A5-18 End of Sales Announcement


#1

Mimosa has officially announced the End of Sale of the A5-18 and A5-18-ETSI, effective immediately.

Here is a link to the official statement and maintenance/support timelines: A5-18 EOS

Alternate access points are the A5-14 or the A5c. For shorter range MicroPOP deployments, Mimosa recommends the A5-14. For longer range tower style deployments we recommend the A5c with sector antennas. For the A5c, you can connect two 2x2 sector antennas, each with a different physical coverage area. See the FAQ for more details on this.

I’ve put this FAQ together to help answer your questions:

Why did Mimosa discontinue the A5-18 access point?

The A5-18 was designed as a higher gain MicroPOP access point for short to mid-range deployment needs. Several factors have led Mimosa to the decision to stop selling the product:

  • The A5-18’s size, gain and down-tilt led to improper product positioning as a multi-sector access point replacement for a tower deployment.
  • The A5-14’s small size and broader antenna elevation pattern outperforms the A5-18’s narrower elevation pattern for typical MicroPOP height deployments. Most MicroPOP deployments are installed on single or two story homes ranging from 20 to 30 feet.
  • The A5c is a significantly higher performance solution for multi-sector tower deployments with its +3 dB beamforming gain when using a 4x4 sector antenna.
  • An A5c can be used with two non-overlapping 2x2 sector antennas providing a cost-effective dual sector solution. Combining multiple A5c access points, each with dual 2x2 sector antennas creates an extremely cost effective 360º tower solution when combined with Mimosa’s GPS-Sync collocation technology.

What if I have an A5-18 on backorder?

Mimosa is not selling the A5-18 anymore. Customers with A5-18 products on backorder should have been contacted by Mimosa or their distributor/reseller to discuss alternative access point options from Mimosa. If you have not been contacted yet, please contact your distributor/reseller. Mimosa recommends one of the following access points as a suitable replacement for the A5-18. Either the A5-14 for shorter range MicroPOP deployments, or the A5c for longer range tower deployments with a sector antenna.

What if I have an A5-18 deployed in my network and need a replacement?

For A5-18 access points that already deployed, Mimosa stands behind the standard product warranty of 1 year from date of unlock or 3 years from date of manufacture. (http://www.mimosa.co/warranty.html) Mimosa has a reasonable RMA stock of A5-18 radios to handle standard warranty servicing.

Can I exchange my new or used A5-18 for an A5-14 or A5c?

Mimosa does not have a product return policy. Mimosa stands behind its standard product warranty policy for the A5-18. (http://www.mimosa.co/warranty.html)

Please feel free to comment and continue the conversation.


#2

Well, we just draw a plan for some development where the A5-18 would have come in use…:frowning:
We need to server an area of a circle of some 200mtrs radius with some 40-50 clients.
The center of this area is also on a cliff type of drop. By tilting this A5-18 4 degrees down towards the lower end we were able to serve some 20 clients in the low part where the other 30 clients would be served byt the -4+4=0 beam.
This could be done by a A5-18 for just under 1000€. Now we have to consider 2x A5c + 4 90º sectors?
This is going to cost 1550€… Thats an 55% increase!

Plus this is way too much capacity for these 50 or so clients.
Plus this still need now 2 frequencies since GPS sync is still not available…

You guys seem to think that where Mimosa is beeing deployed (micro pop high density) provider is the only one.
But in many regions provider has to compete with other and thus not so many clients are to be achieved, even in a high density development, and even againt hight througputs for the client. To make a tower with 2 x A5c with their 2x2Mimo sectors is way over the top in pricing and capacity to deal with only 10-50 clients…
Here a single 4x4 omni would serve and sometimes some center building is that much higher then the rest the A5-18 is still preferred.

You should also not forget that many providers considering Mimosa come from another 3rd party background so probably start with ‘interop’ mode anyway. This however prevents the GPS sync and then a 4x4 Omni is again preferred above 2xA5c’s for the obvious reason of spectrum re-use. Many work in crowded spectrum so the less we can use the better…

I got the feeling Mimosa is aiming at new network development more then on migrating. Imho the big bugs are to be found in the latter. Not in new environments… after all, how many of these are out there waiting somebody finally is giving them the not yet existing internet and just have it coming with 200Mbps from day one?

Now a A5-18 without downtilt would be THE ideal unit in my opinion for migrating WISPs all around the world.
If the A5-18 downtilt would not server your business model, a A5-18 WITHOUT a downtilt would be!


#3

A5-18 without downtilt is the A5-14 that hasn’t been discontinued.


#4

Not completely. The “18” is for the gain, as is the “14”.
As you’d now 3dB is doubling (or halving) the signal strength. Hence the A5-15 would throw 1,15% more power out.
More important, it would ‘listen’ to its associated clients with 1,15% more sensitivity.

Meaning the ‘cel’ range of the A5-18 could be also be bigger in use with the same type of CPE clients.

