirMax do the same thing. However as they both invented their own protocol they aren't mutually compatible.

Of course in a new system one can solve it at the radio layer, but the polling solution can be used with existing radio hardware.

pe1chl · Wed Feb 06, 2019 3:36 pm

Those networks might still be wireless, but will become more and more NFC to a point where it won't be possible to have any obstacle between radio partners (except for their housings) and thus there won't be any "hidden terminal" due to lack of signal coverage.

I thought the movement is towards beamforming accesspoints (with the associated advantages of MIMO) which means that all clients clearly hear the accesspoint, but do not hear eachother at all.
That is the same as with a P2MP outdoor link setup.

mkx · Wed Feb 06, 2019 9:40 pm

Those networks might still be wireless, but will become more and more NFC to a point where it won't be possible to have any obstacle between radio partners (except for their housings) and thus there won't be any "hidden terminal" due to lack of signal coverage.
I thought the movement is towards beamforming accesspoints (with the associated advantages of MIMO) which means that all clients clearly hear the accesspoint, but do not hear eachother at all.
That is the same as with a P2MP outdoor link setup.

Yeah, beamforming is buzzword of the moment. However, there's still EIRP limitation and with all that free air pathloss increasing with frequency and specially with inability to penetrate or refract the obstacles (a moving human body can become a huge problem) ... I'm just not holding my breath.

mistry7 · Thu Feb 07, 2019 8:06 am

We had the same issue in amateur packet radio and it was solved in the nineties. The solution was called DAMA and I think it is much like NV2 and AirMax, maybe less a few of the special optimizations.
It is unbelievable that 25 years later there still is no 802.11xx standard that covers this and can be implemented by every manufacturer.
Then at least you can have rudimentary collision avoidance on a P2MP link, even when the manufacturer-specific protocol works a bit better.

Does 802.11AX still not include such a system? I would think that an "all new High-Efficiency Wireless protocol" would... the "collisions due to hidden terminal" problem should be present even in indoor networks at increasingly higher frequencies...
It does. Look my post one minute before your's.

And you really believe that Mikrotik will support that feature from the beginning, there are so many things missing @ac Hardware, so don’t call for new Chipsets, they are not able to support them

Wave 1 Missings:
- Output Power Managment
- Spectral Scan
- Channel sizes

Wave 2 Missings:
- no wave 2 features available

What’s about 802.11k, 802.11r and 802.11v, we need that in the EU for every Wireless Network in
Government Places / Meeting Rooms etc
Ask Mikrotik for those features and you hear silence
At this point there is no future for Mikrotik Wireless, not Indoor and not Outdoor, only at
low cost markets

r00t · Thu Feb 07, 2019 2:18 pm

This is quite troubling situation, as for WiFi4EU you need these features and no Mikrotik device supports them. So even if you love Mikrotik, you can't use it. And these are often big projects, paid partly by EU money. If Mikrotik had suitable device, they could sell good amount of them.
Instead all that money is spent on competitor products and as a result they get more R&D money to release even better products, meanwhile Mikrotik is stuck in the past with limited R&D budget unable to catch up. We can only hope it's actually not this bad...

Redmor · Thu Feb 07, 2019 9:06 pm

Come on MikroTik support, at least tell us if you're planning something or not.

mistry7 · Thu Feb 07, 2019 9:18 pm

Come on MikroTik support, at least tell us if you're planning something or not.

Normis wrote on other Topic that they develop new driver with wave 2 support, but we don’t no when it is available

Redmor · Thu Feb 07, 2019 9:29 pm

I'd prefer to see ax based protocol than Wave 2 ac

mistry7 · Thu Feb 07, 2019 10:16 pm

I'd prefer to see ax based protocol than Wave 2 ac

You want to get next Beta Tester, buy arm that’s Beta enough

Redmor · Fri Feb 08, 2019 8:16 am

I only use ARMs as clients and I don't see any problem.
I have throughput problem even with mibsbe in some cases, especially for client upload, or with NetMetal + mANT I only get 50Mbps throughput when with other vendors I get 200Mbps, but I think it's a mANT problem.

mistry7 · Fri Feb 08, 2019 8:45 am

I only use ARMs as clients and I don't see any problem.
I have throughput problem even with mibsbe in some cases, especially for client upload, or with NetMetal + mANT I only get 50Mbps throughput when with other vendors I get 200Mbps, but I think it's a mANT problem.

Just setup an Omnitik ac, connect some clients, and do a ping to the MISPBE Clients and the same time to the ARM Clients, Mode 802.11
You will See high Pings on ARM without any reason, impossible to deliver Voice Service over this Hardware.

Fri Feb 08, 2019 2:54 pm

And whooop .... back to topic

@Normis Are you able to talk about 802.11ax ? At least what MUM to look at ....

There are chipsets of your suppliers out there: https://en.wikipedia.org/wiki/IEEE_802.11ax#Silicon. So I guess you have something in your lab and as you did with 60GHz you might want to do some interesting move there ... hopefully

r00t · Fri Feb 08, 2019 5:18 pm

Would be great if Mikrotik released at least minimalistic roadmap, something like: 802.11ax work in progress, expected ~2019Q3
Same for Wave2 and other features. No need to be more specific, just confirmation of "yes, we are really working on it".
This might be enough to delay investment and buying of competitors products. Currently you wait, and wait and wait and nothing is happening, no news.
So you end up buying other company products, only to see Mikrotik release it month later. In the end it only hurts the Mikrotik sales...

DSK · Fri Feb 08, 2019 11:58 pm

+1 for this. Normis and colleagues at Mikrotik, we need a roadmap for wireless LAN products development for the near future. Mikrotik is my favorite and default goto device, but one can barely plan for the future as you guys hardly talk of what's in store for the future! It's by guesswork and keeping on hoping for the better. Put up a roadmap ASAP!!

0ldman · Fri Mar 08, 2019 6:54 pm

You can buy 802.11ax routers on the shelf in retail stores already, yet zero communication from Mikrotik about their timeline... this is rather worrying.

802.11ax isn't finalized yet.

You must be young.

The routers on the market are "draft" ax, as in public beta. Compatibility with the final product is not guaranteed.

https://searchnetworking.techtarget.com ... -late-2019

0ldman · Fri Mar 08, 2019 6:59 pm

I'd prefer to see ax based protocol than Wave 2 ac

ax is going to have better contention protocols and MUMIMO than ac, but not by much. Wave 2 with MUMIMO has that capability and the built in contention control may interfere with nstreme and nv2.

I imagine that ax is going to be a bit of a bear for MT to adapt to. Once they get it going it really does sound like a match made in heaven, however, it has so many WISP related features built in I believe it is going to cause them problems. They'll probably disable most of those features in the initial version and slowly roll out updates with hardware support for the built in contention control.

Fri Mar 08, 2019 7:23 pm

You can buy 802.11ax routers on the shelf in retail stores already, yet zero communication from Mikrotik about their timeline... this is rather worrying.
802.11ax isn't finalized yet.

You must be young.

The routers on the market are "draft" ax, as in public beta. Compatibility with the final product is not guaranteed.

https://searchnetworking.techtarget.com ... -late-2019

Most likely chipsets will do the job already:
https://searchnetworking.techtarget.com ... -late-2019

I am not realy young

But we need some upgrades soon.

Tue Apr 02, 2019 9:51 pm

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?

Normis - re: ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.

If Nv2 is a legacy protocol, then how about Mikrotik create a next generation TDMA protocol. Call it Nv3 or something.

North Idaho Tom Jones

pe1chl · Tue Apr 02, 2019 9:58 pm

I hope there will be no next manufacturer-specific TMDA protocol but rather an established standard that all manufacturers follow and that allows inter-operation.

r00t · Wed Apr 03, 2019 4:53 am

For 802.11ax, there should be no need for any custom TDMA protocol, because standard itself already have what you can call NV2 on steroids: Not only it's TDMA, it can also assign different parts of spectrum to a different users for each timeslot... over multiple MIMO beams. And all this is handled by hardware chips, so with great precision and no requirements of main CPU to do the timing (and cause issues like with NV2 on ARM).

MasterCATZ · Sat Apr 20, 2019 2:21 pm

please 802.11ax IOT is killing the network , even solar panels now clutter up the wifi

zlobster · Sat Apr 20, 2019 10:16 pm

I've been quietly following MT for many moons now, but I'm honestly clueless what is their business strategy. If this is thier tactics, they totally got it right, though!

Maybe MT focus on the last-mile and some backhaul? Maybe ISP edge? Or CPE/SOHO? Or maybe amateur projects? Or maybe stability before features? IDK...

All I now is that it's 2019!!! 5G is being deployed even in Eastern Europe, .ax mainstream devices are almost flooding the streets these days, IoT is already wrecking havoc in some places, more and more spectrum is being made available worldwide. And I'm just sitting here, waiting for MT to release some modems above Cat.4 or at least improve compatibility with other vendors' ones.

taduikis · Mon Apr 22, 2019 2:46 pm

Guys, calm down. Chill a little. Forget MT Wireless. It’ll be like a breakup with a girl: hard, painful, expensive at first, much work, but you’ll get over it.
I did. I miss RouterOS superior management and features, since I hate web GUIs and limited console configuration options, I sweared much, missed mac-telnet, torch, l2 and l3 firewall, super flexible bridge and endless other configurations.. but at the end of the day what matters the most? Right! User experience!. As long as it works ok, you don’t really care any of that above mentioned stuff and have your chilled margaritas

Let MT sell wireless eq in developing countries where 2Mbps are still called broadband. We need much more advanced stuff and get used that MT fail to deliver it, so let’s just jump the boat and stop bragging..

P.S. .ax alone won’t save them, IMHO.

zlobster · Mon Apr 22, 2019 7:05 pm

Guys, calm down. Chill a little. Forget MT Wireless. It’ll be like a breakup with a girl: hard, painful, expensive at first, much work, but you’ll get over it.
I did. I miss RouterOS superior management and features, since I hate web GUIs and limited console configuration options, I sweared much, missed mac-telnet, torch, l2 and l3 firewall, super flexible bridge and endless other configurations.. but at the end of the day what matters the most? Right! User experience!. As long as it works ok, you don’t really care any of that above mentioned stuff and have your chilled margaritas Let MT sell wireless eq in developing countries where 2Mbps are still called broadband. We need much more advanced stuff and get used that MT fail to deliver it, so let’s just jump the boat and stop bragging.. P.S. .ax alone won’t save them, IMHO.

This really made me ROFL.