For micro o small sized cells this is not so much of importance. Here we’d look for 2x2 conectivity anyway.
But due to its high cpu power and advanced chipset and in combination with its high gain (18dBi) this AP-18 could easy compete with many other Ubnt and similar Omni AP setups…

Combined with the future gps-tdma and C5’s even small to medium cell network deployment would be outdate many present omni directional setups…

And many times in countries/regions with hills and mountains indeed a omni serving not too many clients can still sit high so a downtilt is a pro.

I can really see the point in dropping the A5-18 if Mimosa’s strategy is only to go to war with micro cell alternatives and fibre/cable companies.
But imho the scale of numbers in deployments are probably more to be found in medium (and wide) range WISP networks. Like us!

I am a typical operator in an Europen country with mountainous regions etc. and we deploy many duo-pol omni’s in networks only serving 10-20Mb max to the client.
With a Mimosa AP, so also a A5-14 AND a A5-18 on a high tower or hill, we can serve same clients with 50-100Mbps and that would make us a winner!

Let’s hope Mimosa not gambles on the wrong market segment… :wink:


#5

@Rudy - Thanks for the feedback. Just to be clear, you can use an A5c with two dual-pole antennas as illustrated in this diagram. Obviously coverage and gain will determine the reach of the coverage. From the access point standpoint, it doesn’t matter if the antennas are V/H polarized or ±45 slant, we just want to see two different polarizations on each antenna as I’ve indicted in the diagram below.

When using a single A5c, the non-overlapping pattern concept is not that big of an issue as both sector panels are connected to the same radio and using the same channel. You even get some beam-forming advantage in the areas of coverage overlap and you will get higher throughput because of the 2 streams per antenna vs. an A5-18.


#6

Hi David,

I have to agree the setup with an A5c and two 180º sectors didn’t cross my mind. Probably since up to today I have never seen one duo polarised 180º sector. And probably they will have a narrow V-beam.
This latter will make it problematic to use on a high point since the narrow V-sector will limit its pattern reach. (Either only close range or distant)

You’d refer to a bit of beam-forming in the non-overlapping region. But is beam-forming in the A5’s already embedded in the firmware? I thought this is still something we are waiting for?
Also, most sector antenna’s will have some side lobs even outside there working range. So probably even in the non-overlapping sector a high gain CPE could still connect to the A5c.

Actually, I think antennas with some side lobs are now preferred above some highly reduced side-lobe antennas since we don’t have to worry about ‘auto-tower interference’ since the A5c works in one frequency…

So its basically up to manufacturers to develop some ‘dirty’ (=no side-lobe suppression) 180º sectors that have reasonable gain compared with some ‘open’ Vertical beam…


#7

Rudy,

I have one that will be replaced soon with an A5-14. The house is just not tall enough for my application. Coverage area is not broad as I would like. Let me know if you are interested.

Harold


#8

One other point on hooking up two 2x2 sector antennas is operating in the same channel. Using two 2x2 90º sectors connected to one A5c, for example, will provide 180º of coverage on one channel.


#9

What is a source for 90 degree compatible sectors?


#10

@Paul2 MTI Wireless Edge makes a 4x4 90º sector. MT-463042/ND Double Dual Slant, 4.9-5.9 GHz, 14 dBi 90º

It should be available now with our distributors. If you can’t find it let me know.


#11

So can two ubnt 120 19 db antennas be arranged to cover 180 degrees and would have 60 degree overlap where you get better coversge


#12

Along the same lines could two ubnt 120 19 or 16 Db antennas be arranged to cover the same area and separated by 6 or more feet to give the advantages of beam forming to CPEs. After all, the whole advantage of beam forming is to cut down the affects of noise between the AP and CPE. Beam forming was supposed to be the great advantage of Mimosa advanced processing power and antenna design.

I am also concerned on the distance scenario of the A5c. We overlook valleys from the tops of mountains. Because of the rolling hills in the valley and low density of houses, we broadcast from the tops of mountains on each side of the valley so pointing often at each other. So we can self interfere if we do not choose frequencies carefully.

So I need more info on proper use of Mimosa in this scenario. I have held up for two years on implementation waiting for the TDD and GPS capabilities. We have been told this is coming in a couple months every couple months, but it is not here yet.

I understand that this is similar to AC technology which has distance limitations. Is this a reason for discontinuing the A5-18 or can the Mimosa technology operate over longer distances?

Most of the information posted by WISPs is that they are getting great results in the WIFI mode but we never get a real feeling of the implementation. For how does it perform with high number of CPEs (30 or more)? What if the distance between CPEs goes from 4 miles to 10 off a single A5. I tried it with an A5-18 and there seemed to be some real limitations that I could not test as it seems some customers would not reconnect and we cannot take down a live network very long.

I was testing an A5-14 in a place where I have a lot of customers within a short distance and same elevation as A5… But I needed to upgrade the backhaul with a B5. In the process, we visually aligned the B5s and I was changing the cabling, when I fell of the roof shattering my elbow. So now after 2.5 months of recovery later, we are getting around to testing in wifi mode. We have moved 27 customers to it. We will see how they perform this Christmas weekend. We will othervtesting over New Years so stay tuned.