It's sad but true, though. I now only use MT in a few very niche projects. Shame to watch such potential going to waste...

andyhenckel · Thu Jun 27, 2019 7:14 pm

Dear Forum, Cheers to All. May your packets never queue. I've personally not had any problems with ARM, or NV2 as long as you remember that NV2 doesn't work with poor link quality causing re-transmissions. In terms of the performance, it seems to work fine and the faster the processors the better. I really like the dual or quad core processors on the wireless wire and hoped every new unit had that much horsepower. I'm using all 802.11AC radios, and provide my clients with speeds between 10 and 40 mbps x 5, expecting to get no more than 15 clients per device PTMP. I can get 180mbps over the air all over town and 300mbps tcp on the bench one direction with AC. FDX is 100 TCP on bench, or 70mbps outdoors real world. I'm not sure what the rest of you are doing. There were some optimizations I learned early on (through the forum) to make things work right.. and it would be nice if mikrotik enabled some of those things by default as it seems to improve performance. Namely, the multicast helper(for NV2), interface queues (multi-queue ethernet default), ethernet flow control (on). I've been happy with the product and with NV2 compared with no NV2. the key with NV2 is that you can't hookup clients with crappy signal, anything better than -65 is ok, but I prefer all clients in the -59 or better range. I configure all nv2 aps with 70 down and 30 up, typically a 2 or 3 ms, and turn hardware retries down to 2 rather than 7. All minor improvements that boosted overall performance, in addition to a current and stable version of firmware and routerboard software. We have used the Mtik Dish and Netmetal AC to spotlight areas from the tower, for instance a neighborhood with 200 homes 20 KM away from the tower. Then I've started to use the DynaDish and i'm pretty happy with the beam width as it trims out the local interference better than the QRT. Re-transmissions kill NV2. I am very hopeful for 60 G, and have been using two links to backhaul from tower to tower downtown for over a year at 350meters with no issue. (I strapped one to an old Canopy dish, and omg the signal is perfect if you have the patience to align it.) I am hoping that the new wireless wire dish will get an integrated AX radio so we can have a primary and secondary radio set in each link to allow for redundancy to the tower. (hint hint, nudge, nudge, wink, wink) I eagerly await the integration of new standards for AX, to keep us small WISPS ahead of the curve. My clients and I are happy with Mikrotik products and the versatility of such a powerful OS. If MTIK can help with the Wireless wire dish + AX radio, us small wisps can actually provide reliable service for 50 clients from one site and 5 GB of over air TCP data, allowing us to compete in small neighborhoods with the existing Cable operators.

mfr476 · Fri Jun 28, 2019 9:56 pm

Please gif us your ap and cpe configuration It is incredible¡

taduikis · Sat Jun 29, 2019 9:32 am

I'd like to address your comment in step-by-step:
- yes, you're right, the RouterOS is increadible OS with it's instruments, management, capabilities, etc.
- NV2 from ~6.42 has been updated and poor signal/quality clients no longer drag down the AP performance that much, also higher number of clients on AP doesn't reduce the overal capacity of AP anymore (though they still could have done better). Together with TDD ratio, I call this NV2 update one of the best since the NV2 introduction. It actually works.
- Hardware retransmits doesn't have an effect on NV2. It was mainly used back in the day for Nstreme.
- CPU power on AC CPEs is not everything. Higher performance CPUs mostly benefit on APs, especially when there's such magic like GPS, beamforming, TDMA scheduling in play. Newer generation MT CPEs with >=600 MHz CPUs are usually enough.
- Sure, if your network is all on AC, you get some benefit from that. But I bet you're on pretty clean RF over there, and furthermore you use some client-oriented devices for APs to filter out the interference. Because on MT, most advantages of AC come only in the form of higher modulations, which again are useless in noisy areas. You can use narrow beamwidth horn antennas to achieve the very same result and to have some more professional solution. We all think of various solutions to survive, so we've all been there, done what you said. And managed MikroTik wireless to live up to this day.

Now let's get back to reality.. In this 4G LTE and comming 5G era, we have no chance to compete with current MikroTik products. We simply cannot. Sure, 4G is not available in some areas, but soon it will be when equipment rotation will start happening. To buy us few more years we need seriously reworked AP devices from MikroTik. With backward compatibility of course. Because competition is already far ahead of MT, and they showed us that it can be done - MT's AC cannot outperform their 11N. And they also showed us that MT client devices are quite good and capable. So not everything is lost for MT yet. I remind MT few key features they somehow really need to work on: nearby channel filtering, proper GPS-sync for freq reuse, asymmetric fixed downlink/uplink datarate support, little more improvements in NV2 to call it NV3.

pe1chl · Sat Jun 29, 2019 11:49 am

Now let's get back to reality.. In this 4G LTE and comming 5G era, we have no chance to compete with current MikroTik products.

With "we", you likely mean the small WISP that offers internet access to a couple of local paying customers using a WiFi network.
That is of course a niche application that works only in places where it can compete with better solutions like you mention.
In a highly populated area with large demands on internet bandwidth, such solutions never have been viable, no matter what make of equipment you use.

But it is not the only application for MikroTik equipment.

andyhenckel · Tue Jul 09, 2019 11:42 pm

Code: Select all

  

I'm not so sure 5G cellular is going to be the "bees knees".   Cell companies still have to roll out their hardware and i've got no faith in 5G given the current results of tests done in Dallas, LA, and Chicago. It won't work through windows or much very well and i'm anticipating they will be nothing more than little wifi hotspots and it's mostly hype to generate more sales.  I've been installing for them for 14 years since 2.5 G cellular.   They use overpriced hardware that doesn't seem to perform well enough to justify the additional expense.  The only thing the cell companies have is the ability to output at 300 times more power than ISM band WISPS.  They have a huge budget, make record profits, and i'm sure they are lobbying to try and take over the 5Ghz to 6gHz band to further create more of a monopolistic situation to boost bottom lines and reduce competition from highly motivated small WISPS. 



There is a bunch of interference in this valley.  In Reality there are only 3 major tower sites where everyone and their brother co-locates to serve the 70-100 thousand inhabitants of this 6x10 mile valley. Spectrum and other local wired ISPs have most of the market share.    However there are many people on the fringe who cannot use the 4g Network to operate, have low speeds from other ISPs, and the major players have no intention of providing services to the outlying areas, ever - without massive grants funded by further inflation and debt.   I personally own a network that I paid cash for every component, after working for Cell companies and building their infrastructure and being beaten down by soulless huge spreadsheet management firms posing as capable construction management firms that pinch most of the profit and shed any responsibility. 



In any case 802.11AX will likely perform as well or close to the expected performance of a nokia or alcatel or huawai base station.   For the same price we can deploy 20 times the sectors and throughput with Mtik..   We already have a 5G style link performance with 802.11AD for a fraction of the price.    I really don't see much better performance from 4G than 802.11AC in terms of realized throughput. You pay the additional premium for better quality, higher powered components, with redundant power supplies,  HVAC, and proprietary algorithms for TDMA that makes the medium more friendly and adds tons of latency.  Since most providers enabled Unlimited access, at times my service is worse than dial up at times.   Fact is that there are not enough towers, and they are not motivated financially enough to add outside of the core community centers.   



If you've got a small WISP, build out a site, hold the root lease, then sublet to the Cell companies for 2K per month and you can win both ways.  Or start putting fiber in the ground, and use your tower assets to fire licensed links to small fiber loops in select areas where you already have a handful of wireless clients.... that's what I'm doing.   And it's been very successful.  I do feel us Small WISPs can fill a niche, and many folks would rather support a local company anyway than promote the corporate grab for concentrated power.

WirelessRudy · Tue Aug 06, 2019 11:23 pm

I don't understand al that whining abt MT not making it into the future...
I am a small 750 clients WISP in an area where at least 8 other providers are around.
60% of my network is now full 'ac' (with a load of 'arm' devices) and I run on most of my 40+ sectors plain 802.11 which still everytime again shows better results then NV2 (or nstreeme)
Yes, I tend to work with signal levels close to -50, anything higher then -60 will be up for an antenna upgrade or it must be on a small (<10 clients) AP.

Our plans are offering 50Mbps down, 5 up to our clients and so far we have a very satisfied customer base.
In one area to compete with fiber we have a Netmetal with 40+ SXT's and SXT'sq, all working in 802.11ac and we give them the 100/20 Mb plan...
The frequency even overlaps some other but in careful picking the 'pilot' channel it all works to satisfaction.
Due the pricing of MT hardware we can always compete on price with any of the 'bigger' providers and although we loose some at times (and get new ones back) we managed to have a stable and satisfied client base.
We use several 60Ghz links for the short backhal to satisfaction (ditched some Metrolinks over it. Unreliable and too costly). These 60Ghz works so good I tend to forget them in the update cycles we maintain since they simply always work.

Yes, MT has a learning curve and indeed, a lot of fine tuning can be done, and needs to be done. We also tried Mimosa but for the money we spend on that I regret I didn't wait a year and would have no all the same done with Mikrotik for much less money..
All my competitors use ubiquity but I see them struggling to get the same done as us and we see several of them trying other brands as well. So they must have their issues too...

We have 4G LTE since 2 years in our region, but to be honest... it sucks.. Where speeds very good the first weeks (way over 100Mbps, even in a moving car) that fell considerably the moment more people bought themselves 4G phone and started to watch their video over it. (And these same clients still want 'fixed' internet to their home).
For the rural area (but high housing levels) I don't see 5G going to deliver the promises neither. Their simply won't come enough towers to the region.

MT as usual takes time to bring new technology and needs even more to mature their ROS when the hardware is finally there, but for the price it makes still a economic model to wait maybe a year to do the same as any high level product but for a fraction of the price.
For instance eCambium with their new ePMP4000 product line is claiming higher CPE levels per AP but the price is also several times higher. And in the transition process you need to run two networks side by side. In the anology of taduikis this means you still need to keep the present girlfriend with all its merits while investing in the new one. And yes, you get screwed twice! They 'interfere with each other and we all know what that means!

For small WISPS with small budgets the niche market can still be profitable but you have to be smart and stick to the wife once chosen. If you are loyal you have to hope she (MT) is too and delivers us new hardware to sustain our marriage into the future!

Tue Aug 06, 2019 11:43 pm

I don't understand al that whining abt MT not making it into the future...
I am a small 750 clients WISP in an area where at least 8 other providers are around.
60% of my network is now full 'ac' (with a load of 'arm' devices) and I run on most of my 40+ sectors plain 802.11 which still everytime again shows better results then NV2 (or nstreeme)
Yes, I tend to work with signal levels close to -50, anything higher then -60 will be up for an antenna upgrade or it must be on a small (<10 clients) AP.

Our plans are offering 50Mbps down, 5 up to our clients and so far we have a very satisfied customer base.
In one area to compete with fiber we have a Netmetal with 40+ SXT's and SXT'sq, all working in 802.11ac and we give them the 100/20 Mb plan...
The frequency even overlaps some other but in careful picking the 'pilot' channel it all works to satisfaction.
Due the pricing of MT hardware we can always compete on price with any of the 'bigger' providers and although we loose some at times (and get new ones back) we managed to have a stable and satisfied client base.
We use several 60Ghz links for the short backhal to satisfaction (ditched some Metrolinks over it. Unreliable and too costly). These 60Ghz works so good I tend to forget them in the update cycles we maintain since they simply always work.

Yes, MT has a learning curve and indeed, a lot of fine tuning can be done, and needs to be done. We also tried Mimosa but for the money we spend on that I regret I didn't wait a year and would have no all the same done with Mikrotik for much less money..
All my competitors use ubiquity but I see them struggling to get the same done as us and we see several of them trying other brands as well. So they must have their issues too...

We have 4G LTE since 2 years in our region, but to be honest... it sucks.. Where speeds very good the first weeks (way over 100Mbps, even in a moving car) that fell considerably the moment more people bought themselves 4G phone and started to watch their video over it. (And these same clients still want 'fixed' internet to their home).
For the rural area (but high housing levels) I don't see 5G going to deliver the promises neither. Their simply won't come enough towers to the region.

MT as usual takes time to bring new technology and needs even more to mature their ROS when the hardware is finally there, but for the price it makes still a economic model to wait maybe a year to do the same as any high level product but for a fraction of the price.
For instance eCambium with their new ePMP4000 product line is claiming higher CPE levels per AP but the price is also several times higher. And in the transition process you need to run two networks side by side. In the anology of taduikis this means you still need to keep the present girlfriend with all its merits while investing in the new one. And yes, you get screwed twice! They 'interfere with each other and we all know what that means!

For small WISPS with small budgets the niche market can still be profitable but you have to be smart and stick to the wife once chosen. If you are loyal you have to hope she (MT) is too and delivers us new hardware to sustain our marriage into the future!

Don t know what your available Spektrum is. We need/use gps sync to get the performance we need.

Wed Aug 07, 2019 12:00 am

!ste:

Is it necessary to quote FULL previous post?

Wed Aug 07, 2019 12:15 am

!ste:

Is it necessary to quote FULL previous post?

No. As forum SW handles this quite good (scrollbars) my lazyness does not hurt too much

Wed Aug 07, 2019 12:22 am

How to educate other users if you do not set a good example? Laziness is not a good excuse.

WirelessRudy · Wed Aug 07, 2019 2:12 am

I don't understand al that whining abt MT not making it into the future.............................................. our marriage into the future!
Don t know what your available Spektrum is. We need/use gps sync to get the performance we need.

This is Spain... we use like all the others, that what is available.
Let me count, in total some 40+ AP's and at least some 25 backhauls. All in 5Ghz, most 40Mhs, many 80Mhz. All in a range of 15km...
Count my blessings. Every sector I have at leasts overlaps some other radio and vice versa in some corner of our valley. It's a big puzzle to find working freq's.
I know some operators even work 'out of band' in 100% illegal frequencies, some in 'grey' area (legal but too much power....) etc. just to keep the business going. Nobody denounces since everybody will see it then boomerang back.....

Yes, MT's sync sucks. I tried it several times but it simply is worse then without. But then again, we even have 4 Mimosa AP's that we tried to do sync and we could only manage to get 2 pairs out of the 4 AP's to work seamless. The others interfere with each other.
Some new provider entered this region a year ago with two towers with each 4 sectors, synch'd but each full 80Mhz channel, ubnt radios. It caused problems for some of my sectors so shifted some frequencies and now even have one of my sectors at distance but serving the same region. It sits in the middle of their 80Mhz and I am serving my clients with 40Mhz...The 802.11ac protocol works fine, don't know about theirs.... (All my signals to clients have to be -60 or better. And ping times are not very stable but below 40ms which is ok for a P2MP setup. I do always make sure my "C" channel ('pilot channel'?) is not the same as theirs though. The 'ac' protocol should work 'around' it.
We don't do VOIP, it's cheaper to give clients a terminal with a simm of one of the cell operators. We have a deal with them. They bill us wholesale and we bill the clients. Problems are theirs and we still make a couple of bucks per month on these clients...! Clients don't care if this is VOIP over our wire or Cell (with VOIP over cell). As long as they have a phone line with a geographic number assigned they are happy.
Bulk of our clients use video and IPTV and that works absolutely fine for them. Our 90% issues come from home Wifi.

When I started this WISP 18 years ago the spectrum was sort of desert. Now it's more like the battle of Austerlitz with too many parties and even the big army's (cell operators with their G4) make a mess with low coverage and illegal towers.... It's a war out here and if there would be any arbitrage 80% had to shut down and probably the richest (or the best 'friends in high places' = money) win. That is not me.....

As long as we all earn our daily bread and people get their internet we will survive... And MT makes it possible!

mistry7 · Wed Aug 07, 2019 9:13 am

[/quote]

We don't do VOIP, it's cheaper to give clients a terminal with a simm of one of the cell operators. We have a deal with them. They bill us wholesale and we bill the clients. Problems are theirs and we still make a couple of bucks per month on these clients...! Clients don't care if this is VOIP over our wire or Cell (with VOIP over cell). As long as they have a phone line with a geographic number assigned they are happy.
[/quote]

We Are in Germany and we have 10 Village Networks without any CellPhone Service.
The Ping over our Networks is in 1-2ms
We have No AP with more Then 15 Clients for
Staible VoIP Service.

Our 90% issues come from home Wifi.

Yes that is the same here

When I started this WISP 18 years ago the Spectrum was sort of desert. Now it's more like the battle of Austerlitz with too many parties and even the big army's (cell operators with their G4) make a mess with low coverage and illegal towers.... It's a war out here and if there would be any arbitrage 80% had to shut down and probably the richest (or the best 'friends in high places' = money) win. That is not me.....
As long as we all earn our daily bread and people get their internet we will survive... And MT makes it possible!

MT Could have much better Soloutions if they here there coustomers

pe1chl · Wed Aug 07, 2019 10:55 am

I know some operators even work 'out of band' in 100% illegal frequencies, some in 'grey' area (legal but too much power....) etc. just to keep the business going. Nobody denounces since everybody will see it then boomerang back.....

Is there no telecom authority in Spain?
Here in the Netherlands we get calls and/or letters from them when an outside AP operates on the indoor (low) channels, or on a radar channel without using DFS, or outside the band.
Operating too much power is a bit harder for them to prove. But on several occasions they actually mapped the frequencies of tower-mounted APs.

Wed Aug 07, 2019 11:46 am

I know some operators even work 'out of band' in 100% illegal frequencies, some in 'grey' area (legal but too much power....) etc. just to keep the business going. Nobody denounces since everybody will see it then boomerang back.....
Is there no telecom authority in Spain?
Here in the Netherlands we get calls and/or letters from them when an outside AP operates on the indoor (low) channels, or on a radar channel without using DFS, or outside the band.
Operating too much power is a bit harder for them to prove. But on several occasions they actually mapped the frequencies of tower-mounted APs.

There are comparable rules/laws. But very different efforts enforcing this. And there are different capabilities (my guys here have an anritsu but bad antennas). I follow rules to avoid problems. And I see the sense in these rules. (Almost, as these DFS rules are not practicable (lead to service interuptions at times) and no vendor has a good/correct handling).
One reason we move away from MT Wireless is: You cant follow the rules and keep the service going smooth.
There is still no ATPC. MT follows the rules and set TX-Power to the lower powerlimit. So if you enter correct antenna gain you end up with low TX-power.
So WirelessRudy operates illegal for sure to get reasonable Signal. He will not call regulations as they may shut him down. You cant enforce a rule you break.
I gave up on asking for ATPC for a long time and decided to use a vendor which is able to implement it and allows me to operate with higher power according to the rules. If everybody would follow rules MT had implemented ATPC long time ago. But there are too much cheaters and "amateurs" (which do not see the benefit of ATPC for their self-interference). So they dont care.

WirelessRudy · Wed Aug 07, 2019 1:01 pm

I am from Holland too and know the differences between the two countries. But where in Holland the rules are strict, and policing is everywhere, in Spain only those that get denounced by others get a checkup.
It's with housing rules, business rules, safety rules, traffic rules etc. etc. You can basically do what you want until someone doesn't like it/you and denounces it/you.
In Holland a policeman seeing something that could be considered illegal or just lacking documentations would trigger him to ask/investigate it and if found something not in order shut things down.
In Spain the police doesn't do nothing if somebody has not been asking him to do so.... (Well, traffic nowadays is slightly different). Not seldom that same police officer (or major, civil servant etc.) is not having all papers for his house/reform/whatever rule in place himself...
In Holland you have dedicated civil servants that are looking around if all that falls under his area of interest is having the proper licenses etc. In Spain most townhals haven't and even at national or provincial levels they are ample.

I always stated, and still think this is valid, that if all WISP's in Spain tomorrow would work according regulations, the day after 60% of fixed wireless clients wouldn't have no more service and by the end of the month 50% of all WISP would go bankrupt...
Frequencies 'out of band' or in the lower band, too high signal levels (Spain doesn't even allow directional antenna in the free 5Ghz band or you have to tune the power output that low nothing comes out of it anymore!) is more the standard way of doing things here .... off course nobody openly admits, but I see what I see when I run a scan....
I tried Mimosa in regulatory domain and almost all my clients disconnect. Same with Mikrotik. Or signals become that bad that no more proper service can be delivered.
Running DFS? Well, we have rain radar and an approach path of one of Span's busiest airfield over our valley. If we have to work in those frequencies that have no influence of these two I am left with almost nothing. If possible (Mikrotik) I switch it off. In Mimosa we cannot but in using 'grey' frequency at least it doesn't harm the network every minute. Now the network just goes down for almost 10 minutes when we reboot the Mimosa AP.. I can live with that....
Basically all my competitors, even the national operating ones, use frequencies and EIRPS way out of the regulations....
Also, once I've red an article stating that in Spain probably some 16.000 (!) communications towers on the top of mountains are lacking any form of documentation or permits and not seldom are protruding local as well as national laws.... Many towers it's not even known who owns them....

Our reason to work as much as possible with frequencies inside the allowed band is not out of abiding to the law... more that I don't want to give my competitor too much sticks to hit me with in front of a judge...
If tomorrow all housing laws would be enforce I would probably lose half of my clients too... this is rural Spain!

Fri Aug 23, 2019 11:13 am

So ordered my first laptop with 802.11ax. Where is wAP ax?

Fri Aug 23, 2019 11:15 am

Coming

honzam · Fri Aug 23, 2019 12:02 pm

Coming

Wau

mfr476 · Fri Aug 23, 2019 2:21 pm

Only 802.11x wap?

Fri Aug 23, 2019 2:30 pm

I didn't say wAP and I didn't say when

But yes, we are always working on new and exciting products

Fri Aug 23, 2019 2:49 pm

I didn't say wAP and I didn't say when But yes, we are always working on new and exciting products

Dont step back

WirelessRudy · Fri Aug 23, 2019 2:53 pm

I didn't say wAP and I didn't say when But yes, we are always working on new and exciting products
Dont step back

cmanuelsantos · Sat Aug 24, 2019 7:33 am

@WirelessRudy,
Do you still use Mimosa? Is it woth deploying in interop mode?

anuser · Sat Aug 24, 2019 11:20 am

Guys, perhaps you heard about a tiny vendor called Hew Pac Ent I think. They started selling pre-802.11ax hardware since last year. After 1 year they finally published an updated roadmap for 802.11ax software features:

So you hopefully understand that making software is hard for a smaller vendor especially also for that one coming from a small cottage on Aruba.

Sat Aug 24, 2019 11:47 am

Guys, perhaps you heard about a tiny vendor called Hew Pac Ent I think. They started selling pre-802.11ax hardware since last year. After 1 year they finally published an updated roadmap for 802.11ax software features:
So you hopefully understand that making software is hard for a smaller vendor especially also for that one coming from a small cottage on Aruba.

Not that hard for a vendor doing this stuff for a longer time now like MT. They are used to take Atheros chipsets and build their gear based on this. They managed to build this 60GHz gear so I guess a basic implementation of 802.ax would not be magic for MT. Tuning this for WISP usage is a another step which might be harder. This heavily depends on developer kits from chipset vendors and what is already done by them.

server8 · Sat Aug 24, 2019 4:02 pm

Coming

The main question what and when

More info are welcome

r00t · Sat Aug 24, 2019 4:23 pm

Coming

Certainly good news... but would be nice to know if it's weeks, months or years before it's available... at least give us some timeframe, like Q1-2020 etc.

anuser · Sat Aug 24, 2019 6:29 pm

Time to Market would be rather short, If you take the design of the RB4011, add 2 antennas for 2.4 GHz, remove the QCA9984 and the R11e-2HnD and put in 2 QCA6390 so finally can achieve a Quad Radio 802.11ax access point with 2x 2.4GHz and 2x 5.0GHz. This would be something no other vendor currently offers.
Also Reinis will have a working MU-MIMO firmware ready when this device arrives.

server8 · Sat Aug 24, 2019 7:10 pm

We need asap something like EPMP3000 mu-mimo 4x4 with GPS sync compatible with NV2 and 802.11N/AC client

WirelessRudy · Sat Aug 24, 2019 7:26 pm

We need asap something like EPMP3000 mu-mimo 4x4 with GPS sync compatible with NV2 and 802.11N/AC client

YES!
Beamforming, mu-mimo 4 and all the rest will bring back MT on my top wanted gear again!

WirelessRudy · Sat Aug 24, 2019 7:32 pm

@WirelessRudy,
Do you still use Mimosa? Is it woth deploying in interop mode?

I still use is yes. But am not investing any further in it. I can do almost the same with MT gear for a fraction of the costs.
Also, further developments in Mimosa go slow. MU-Mimo promised years ago only since a few months available if you are ready to invest in new AP's. My previous ones are still not paid off.....
Mimosa 'interop' mode? Didn't use is since the initial test case I had. It worked fine but excludes GPS sync. Something I do need to run several AP's in 80Mhz band in dense urbanised area.

I am waiting in better performing sync for Mikrotik and Mu-Mimo and possibly beamforming in 5Ghz mode. That would last my marriage with Mikrotik last much longer....

mistry7 · Sat Aug 24, 2019 9:45 pm

We need asap something like EPMP3000 mu-mimo 4x4 with GPS sync compatible with NV2 and 802.11N/AC client

and spectral scan / history working

anuser · Sun Aug 25, 2019 10:15 am

One question I asked myself multiple times is why MikroTik doesn't use the drivers from chipset vendors. There are some different opinions about this:
1. Philosophy: MikroTik doesn't want to be depending on the vendor. Keep full control of RouterOS. If this is the management's decision, that's it.
- Technical reasons: Perhaps it is not possible to integrate certain drivers into current RouterOS.
- Licensing costs: I read Qualcomm charges for each mobile device being sold. I don't know if this is also true for routers. But how much more would a cAP ac cost If the Qualcomm driver was used? 99$ instead of 69$?
With 802.11ax MikroTik could start with completely new high performance wireless devices that utilize chipset vendors drivers. Let us customers charge for those better devices. We will happily pay the higher price.

Sun Aug 25, 2019 10:43 am

One question I asked myself multiple times is why MikroTik doesn't use the drivers from chipset vendors.

I believe the main reason was the ability to implement protocols like nstream and nv2.

pe1chl · Sun Aug 25, 2019 12:11 pm

I believe the main reason was the ability to implement protocols like nstream and nv2.

Ok, but when I understand correctly, nstream and nv2 can be phased out once we have 802.11ax ?
Maybe there could be 2 drivers which are selected depending on the protocol mode in use?

WirelessRudy · Sun Aug 25, 2019 2:24 pm

One question I asked myself multiple times is why MikroTik doesn't use the drivers from chipset vendors.
I believe the main reason was the ability to implement protocols like nstream and nv2.

I get better results in 802.11ac anyway. If tuned properly it outperforms NV2 and nStreame in almost any circumstance. So maybe they continue to build on the standard drivers and make it even better.
(Although we still would need sync, and for that tdma in any form is needed......)

Thu Aug 29, 2019 3:06 pm

So qualcomm announced their second generation of 802.11ax Chipsets:

https://www.qualcomm.com/products/qualc ... 0-platform

cmanuelsantos · Thu Sep 05, 2019 8:32 am

@WirelessRudy,
Do you still use Mimosa? Is it woth deploying in interop mode?
I still use is yes. But am not investing any further in it. I can do almost the same with MT gear for a fraction of the costs.
Also, further developments in Mimosa go slow. MU-Mimo promised years ago only since a few months available if you are ready to invest in new AP's. My previous ones are still not paid off.....
Mimosa 'interop' mode? Didn't use is since the initial test case I had. It worked fine but excludes GPS sync. Something I do need to run several AP's in 80Mhz band in dense urbanised area.

I am waiting in better performing sync for Mikrotik and Mu-Mimo and possibly beamforming in 5Ghz mode. That would last my marriage with Mikrotik last much longer....

What would you do if mikrotik does not release a stable tdma protocol or new Access Points. Netmetal is a very old AP. And in the future products shown in the mums they don't have plans for new wireless products.

Mikrotik seems to be focused on routing and the wireless market seems to be not their interest.

I need a very good tdma protocol since even noise floor here is -79.

Bases on your reviews I am trying Mimosa. But in the end I think I will go with Cambium if nothing new from mikrotik next month.

taduikis · Thu Sep 05, 2019 8:54 am

I repeat once again:
- backward compatibility for legacy devices
- asymmetric manual downlink/uplink rate selection
- automatic closed-loop tx-power control
- adjacent channel filtering (like airprism or hypure)
- gps sync for channel reuse

This was the recipe (in given order) for MT to save their wireless. I said that in multiple MUMs, here on forum and via email when communicating with their support. But I guess they know better or simply do not have necessary human resources to do that. Some are only software-based features requiring only programming. Unfortunately it’s too late for them now. I, for example, haven’t bought anything MT wireless for more than a year now and those who have jumped the boat with much financial pain, I doubt ever going back to MT again whatever they introduce. It’s simply a matter of trust and faith now.

mistry7 · Thu Sep 05, 2019 10:39 am

What would you do if mikrotik does not release a stable tdma protocol or new Access Points. Netmetal is a very old AP. And in the future products shown in the mums they don't have plans for new wireless products.

Mikrotik seems to be focused on routing and the wireless market seems to be not their interest.

I need a very good tdma protocol since even noise floor here is -79.

Bases on your reviews I am trying Mimosa. But in the end I think I will go with Cambium if nothing new from mikrotik next month.

And it is time to tell that costumers, but they don't do.....
I can only say, don't buy any new Mikrotik staff because it is one way road....
There is only cheapest crap available, only plastic, no shielding, we have polluted spectrum here,
and Mikrotik hast no spectral scan, this Hardware is unusable!

in 2017 we spend about 25k in Mikrotik, this year something between 2-3k (only switches)
They lost these Market.......

The reasons?
- NV2 delivers about 30% of competitors speed (802.11 is not usable in crowded spectrum)
- NV2 latency is scrap (Mimosa SRS delivers stable 1-2ms, P2P without GPS)
- spectral scan is needed to find working slot
- 802.11ac Hardware perform worse then 802.11n

WirelessRudy · Sun Sep 08, 2019 1:39 am

What would you do if mikrotik does not release a stable tdma protocol or new Access Points. Netmetal is a very old AP. And in the future products shown in the mums they don't have plans for new wireless products.

Mikrotik seems to be focused on routing and the wireless market seems to be not their interest.

I need a very good tdma protocol since even noise floor here is -79.

Bases on your reviews I am trying Mimosa. But in the end I think I will go with Cambium if nothing new from mikrotik next month.
And it is time to tell that costumers, but they don't do.....
I can only say, don't buy any new Mikrotik staff because it is one way road....
There is only cheapest crap available, only plastic, no shielding, we have polluted spectrum here,
and Mikrotik hast no spectral scan, this Hardware is unusable!

in 2017 we spend about 25k in Mikrotik, this year something between 2-3k (only switches)
They lost these Market.......

The reasons?
- NV2 delivers about 30% of competitors speed (802.11 is not usable in crowded spectrum)
- NV2 latency is scrap (Mimosa SRS delivers stable 1-2ms, P2P without GPS)
- spectral scan is needed to find working slot
- 802.11ac Hardware perform worse then 802.11n

Don't agree:

On almost none of my P2MP networks I use NV2. 802.11ac is much better, even in crowded spectrum. We deliver speeds (50 to 100Mbps) to clients where other providers don't. And we still make clients from other wireless providers that in general run the latest UBNT line of products.
We also have 4 Mimosa AP's and although their needed investment is about 3 times that of Mikrotik the extra capacity is only 10-25% at best. And their 48V Ethernet adapters are very vulnerable. Every thunderstorm we lose some. We almost never loose Mikrotik units in the same storms.. (and we do have at least 3 times as much of them....
Last storm we had we lost 6 C5's out of 100 but 0 of 600 Mikrotiks. Now we needed to order new C5's we found they are EOL and we need to buy the new units. ROI for these is 2 years so that is simply not feasible anymore unless the clients pays the whole investment. Some do, other don't and we simply said goodbye to them. Mimosa is a dead end for us....

Spectral scan is needed yes, but in Mimosa its pretty useless as well. It doesn't tell me SSID's nor protocols, only noise. In a crowded spectrum you need a lot more info than only a signal graph that isn't even very 'ergonomic' to work with.
We also have one eCambium EPMS2000 with beamforming. But although their radio works good the price tag is even higher and to run a spectral scan all clients will be dropped. Un usable....

In regard to 802.11ac versus 'n' with Mikrotik. We are hastily upgrading all 'n' networks to 'ac' since we double the capacity of our networks and its much better resilient towards interferences and overlapping AP's. (But in plain 802.11ac mode!)
You guys must be doing something wrong.

example:
We have a coverage by 4 Mimosa A5-AP's that server some 110 clients and one Netmetal serving 40 clients (Mimosa's have 35-40 clients).
They all work in the same housing estate within a circle of only 500 meters.
They all work in 80Mhz wide channel where 2 x 2 Mimosa's are in sync.
Apart from that we have at least 8 sectors from competitors with ubnt pointing towards this estate.
And we have at least some 10 802.11n/ac backhauls leaving/pointing from/towards this estate, 2 of these are airfibres that are run by me.

So, any frequency we use have some overlap with another sector with signal in the range of -60/-80 range.

If we do speedtest from the clients towards our main router we can reach 200-250Mbps on the Mimosa clients under good conditions (not a lot of other traffic) and running on several clients at the same time I saw a couple of times the 300Mbps mark reached.
When we do the same on the Mikrotik clients we can now run 180-220Mbps to clients and aggregated I can push a Netmetal up to 200-230Mbps (35 associated clients!)

If we not take in consideration the easy of the Mikrotik ROS compared to the limited OS for the Mimosa's (all IP level!), the price difference, the vulnerability of the ethernet ports for power spikes and water ingress, and the slow progress of their technology then Mikrotik is again my preferred brand....

Yes Mikrotik could do a lot better. And yes I hope they come up with new radio's.
But we have also several 60HGhz Mikrotik links and 4 sectors and they work splendid with speeds up to the full GB!
We tried Metrolink before but they are all facing the bins now... Never buy them again. Very expensive, poor support, lack of troubleshooting, no follow ups on promised RMS's and the 5Ghz failover is useless. We preferred to have our 'old' Mikrotik 5Ghz backhauls for backup. They at least always worked and with proper BGP failover it works fine when their 60Ghz went down...

cmanuelsantos · Sun Sep 08, 2019 7:52 pm

What would you do if mikrotik does not release a stable tdma protocol or new Access Points. Netmetal is a very old AP. And in the future products shown in the mums they don't have plans for new wireless products.

Mikrotik seems to be focused on routing and the wireless market seems to be not their interest.

I need a very good tdma protocol since even noise floor here is -79.

Bases on your reviews I am trying Mimosa. But in the end I think I will go with Cambium if nothing new from mikrotik next month.
And it is time to tell that costumers, but they don't do.....
I can only say, don't buy any new Mikrotik staff because it is one way road....
There is only cheapest crap available, only plastic, no shielding, we have polluted spectrum here,
and Mikrotik hast no spectral scan, this Hardware is unusable!

in 2017 we spend about 25k in Mikrotik, this year something between 2-3k (only switches)
They lost these Market.......

The reasons?
- NV2 delivers about 30% of competitors speed (802.11 is not usable in crowded spectrum)
- NV2 latency is scrap (Mimosa SRS delivers stable 1-2ms, P2P without GPS)
- spectral scan is needed to find working slot
- 802.11ac Hardware perform worse then 802.11n
Don't agree:

On almost none of my P2MP networks I use NV2. 802.11ac is much better, even in crowded spectrum. We deliver speeds (50 to 100Mbps) to clients where other providers don't. And we still make clients from other wireless providers that in general run the latest UBNT line of products.
We also have 4 Mimosa AP's and although their needed investment is about 3 times that of Mikrotik the extra capacity is only 10-25% at best. And their 48V Ethernet adapters are very vulnerable. Every thunderstorm we lose some. We almost never loose Mikrotik units in the same storms.. (and we do have at least 3 times as much of them....
Last storm we had we lost 6 C5's out of 100 but 0 of 600 Mikrotiks. Now we needed to order new C5's we found they are EOL and we need to buy the new units. ROI for these is 2 years so that is simply not feasible anymore unless the clients pays the whole investment. Some do, other don't and we simply said goodbye to them. Mimosa is a dead end for us....

Spectral scan is needed yes, but in Mimosa its pretty useless as well. It doesn't tell me SSID's nor protocols, only noise. In a crowded spectrum you need a lot more info than only a signal graph that isn't even very 'ergonomic' to work with.
We also have one eCambium EPMS2000 with beamforming. But although their radio works good the price tag is even higher and to run a spectral scan all clients will be dropped. Un usable....

In regard to 802.11ac versus 'n' with Mikrotik. We are hastily upgrading all 'n' networks to 'ac' since we double the capacity of our networks and its much better resilient towards interferences and overlapping AP's. (But in plain 802.11ac mode!)
You guys must be doing something wrong.

example:
We have a coverage by 4 Mimosa A5-AP's that server some 110 clients and one Netmetal serving 40 clients (Mimosa's have 35-40 clients).
They all work in the same housing estate within a circle of only 500 meters.
They all work in 80Mhz wide channel where 2 x 2 Mimosa's are in sync.
Apart from that we have at least 8 sectors from competitors with ubnt pointing towards this estate.
And we have at least some 10 802.11n/ac backhauls leaving/pointing from/towards this estate, 2 of these are airfibres that are run by me.

So, any frequency we use have some overlap with another sector with signal in the range of -60/-80 range.

If we do speedtest from the clients towards our main router we can reach 200-250Mbps on the Mimosa clients under good conditions (not a lot of other traffic) and running on several clients at the same time I saw a couple of times the 300Mbps mark reached.
When we do the same on the Mikrotik clients we can now run 180-220Mbps to clients and aggregated I can push a Netmetal up to 200-230Mbps (35 associated clients!)

If we not take in consideration the easy of the Mikrotik ROS compared to the limited OS for the Mimosa's (all IP level!), the price difference, the vulnerability of the ethernet ports for power spikes and water ingress, and the slow progress of their technology then Mikrotik is again my preferred brand....

Yes Mikrotik could do a lot better. And yes I hope they come up with new radio's.
But we have also several 60HGhz Mikrotik links and 4 sectors and they work splendid with speeds up to the full GB!
We tried Metrolink before but they are all facing the bins now... Never buy them again. Very expensive, poor support, lack of troubleshooting, no follow ups on promised RMS's and the 5Ghz failover is useless. We preferred to have our 'old' Mikrotik 5Ghz backhauls for backup. They at least always worked and with proper BGP failover it works fine when their 60Ghz went down...

You are saing that you can connect more clients with 802.11ac than using NV2?
For some reason when using pure 802.11 I get clients disconnecting, in mu situation.
Maybe the noise floor here is higher.
One clear advantage of using NV2 is that if you have a client who does not have line of sight performance is still regular.
I guess 802.11 does not have that advantage.

Anyway I will consider using pure 802.11AC from mikrotik for small sites and very close clients.

Sun Sep 08, 2019 8:32 pm

What would you do if mikrotik does not release a stable tdma protocol or new Access Points. Netmetal is a very old AP. And in the future products shown in the mums they don't have plans for new wireless products.

Mikrotik seems to be focused on routing and the wireless market seems to be not their interest.

I need a very good tdma protocol since even noise floor here is -79.

Bases on your reviews I am trying Mimosa. But in the end I think I will go with Cambium if nothing new from mikrotik next month.
And it is time to tell that costumers, but they don't do.....
I can only say, don't buy any new Mikrotik staff because it is one way road....
There is only cheapest crap available, only plastic, no shielding, we have polluted spectrum here,
and Mikrotik hast no spectral scan, this Hardware is unusable!

in 2017 we spend about 25k in Mikrotik, this year something between 2-3k (only switches)
They lost these Market.......

The reasons?
- NV2 delivers about 30% of competitors speed (802.11 is not usable in crowded spectrum)
- NV2 latency is scrap (Mimosa SRS delivers stable 1-2ms, P2P without GPS)
- spectral scan is needed to find working slot
- 802.11ac Hardware perform worse then 802.11n
Don't agree:

On almost none of my P2MP networks I use NV2. 802.11ac is much better, even in crowded spectrum. We deliver speeds (50 to 100Mbps) to clients where other providers don't. And we still make clients from other wireless providers that in general run the latest UBNT line of products.
We also have 4 Mimosa AP's and although their needed investment is about 3 times that of Mikrotik the extra capacity is only 10-25% at best. And their 48V Ethernet adapters are very vulnerable. Every thunderstorm we lose some. We almost never loose Mikrotik units in the same storms.. (and we do have at least 3 times as much of them....
Last storm we had we lost 6 C5's out of 100 but 0 of 600 Mikrotiks. Now we needed to order new C5's we found they are EOL and we need to buy the new units. ROI for these is 2 years so that is simply not feasible anymore unless the clients pays the whole investment. Some do, other don't and we simply said goodbye to them. Mimosa is a dead end for us....

Spectral scan is needed yes, but in Mimosa its pretty useless as well. It doesn't tell me SSID's nor protocols, only noise. In a crowded spectrum you need a lot more info than only a signal graph that isn't even very 'ergonomic' to work with.
We also have one eCambium EPMS2000 with beamforming. But although their radio works good the price tag is even higher and to run a spectral scan all clients will be dropped. Un usable....

In regard to 802.11ac versus 'n' with Mikrotik. We are hastily upgrading all 'n' networks to 'ac' since we double the capacity of our networks and its much better resilient towards interferences and overlapping AP's. (But in plain 802.11ac mode!)
You guys must be doing something wrong.

example:
We have a coverage by 4 Mimosa A5-AP's that server some 110 clients and one Netmetal serving 40 clients (Mimosa's have 35-40 clients).
They all work in the same housing estate within a circle of only 500 meters.
They all work in 80Mhz wide channel where 2 x 2 Mimosa's are in sync.
Apart from that we have at least 8 sectors from competitors with ubnt pointing towards this estate.
And we have at least some 10 802.11n/ac backhauls leaving/pointing from/towards this estate, 2 of these are airfibres that are run by me.

So, any frequency we use have some overlap with another sector with signal in the range of -60/-80 range.

If we do speedtest from the clients towards our main router we can reach 200-250Mbps on the Mimosa clients under good conditions (not a lot of other traffic) and running on several clients at the same time I saw a couple of times the 300Mbps mark reached.
When we do the same on the Mikrotik clients we can now run 180-220Mbps to clients and aggregated I can push a Netmetal up to 200-230Mbps (35 associated clients!)

If we not take in consideration the easy of the Mikrotik ROS compared to the limited OS for the Mimosa's (all IP level!), the price difference, the vulnerability of the ethernet ports for power spikes and water ingress, and the slow progress of their technology then Mikrotik is again my preferred brand....

Yes Mikrotik could do a lot better. And yes I hope they come up with new radio's.
But we have also several 60HGhz Mikrotik links and 4 sectors and they work splendid with speeds up to the full GB!
We tried Metrolink before but they are all facing the bins now... Never buy them again. Very expensive, poor support, lack of troubleshooting, no follow ups on promised RMS's and the 5Ghz failover is useless. We preferred to have our 'old' Mikrotik 5Ghz backhauls for backup. They at least always worked and with proper BGP failover it works fine when their 60Ghz went down...
You are saing that you can connect more clients with 802.11ac than using NV2?
For some reason when using pure 802.11 I get clients disconnecting, in mu situation.
Maybe the noise floor here is higher.
One clear advantage of using NV2 is that if you have a client who does not have line of sight performance is still regular.
I guess 802.11 does not have that advantage.

Anyway I will consider using pure 802.11AC from mikrotik for small sites and very close clients.

He cranks up power until all clients are hot enough to stay connected. Nothing which scales but he lives in a hostile environment where every radio cries out loud. If you want to build a sane good performing network you need gps sync, atpc and good antennas.

anuser · Mon Sep 09, 2019 4:45 pm

- 802.11ac Hardware perform worse then 802.11n

Well, not really.I did some tests with a single Laptop, Windows 10, Intel 8265 connected at 866 Mbit/s, clear spectrum
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)

TCP
1. Mikrotik cAP AC
Interval Transfer Bandwidth
0.00-5.00 sec 154 MBytes 258 Mbits/sec
5.00-10.00 sec 157 MBytes 264 Mbits/sec
10.00-15.00 sec 179 MBytes 301 Mbits/sec
15.00-20.00 sec 178 MBytes 298 Mbits/sec
20.00-25.00 sec 175 MBytes 294 Mbits/sec
25.00-30.00 sec 186 MBytes 313 Mbits/sec
30.00-35.00 sec 163 MBytes 273 Mbits/sec
35.00-40.00 sec 163 MBytes 274 Mbits/sec
40.00-45.00 sec 137 MBytes 230 Mbits/sec
45.00-50.00 sec 97.8 MBytes 164 Mbits/sec
50.00-50.04 sec 1017 KBytes 213 Mbits/sec

2. Cambium E410
Interval Transfer Bandwidth
0.00-5.00 sec 175 MBytes 294 Mbits/sec
5.00-10.00 sec 177 MBytes 296 Mbits/sec
10.00-15.00 sec 176 MBytes 295 Mbits/sec
15.00-20.00 sec 170 MBytes 285 Mbits/sec
20.00-25.00 sec 171 MBytes 287 Mbits/sec
25.00-30.00 sec 174 MBytes 293 Mbits/sec
30.00-35.00 sec 170 MBytes 284 Mbits/sec
35.00-40.00 sec 177 MBytes 297 Mbits/sec
40.00-45.00 sec 171 MBytes 288 Mbits/sec
45.00-50.00 sec 179 MBytes 301 Mbits/sec
50.00-50.04 sec 1.37 MBytes 320 Mbits/sec

UDP
1. MikroTik cAP ac
Interval Transfer Bandwidth Jitter Lost/Total Datagrams
0.00-5.00 sec 140 MBytes 235 Mbits/sec 0.050 ms 11280/29163 38%
5.00-10.00 sec 138 MBytes 232 Mbits/sec 0.129 ms 12188/29888 41%
10.00-15.00 sec 138 MBytes 231 Mbits/sec 0.439 ms 12148/29754 41%
15.00-20.00 sec 141 MBytes 237 Mbits/sec 0.321 ms 11506/29577 39%
20.00-25.00 sec 144 MBytes 242 Mbits/sec 0.045 ms 11678/30162 39%
25.00-30.00 sec 143 MBytes 240 Mbits/sec 0.286 ms 12023/30350 40%
30.00-35.00 sec 139 MBytes 233 MBits/sec 0.107 ms 13072/30863 42%
35.00-40.00 sec 146 MBytes 244 MBits/sec 0.122 ms 10546/29171 36%
40.00-45.00 sec 146 MBytes 244 MBits/sec 0.156 ms 13234/31881 42%
45.00-50.00 sec 138 MBytes 231 Mbits/sec 0.095 ms 14926/32562 46%
50.00-50.05 sec 960 KBytes 159 MBits/sec 0.342 ms 182/302 60%

2. Cambium E410
Interval Transfer Bandwidht Jitter Lost/Total Datagrams
0.00-5.00 sec 166 MBytes 278 Mbits/sec 0.214 ms 10218/31467 32%
5.00-10.00 sec 161 MBytes 270 Mbits/sec 0.279 ms 5195/25750 20%
10.00-15.00 sec 175 MBytes 293 Mbits/sec 0.135 ms 9544/32924 30%
15.00-20.00 sec 166 MBytes 279 Mbits/sec 0.309 ms 12749/34044 37%
20.00-25.00 sec 161 MBytes 271 Mbits/sec 0.358 ms 11902/32564 37%
25.00-30.00 sec 166 MBytes 278 Mbits/sec 0.099 ms 11932/33164 36%
30.00-35.00 sec 168 MBytes 282 Mbits/sec 0.432 ms 8056/29558 27%
35.00-40.00 sec 161 MBytes 270 Mbits/sec 0.072 ms 9822/90396 32%
40.00-45.00 sec 164 MBytes 276 Mbits/sec 0.366 ms 8043/29075 28%
45.00-50.00 sec 164 MBytes 275 Mbits/sec 0.199 ms 10414/31388 33%
50.00-50.05 sec 952 KBytes 184 Mbits/sec 0.440 ms 149/268 56%

Mon Sep 09, 2019 5:38 pm

- 802.11ac Hardware perform worse then 802.11n
Well, not really.I did some tests with a single Laptop, Windows 10, Intel 8265 connected at 866 Mbit/s, clear spectrum
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)

...

Quite slow for a 80MHz Channel. With proprietary 5GHz Gear I get near 600MBit capacity on a 50MHz Channel. Hopefully .ax shows better numbers/spectral efficiency.

WirelessRudy · Mon Sep 09, 2019 6:22 pm

He cranks up power until all clients are hot enough to stay connected. Nothing which scales but he lives in a hostile environment where every radio cries out loud. If you want to build a sane good performing network you need gps sync, atpc and good antennas.

True. Anything worse than -60 is considered a 'bad' client. But that same counts for our Mimosa's. They are nothing better in MCS rates then Mikrotik.
The only difference I noticed some time ago (over a year) is where two chains have a difference over 3-4db in signal Mikrotik seems to use only one chain (the best one) for data transmission.
I have seen Mimosas that had more than 8dB difference in signal over the two chains but still mcs (and real data transport) represented the use of both chains.
How that actually is done I don't know but that has been one of the advantages I found in Mimosa over Mikrotik 2 years ago when I tested them side by side.

(In fact, the A5 AP needs to be able to do that since it has only one array antenna pointing towards the client where Mikrotik AP's always have two. So it is pretty common for a Mimosa AP to have quite a difference between the two used chains. That is probably also the reason why at bigger distances (over 500meters) Mimosa only uses single chain MCS simply because the 'not facing' antenna strips don't pick up signal (or send) from the client anymore where the MT always has two arrays pointing in the same direction.

The CPU of the Mimosa AP's is much faster then the Netmetal so that is the advantage in P2MP where signals and multi channel receipt is good.
But limit in the Mimosa radio's power is a disadvantage.

If I was on my own, like 10 years ago, yes we could get good results with even -80 or 85 results. Off course, demands were less too.
But now where the competition hits me with signals from remote locations that can hardly be met by my own radio's pointing in the same direction there is not other way than also go for full power and big antennas. -50 at the client is the aim, -45 ideal. -40 superb but stronger becomes questionable. (Although I have worked with signals in the past of -20 (yes, even -10!) it never did 'melt' or break a Mikrotik radio...... they still worked fine......

syadnom · Mon Sep 09, 2019 8:03 pm

Mimosa uses the Quantenna chipset which is a BEAST. That said, their deployment model doesn't seem to take advantage of all it's power, and they still don't have a firmware with proper MU-MIMO despite chipset support.

If you are talking any apples to apples scenario though, Mimosa will beat mikrotik and ubiquiti's qualcomm hand's down.

The thing about 'ax is that it takes all the secret sauce out of the mix and basically levels the playing field. Airmax? legacy junk. NV2, NStreme? legacy junk. Custom TDMA implementation on top of 802.11 silicon? legacy junk.

I really expect big things out of Mikrotik and Ignitenet on this front. Two companies that either have a flexible enough platform to just need a drive (Mikrotik) or a business agility to early adopt (Ignitenet).

I should add that I have LTU gear, and I've done some testing on epmp3k. On technical merits, I think gen1 'ax is going to outperform this in even a basic configuration until density gets high enough that GPS sync is needed. At that point, 'ax does have an extension that GPS sync fits so that should be a pretty straight forward development.

I think 'ax is a real game changer and only LTE in licensed bands or CBRS will be a real competitor because if it's superior sensitivity and clean bands.

cmanuelsantos · Tue Sep 10, 2019 5:53 am

If we do speedtest from the clients towards our main router we can reach 200-250Mbps on the Mimosa clients under good conditions (not a lot of other traffic) and running on several clients at the same time I saw a couple of times the 300Mbps mark reached.
When we do the same on the Mikrotik clients we can now run 180-220Mbps to clients and aggregated I can push a Netmetal up to 200-230Mbps (35 associated clients!)

Rudy,
I tried to reproduce your test using 802.11.
I set an AP with 26 clients connected. I am using mostly SXT Lite 5AC but I have like 6 clients that are either SXT Lite 5 or LHG 5 (N Clients).
I have setup all clients to use RTS/CTS with hardware protection set to 0 (always on).

When I tested a single client (the AP was using like 20Mbps from the other clients) I could get only 12-15Mbps of traffic using a 40Mhz channel.
I then tried using NV2 from the same client, same channel same everything, I could push 60-80Mbps.
It is important to mention that this was an N client.

I tried the same thing with AC client very closed and I could never exceed 25Mbps using 802.11 protocol.

This means, that 802.11AC could deliver very good bandwidth if the spectrum is not so crowded.

Maybe you have very good contion. But we need a better NV2.

anuser · Tue Sep 10, 2019 8:47 am

I tried the same thing with AC client very closed and I could never exceed 25Mbps using 802.11 protocol.

Have you enabled adaptive-noise-immunity on wireless interfaces, i.e. set it to "AP and Client mode"?

anuser · Tue Sep 10, 2019 8:54 am

- 802.11ac Hardware perform worse then 802.11n
Well, not really.I did some tests with a single Laptop, Windows 10, Intel 8265 connected at 866 Mbit/s, clear spectrum
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)

...

Quite slow for a 80MHz Channel. With proprietary 5GHz Gear I get near 600MBit capacity on a 50MHz Channel. Hopefully .ax shows better numbers/spectral efficiency.

Indeed, only ~300Mbit/s when connected at 866 Mbit/s. But with Cambium I have similar results. IPQ4019 might not be the best chipset?

syadnom · Tue Sep 10, 2019 8:57 am

80Mhz is near impossible to pull off full performance outdoors in 5Ghz. You’d have to be in the wilderness 20 miles from any homes and well below the tree line.

mistry7 · Tue Sep 10, 2019 9:01 am

- 802.11ac Hardware perform worse then 802.11n
Well, not really.I did some tests with a single Laptop, Windows 10, Intel 8265 connected at 866 Mbit/s, clear spectrum
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)

TCP
1. Mikrotik cAP AC
Interval Transfer Bandwidth
0.00-5.00 sec 154 MBytes 258 Mbits/sec
5.00-10.00 sec 157 MBytes 264 Mbits/sec
10.00-15.00 sec 179 MBytes 301 Mbits/sec
15.00-20.00 sec 178 MBytes 298 Mbits/sec
20.00-25.00 sec 175 MBytes 294 Mbits/sec
25.00-30.00 sec 186 MBytes 313 Mbits/sec
30.00-35.00 sec 163 MBytes 273 Mbits/sec
35.00-40.00 sec 163 MBytes 274 Mbits/sec
40.00-45.00 sec 137 MBytes 230 Mbits/sec
45.00-50.00 sec 97.8 MBytes 164 Mbits/sec
50.00-50.04 sec 1017 KBytes 213 Mbits/sec

2. Cambium E410
Interval Transfer Bandwidth
0.00-5.00 sec 175 MBytes 294 Mbits/sec
5.00-10.00 sec 177 MBytes 296 Mbits/sec
10.00-15.00 sec 176 MBytes 295 Mbits/sec
15.00-20.00 sec 170 MBytes 285 Mbits/sec
20.00-25.00 sec 171 MBytes 287 Mbits/sec
25.00-30.00 sec 174 MBytes 293 Mbits/sec
30.00-35.00 sec 170 MBytes 284 Mbits/sec
35.00-40.00 sec 177 MBytes 297 Mbits/sec
40.00-45.00 sec 171 MBytes 288 Mbits/sec
45.00-50.00 sec 179 MBytes 301 Mbits/sec
50.00-50.04 sec 1.37 MBytes 320 Mbits/sec

Point one, did you see how stable the Bandwidth with the Cadmium is?

With my macbook to my NAS over the Ligowave (NFT 3ac Lite) with 80Mhz Channel i do 580-600MBit
and 270-280 Mbit with 40 MhZ Channel.
Impossible with WAP AC

Tue Sep 10, 2019 9:29 am

80Mhz is near impossible to pull off full performance outdoors in 5Ghz. You’d have to be in the wilderness 20 miles from any homes and well below the tree line.

Or use high gain antennas with good isolation on short links.

anuser · Tue Sep 10, 2019 12:57 pm

Point one, did you see how stable the Bandwidth with the Cadmium is?

With my macbook to my NAS over the Ligowave (NFT 3ac Lite) with 80Mhz Channel i do 580-600MBit
and 270-280 Mbit with 40 MhZ Channel.
Impossible with WAP AC

For 1.: Yes it´s not perfect.
Cambium E410 (~220$) and MikroTik cAP ac (~70$) use IPQ4019 chipset with 2x2:2.
What chipset is being used on the Ligowave NFT 3ac Lite (~170$)? What about streams? Is it 3x3:3?

WirelessRudy · Tue Sep 10, 2019 1:36 pm

80Mhz is near impossible to pull off full performance outdoors in 5Ghz. You’d have to be in the wilderness 20 miles from any homes and well below the tree line.

True, but one of the advantages of using 80Mhz channel over 40 or 20 even in a 'noisy' 802.11 environment is the ability of the 'ac' standard to skip one of the 20Mhz channels that has to much interference to be usable. The AP will communicate with that client in 2 20Mhz bands only or even one, that give acceptable results.
This way it is possible to work with overlapping 80Mhz channels.

In a P2MP environment where several AP's have to share the same spectrum and where clients have high gain directional antenas pointed towards the AP some of those clients could have interference on channels A+B but another client on C+D or another on B+C. The AP picks then the best set for each specific client and still has a pretty high CCQ on a 40Mhz channel. Other clients that have no interference are still using the full 80Mhz bandwidth so their traffic clears the AP's time superfast so it has more time to deal with the 'slower' clients.

The problem with working in a mixed environment is that although the 'ac' AP is capable to work with both 'ac' as 'n' clients is that it needs too much standard changes which degrades de network pretty fast. I found in such environment it is better to still use 'n-only' then mixed. So my AP's only go to 'ac-only' if all clients are swapped and straight away 'n' clients are no longer allowed to associate.

I work now with ROS v.6.45.5 and still on some tests I'v done plain 802.11ac works much better for the client than NV2. At least 50% more speed towards the client....

WirelessRudy · Tue Sep 10, 2019 1:49 pm

Mimosa uses the Quantenna chipset which is a BEAST. That said, their deployment model doesn't seem to take advantage of all it's power, and they still don't have a firmware with proper MU-MIMO despite chipset support.

If you are talking any apples to apples scenario though, Mimosa will beat mikrotik and ubiquiti's qualcomm hand's down.

True. In the beginning I had my first Mimosa A5 AP with 60 (!) associated clients and these clients all where Mikrotik SXT-5ac (mipsbe). I could easily run 180Mbps to these clients, or 2 with some 100Mbps and still had some more traffic from the regular clients. Saw 250-300Mbps traffic aggregated over the AP....
That a wet dream for Mikrotik still...

We then converted the whole area in 4x Mimosa A5 with 120 C5's and 2 x 2 pairs syncd. All in 80Mhz channel. But to be honest, the huge investment didn't really pay itself off. We still have interferences at times, we have some 15% failures of the C5's in 3 years time (which is huge! On my 600 mikrotiks I have less then 1% failures over the last 10 years....) which is a big disappointment in hardware reliability. We also have an ever higher failure rate in the G2's that do the power supply of the antennas. We are facing them our for normal (cheaper!) plain 48V supplies with a cheap Wifi router.
Mimosa had many promises when I started with them 3,5 years ago and several still have to materialize. You think Mikrotik is slow ....?

We are stuck with Mimosa now in this region but I am not expanding its use anywhere else what initially was the idea.
3 years ago the difference between Mikrotik performance and Mimosa was worth the investment, but the new ROS of Mikrotik is good enough for me and we manage to get similar results with Mikrotik as with Mimosa for 1/4 of the price tag. And the client? They don't even notice the difference apart from with every rainstorm we lose another C5......
Now the C5's are EOL I am facing to buy their new line of client devices. And although they might be better in hardware, they are also even more expensive then the C5's.
I made some calculations and see that my ROI (return of investment) for this new line of Mimosa's is 2 years or more... it's hardly worth the effort.....

anuser · Tue Sep 10, 2019 4:29 pm

Update see next post

syadnom · Tue Sep 10, 2019 6:23 pm

fire up excel and build some charts with that data!

anuser · Tue Sep 10, 2019 9:46 pm

Update:
Windows 10, Thinkpad x270 with Intel 8265 connected at 866 Mbit/s (2x2:2), clear spectrum, iPerf 3.1.3, about 50 cm between Thinkpad and the access points.
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)
3. Mikrotik RB4011, RouterOS 6.45.5, CH36 (80 MHz)
4. Mikrotik wAP AC, RouterOS 6.45.5, CH36 (80 MHz)
5. Aruba AP-535, ArubaOS 8.3.0.2, CH36 (80 MHz)
6. MikroTik cAP ac, RouterOS 7.0beta1 (development), CH36 (80 MHz)

Thu Sep 19, 2019 5:35 pm

please share the driver version on intel wireless NIC

heidarren · Thu Sep 19, 2019 6:32 pm

Update:
Windows 10, Thinkpad x270 with Intel 8265 connected at 866 Mbit/s (2x2:2), clear spectrum, iPerf 3.1.3, about 50 cm between Thinkpad and the access points.
1. MikroTik cAP ac with IPQ4019, RouterOS 6.45.5, CH36 (80 MHz)
2. Cambium E410 with IPQ4019, CambiumOS 3.11.2, CH36 (80 MHz)
3. Mikrotik RB4011, RouterOS 6.45.5, CH36 (80 MHz)
4. Mikrotik wAP AC, RouterOS 6.45.5, CH36 (80 MHz)
5. Aruba AP-535, ArubaOS 8.3.0.2, CH36 (80 MHz)
6. MikroTik cAP ac, RouterOS 7.0beta1 (development), CH36 (80 MHz)

Happy to see the improvement on cAP AC, seems like they are working hard to bring the best out of it. I'd removed my cAP AC from the listing on some online selling platform, will keep it and see how it performs in ROS v7

Thu Sep 19, 2019 6:40 pm

i think there is a lot of marketing hype around each new wifi generation, this causes a lot of confusion and excessive expectations

Thu Sep 19, 2019 8:09 pm

i think there is a lot of marketing hype around each new wifi generation, this causes a lot of confusion and excessive expectations

Of course. But for sure every generation was a big step forward. Not as big as announced but big.

szalkerous · Tue Sep 24, 2019 10:19 pm

Wow, did this thread derail left and right....

It seems to me that Mikrotik doesn't announce hardware they haven't started final QC on. Many announcements lately are only a month or two before release it feels like.

I have a grand total of 0 ax-compatible devices. Even if I did have one, the cost to replace all my CAP network would prevent me from jumping to the newly announced ax hardware anyways until I have a strong demand for it. ac is simply good enough for now. I'll let the daring folks beta test for me and rotate out my cAP AC units when I have an honest value in doing so.

I'm suspecting that I'm far from the only one.

I've also heard that, at least in the USA, some more of the spectrum is being considered for release to wifi general use. I'm guessing this will require new chips and/or radios, especially if anything in 6Ghz is offered. So I may just wait and see what happens.

P.S. Anyone else hate that they called it "Wifi 6"? Did we really need to simplify the name?

syadnom · Tue Dec 03, 2019 8:39 pm

Trying to get this back on track....

as was mentioned, Wave2 AC can segment transmissions on 20Mhz increments. Ie, an 80Mhz Wave2 AC channel can be 4 separate transmissions to a client. This requires Wave2 APs though, so very limited scope of value here. Wave1 APs won't do this.

802.1ax and OFDMA on the other hand can separate a channel out from 26 to over 400 subchannels and can track and isolate clients into best spectrum pairs. In other words, you can drop a 160Mhz channel down right over the top of other channels and the AP can send as little as about .5Mhz channels through on the cleanest subchannels and the client can be told to transmit similarly so the AP's noise and the client's noise are measured separately.

Think about this for an AP. It can see interference all over the place and so schedule clients to 'upload' in the clean gaps at virtually any beam width to avoid the noise on the upload. Meanwhile, it will track what the clients deem as clean and transmit to them on that. Noise on one client is irrelevant to noise on another client except that the AP might transmit at the same time to both clients thus MU-MIMO.

People that aren't excited about this are truly failing to understand it.

I haven't been able to get my hands on a m.2/mpcie card that can be an AP yet so I can only test the AX200 2x2 cards to an ASUS AX88U AP set on 80+80 overlapping my 3 Unifi APs.
Get this, If I stream a 4k video over a Unifi AC AP to eat up the channel and then run some iperf over the AX AP, the Unifi AC app keeps running. The AX AP avoids the 40Mhz channel that is busy and transmits around it. Latency doesn't fall apart, still pushing over 600Mbps which is a CPU limit on my iperf server. It avoids the in-use channel gracefully.

Think about that for a second. Think about how you might use that capability with a 4x4 Omni (N5-360) for example, or a single 8x8 AP with 4x 90/120's and a 160Mhz channel. Or with 4 2x2 overlapping horns.

All with a built in TDM OFDMA scheduler stock, no special sauce needed.

So, to make the case for Mikrotik to get REALLY serious about this. 'ac mikrotik for PtMP sucks. NV2/Stream simply don't work right as they weren't designed for AC, and 802.11 on AC is still a CSMA model so has the severe issue of hidden node when using in long range PtMP mode.

'AX support would mean instant catch up and overtake to all the other vendors using a TDMA approach like NV2 but functional on AC.

Let me lay out a couple ridiculous part number here that would *change the game* if it came out in ~Q2 2020.

RBX11UAGS-5HPaxD-NM (X for 10 ie >9)
RBX11UAGS-5HPaxQ-NM (Q for Quad)
RBX11UAGS-5HPaxO-NM (O for Octa)

The D model just to have a basic device to 2x2 Omnis and small sectors. Also do a mANT model.
The Q model perfect for a 4x4 Omni and takes direct aim at the A5c+N360 setup that is kind of micro-pops.
The O model is that single super site kit for 4x 2x2 sectors, either 4x 90s or 4x 120's for some overlap, or for high density deployments with the fantastic RFE Asymmetric horns with overlaps doing 90-180 degree shots for ultra density.

IMO, this would put Mikrotik right up on top of the heap and move a LOT of units. No special sauce, just getting the AX driver into routeros (v7?).

mfr476 · Tue Dec 03, 2019 8:55 pm

At the end mikrotik have a dream¡ jajajaja

xbliss · Wed Jan 08, 2020 9:46 pm

Trying to get this back on track....

as was mentioned, Wave2 AC can segment transmissions on 20Mhz increments. Ie, an 80Mhz Wave2 AC channel can be 4 separate transmissions to a client. This requires Wave2 APs though, so very limited scope of value here. Wave1 APs won't do this.

802.1ax and OFDMA on the other hand can separate a channel out from 26 to over 400 subchannels and can track and isolate clients into best spectrum pairs. In other words, you can drop a 160Mhz channel down right over the top of other channels and the AP can send as little as about .5Mhz channels through on the cleanest subchannels and the client can be told to transmit similarly so the AP's noise and the client's noise are measured separately.

Think about this for an AP. It can see interference all over the place and so schedule clients to 'upload' in the clean gaps at virtually any beam width to avoid the noise on the upload. Meanwhile, it will track what the clients deem as clean and transmit to them on that. Noise on one client is irrelevant to noise on another client except that the AP might transmit at the same time to both clients thus MU-MIMO.

People that aren't excited about this are truly failing to understand it.

I haven't been able to get my hands on a m.2/mpcie card that can be an AP yet so I can only test the AX200 2x2 cards to an ASUS AX88U AP set on 80+80 overlapping my 3 Unifi APs.
Get this, If I stream a 4k video over a Unifi AC AP to eat up the channel and then run some iperf over the AX AP, the Unifi AC app keeps running. The AX AP avoids the 40Mhz channel that is busy and transmits around it. Latency doesn't fall apart, still pushing over 600Mbps which is a CPU limit on my iperf server. It avoids the in-use channel gracefully.

Think about that for a second. Think about how you might use that capability with a 4x4 Omni (N5-360) for example, or a single 8x8 AP with 4x 90/120's and a 160Mhz channel. Or with 4 2x2 overlapping horns.

All with a built in TDM OFDMA scheduler stock, no special sauce needed.

So, to make the case for Mikrotik to get REALLY serious about this. 'ac mikrotik for PtMP sucks. NV2/Stream simply don't work right as they weren't designed for AC, and 802.11 on AC is still a CSMA model so has the severe issue of hidden node when using in long range PtMP mode.

'AX support would mean instant catch up and overtake to all the other vendors using a TDMA approach like NV2 but functional on AC.

Let me lay out a couple ridiculous part number here that would *change the game* if it came out in ~Q2 2020.

RBX11UAGS-5HPaxD-NM (X for 10 ie >9)
RBX11UAGS-5HPaxQ-NM (Q for Quad)
RBX11UAGS-5HPaxO-NM (O for Octa)

The D model just to have a basic device to 2x2 Omnis and small sectors. Also do a mANT model.
The Q model perfect for a 4x4 Omni and takes direct aim at the A5c+N360 setup that is kind of micro-pops.
The O model is that single super site kit for 4x 2x2 sectors, either 4x 90s or 4x 120's for some overlap, or for high density deployments with the fantastic RFE Asymmetric horns with overlaps doing 90-180 degree shots for ultra density.

IMO, this would put Mikrotik right up on top of the heap and move a LOT of units. No special sauce, just getting the AX driver into routeros (v7?).

Very interesting. Hence I am doing this for now...
Cheap APs & Routers to pair w low end Mikrotik hAPlite (N) Home/ SOHO setup for Wireless AC? (Non ubiquiti etc)
And I might go with the Tp-link AX 1500 (lowest priced Wi Fi 6 on market) as an AP with Hap Lite as Router/ Manager. Thoughts? Better WiFi with AC & AX until Mktk gets AX etc?

CyB3RMX · Thu Feb 27, 2020 8:25 am

ARM works very well. The only issue I know of, is lack of full compatibility with legacy Nv2 protocol.
Any other problems?

the lack of Spectral-scan....

WirelessRudy · Thu Feb 27, 2020 1:37 pm

Trying to get this back on track....

as was mentioned, Wave2 AC can segment transmissions on 20Mhz increments. Ie, an 80Mhz Wave2 AC channel can be 4 separate transmissions to a client. This requires Wave2 APs though, so very limited scope of value here. Wave1 APs won't do this.

802.1ax and OFDMA on the other hand can separate a channel out from 26 to over 400 subchannels and can track and isolate clients into best spectrum pairs. In other words, you can drop a 160Mhz channel down right over the top of other channels and the AP can send as little as about .5Mhz channels through on the cleanest subchannels and the client can be told to transmit similarly so the AP's noise and the client's noise are measured separately.

Think about this for an AP. It can see interference all over the place and so schedule clients to 'upload' in the clean gaps at virtually any beam width to avoid the noise on the upload. Meanwhile, it will track what the clients deem as clean and transmit to them on that. Noise on one client is irrelevant to noise on another client except that the AP might transmit at the same time to both clients thus MU-MIMO.

People that aren't excited about this are truly failing to understand it.

I haven't been able to get my hands on a m.2/mpcie card that can be an AP yet so I can only test the AX200 2x2 cards to an ASUS AX88U AP set on 80+80 overlapping my 3 Unifi APs.
Get this, If I stream a 4k video over a Unifi AC AP to eat up the channel and then run some iperf over the AX AP, the Unifi AC app keeps running. The AX AP avoids the 40Mhz channel that is busy and transmits around it. Latency doesn't fall apart, still pushing over 600Mbps which is a CPU limit on my iperf server. It avoids the in-use channel gracefully.

Think about that for a second. Think about how you might use that capability with a 4x4 Omni (N5-360) for example, or a single 8x8 AP with 4x 90/120's and a 160Mhz channel. Or with 4 2x2 overlapping horns.

All with a built in TDM OFDMA scheduler stock, no special sauce needed.

So, to make the case for Mikrotik to get REALLY serious about this. 'ac mikrotik for PtMP sucks. NV2/Stream simply don't work right as they weren't designed for AC, and 802.11 on AC is still a CSMA model so has the severe issue of hidden node when using in long range PtMP mode.

'AX support would mean instant catch up and overtake to all the other vendors using a TDMA approach like NV2 but functional on AC.

Let me lay out a couple ridiculous part number here that would *change the game* if it came out in ~Q2 2020.

RBX11UAGS-5HPaxD-NM (X for 10 ie >9)
RBX11UAGS-5HPaxQ-NM (Q for Quad)
RBX11UAGS-5HPaxO-NM (O for Octa)

The D model just to have a basic device to 2x2 Omnis and small sectors. Also do a mANT model.
The Q model perfect for a 4x4 Omni and takes direct aim at the A5c+N360 setup that is kind of micro-pops.
The O model is that single super site kit for 4x 2x2 sectors, either 4x 90s or 4x 120's for some overlap, or for high density deployments with the fantastic RFE Asymmetric horns with overlaps doing 90-180 degree shots for ultra density.

IMO, this would put Mikrotik right up on top of the heap and move a LOT of units. No special sauce, just getting the AX driver into routeros (v7?).

Would be nice to know if any Mikrotik guys took this info....

If this would make it in 2020 I can wait and should blow my competition out of the sky by 2021.....

modsx · Thu Nov 19, 2020 10:19 am

I only use ARMs as clients and I don't see any problem.
I have throughput problem even with mibsbe in some cases, especially for client upload, or with NetMetal + mANT I only get 50Mbps throughput when with other vendors I get 200Mbps, but I think it's a mANT problem.

We had a similar problem. mANT good signals, poor throughput. We have removed all mANTs and do not plan to use them again. SXT SA (90 degrees) and QRT (10 degrees) are currently the best solutions (seemingly more concentrated directional energy). I hope MT will build a QRT with 30,45,60,90 degrees.

mistry7 · Thu Nov 19, 2020 11:02 pm

We had a similar problem. mANT good signals, poor throughput. We have removed all mANTs and do not plan to use them again. SXT SA (90 degrees) and QRT (10 degrees) are currently the best solutions (seemingly more concentrated directional energy). I hope MT will build a QRT with 30,45,60,90 degrees.

if you won’t to buy good Antennas, look for RF Element Horns, Radio is easy swap able, and you need no alignment after exchange, easy way to test mikrotik, mimosa, ubnt, cambium etc

modsx · Fri Nov 20, 2020 1:39 pm

if you won’t to buy good Antennas, look for RF Element Horns, Radio is easy swap able, and you need no alignment after exchange, easy way to test mikrotik, mimosa, ubnt, cambium etc

Goodl idea. What good antennas can you recommend?

mistry7 · Fri Nov 20, 2020 1:48 pm

depends on your needs.....

https://rfelements.com/products/

WirelessRudy · Thu Nov 26, 2020 6:34 pm

We had a similar problem. mANT good signals, poor throughput. We have removed all mANTs and do not plan to use them again. SXT SA (90 degrees) and QRT (10 degrees) are currently the best solutions (seemingly more concentrated directional energy). I hope MT will build a QRT with 30,45,60,90 degrees.
if you won’t to buy good Antennas, look for RF Element Horns, Radio is easy swap able, and you need no alignment after exchange, easy way to test mikrotik, mimosa, ubnt, cambium etc

You also need to swap client units if you'd really want to compare the different makes.....

WirelessRudy · Thu Nov 26, 2020 6:39 pm

I only use ARMs as clients and I don't see any problem.
I have throughput problem even with mibsbe in some cases, especially for client upload, or with NetMetal + mANT I only get 50Mbps throughput when with other vendors I get 200Mbps, but I think it's a mANT problem.
We had a similar problem. mANT good signals, poor throughput. We have removed all mANTs and do not plan to use them again. SXT SA (90 degrees) and QRT (10 degrees) are currently the best solutions (seemingly more concentrated directional energy). I hope MT will build a QRT with 30,45,60,90 degrees.

We use RF Horns for a while and they are good. The only problen now is we need to have an upgrade on the RB922 boards. They are too weak to serve a lot of clients that have 100Mbps plans.
We need the same board as the new Netmetal ac but in the 922 form factor size. This way we can swap many of our sectors for more powerfull (CPU) sectors and yet again beat the competition!

modsx · Thu Nov 26, 2020 6:47 pm

We use RF Horns for a while and they are good.

RF horn 90* vs SXT 90* who wins and why?

Thu Nov 26, 2020 9:01 pm

We use RF Horns for a while and they are good.
RF horn 90* vs SXT 90* who wins and why?

Cleaner pattern, same gain with different frequencies, low sidelobe, much better FB Ratio.
It is comparing a big speaker box to a smartphone speaker. Both make sound.

modsx · Thu Nov 26, 2020 9:49 pm

It is comparing a big speaker box to a smartphone speaker. Both make sound.

Hard to say...
SXT Maybe not so evenly, but the SXT will probably deliver more concentrated energetic.
I assume that the front to back ratio is not big for them but similar 30-35 (because horn high-frequency currents do not have time to pass into the electric field and flow to the external surface of the speaker and create side lobes, but if in case they have a ribbed inside, it helps a bit).
The side lobes will be both, but you can try to save bandwidth by putting it in reverse and using frequency reuse.

mistry7 · Thu Nov 26, 2020 10:49 pm

It is comparing a big speaker box to a smartphone speaker. Both make sound.
Hard to say

Did you ever use them?
We saw much better CCQ with horns!
We will never go back!

Thu Nov 26, 2020 10:58 pm

It is comparing a big speaker box to a smartphone speaker. Both make sound.
Hard to say
Did you ever use them?
We saw much better CCQ with horns!
We will never go back!

Yes. Same experience.

syadnom · Fri Nov 27, 2020 7:03 am

SXT and virtually all MikroTik antennas except parabolics are patch antennas, which are the cheapest and craziest antenna design.

Horns are wave guide with isolating layers.

The horn will put more signal towards your customer for the same dB rating because none is lost to sideloges and backside radiation. It will reject all off-axis signals.

There really is no comparison. The SXT’s patch has zero advantages.

modsx · Fri Nov 27, 2020 12:15 pm

It's not about cheap.
SXT works great.
CCQ on the client side is achieved with a professional installation, searching for a place (depending on the angle, Fresnel / Arago / Stand waves specific). Take MQS + Powerbank + ex.SXT and look for measurements.
For example, the old SXT.

mkx · Fri Nov 27, 2020 5:41 pm

The horn will put more signal towards your customer for the same dB rating because none is lost to sideloges and backside radiation. It will reject all off-axis signals.

The first part is not exactly true. Antenna gain (measured in dBi), to which you're referring as dB rating, is maximum signal level radiated in any direction compared to signal level of (theoretical) dipole antenna when driven by same radio Tx power.
But, as you noted, the maximum signal stregth is not the whole story, side-beams are important as well because they contribute to interference (in both directions). And main beam width and height of course, after all making main beam narrower is the only way of getting antenna gain higher (it's easy to make gain lower of course but usually nobody wants that).

Fri Nov 27, 2020 8:03 pm

The horn will put more signal towards your customer for the same dB rating because none is lost to sideloges and backside radiation. It will reject all off-axis signals.

The first part is not exactly true. Antenna gain (measured in dBi), to which you're referring as dB rating, is maximum signal level radiated in any direction compared to signal level of (theoretical) dipole antenna when driven by same radio Tx power.
But, as you noted, the maximum signal stregth is not the whole story, side-beams are important as well because they contribute to interference (in both directions). And main beam width and height of course, after all making main beam narrower is the only way of getting antenna gain higher (it's easy to make gain lower of course but usually nobody wants that).

Best is to look at the antenna pattern of both polarisations. This shows where the signal goes along. Some vendors do not show them so you know they are bad. There is also a graph of gain vs. frequency. No antenne performs the same for the whole band from 5 to 6GHz.

anuser · Sun Nov 29, 2020 11:51 am

SXT and virtually all MikroTik antennas except parabolics are patch antennas, which are the cheapest and craziest antenna design.

I recently seen some advertisement of a vendor saying that its new access point has beamforming inbuilt. Looking at those antennas, I actually cannot see anything special. Aren´t those aswell the same cheap antennas which can be found in MikroTik access points?

r00t · Sun Nov 29, 2020 6:25 pm

Beamforming is done by changing phase of the transmitted signal going into multiple antennas. That way it's possible to make all the transmitted copies of signal add exactly at the spot where client is receiving. This technology itself is antenna type agnostic. Patch antennas and arrays of phased patch antennas are very common because they are cheap to make and for higher bands they are also have a decent gain on a small footprint. But as said by other posters above, resulting beam is not as clean as from waveguide horn or other "clean" antennas like a helixes. Ugly sidelobes and often "fractured" main beams are common for arrays of patch antennas. That's why they aren't very good when used as a feed for a bigger dishes, but usually acceptable for CPE use where being small and cheap often wins.
But in some cases there is simply no other alternative. For example for 60GHz you have to have phased array on a very small PCB and patch antennas is the only way to do it, can't have an array of 64 waveguides. But also on these high bands the adverse effects of multiple elements are less pronounced.

Lazaro · Sun Dec 06, 2020 4:44 pm

Hello good day. I have with the 802.11 only ac a PaP link between disc ac and sxt sq ac signal -65 noise 107 working at 80 MHz all fine but at the moment it gives packet loss and 2000 pins this only lasts a second and returns to normal again someone could help me with that thanks

Who is online