I should note that all customers are using ubiquity M5 radios. They are all 22 or 25 db radios and a heavy mix of older nano bridges versus new nano beams.

When and if Mimosa releases the TDD and GPS software, we will upgrade all of these to C5s with G2s in wifi mode and then switch all at once to the new Mimosa software.


#13

@eden: I would not use any narrow vertical beam sectors from any brand to cover a valley from a high point. Due the narrow vertical beam the footprint on the bottom of the valley is only a small distant ring section from AP that gets best signal from these sectors. Special the ubnt’s are not a good sector for high towers that have to face down and cover big ranges. I replaced them already years ago for better ones.
Better in that case is the RF elements cone range of antennas but I also use their carrier class sectors. They are soo much better then ubnt’s sectors…

Beam forming, tdma and GPS sync is still not there from Mimosa, that’s a real setback if you’d ask me and should start to give them the shivers since after almost 2 years of promises and delays yet still not available…

In the meantime I use one A5-14 with 53 (!) associated clients, all Mikrotik, all at distances from 50 meters up to 400 meters max and the performance is great! Clients are allowed to use up to 25Mbps downloads and we see peak usage of the AP of 50-60Mb at times. We have 2 SXT’s and one Mimosa C5 connected and at any time I can start downloads on these and they all get 40-50Mbps at the same time! (Meaning total througput on the A5 reaches 200Mb!) All ‘n’ type of SXT’s! I can easy push to their wire speed…
On a single mikrotik CPE (we used a SEXTANT for the gigabit port) we could push a download to 150-200Mbps relative easy while the rest of the network still worked. The C5 we could push even a bit higher… I saw a peak of 225Mbps…

This all in a ‘interop’ mode (=802.11 legacy) and apart from 1 SXT and the C5 all CPE’s where ‘n’ models only.
We used 40Mhz channel. We tried 80Mhz for the test which gave the high througput on the Mimosa C5 but since none of the ‘n’ models do support that and we needed to conserve some spectrum the normal working bandwidth is 40Mhz. And that works fine…

So when it comes to 802.11a© legacy mode this is the best performing AP I have seen for a long time. I am very happy and since none of the connected customers complained (that has been different in the past!) I’d presume they are happy too!

The only negative I noticed is that it seems sometimes the network sort of collapses when some antennas seem to consuming lots of bandwidth. I’d presume this is because the ‘legacy’ mode since with so many associated stations even the RTS/CTS CSMA helper tool can’t stop collisions in the network which brings the capacity down.
But mind you, I only saw this when I was pushing the A5 to deliver 100Mbps to some stations at the same time. Under normal conditions as said I never see the network collapse.

The use of M5’s in comparison to SXT’s is both a pro as a con. The pro is that the dishes are better in isolating from unwanted noise. And for bigger distance you still can get relative high signal levels. (An SXT is only 16dbi).
But in a CSMA network the directional beam is in fact a disadvantage. Since the probability of a ‘hidden node’ issue is even bigger when all clients stations are heavily focussed on the AP only. The simply won’t ‘hear’ eachother anymore. In tdma this is not an issue, in CSMA it is…

Also their signal gain is too high for short distant associations to the Mimosa. It could be that both on the CPE’s end as on the A5’s end the signals are too high. Signal from CPE to A5 can be tuned down but from A5 to CPE it all depends on the weakest clients you’d want to reach.

I tuned all client stations output power such that their signals ‘hit’ the A5 in the region of -52/-58dB. First we tuned them down even more but then the stations couldn’t ‘hear’ all the others anymore and the ‘hidden node effect’ made the network complete degrade to almost useless. I found this level for a cell of about 400mtrs works good.

We also set all CPE’s (well, on the Mimosa C5 its not possible!) to use RTS/CTS at all times. Without this the network is crap… (Mind that most CPE’s, Mikrotik, Ubnt, etc by default have a setting not to use RTS/CTS. That is wrong. It should always work, under any circumstances.)

So, this is all in ‘interop’ mode, or “migrating network” mode.
To prepare for the future GPS sync with tdma we need to replace all stations too into C5’s. But due the costs and the fact its still not available we developed another plan;
Before converting all our 55 clients that work with the Mimosa into C5’s we are thinking to see what would happen if we’d replaced all SXT ‘n’ protocol antennas for ‘ac’ units. Its a lot of work but if it would improve the network a bit further then we already achieved (we then could use 80Mhz bandwidth and the higher MCS rates) I might decide not to spend my money on a whole bunch new C5’s but buy more Mimosa AP’s for the rest of my network.
After all, for the price of one C5 I can buy almost 3 SXT ac’s…
On this moment it doesn’t make a lot of sense to spend 3 times as much on C5’s only to squeeze the extra 20% performance gain out of the network once the new promised features finally comes available.
And then again, who is waiting for 200-300Mb clients connections anyway where 20-30Mb is already considered ‘top speed’??

Imho only new deployments that have to compete in dense areas with fibre should go for the full A5+C5’s PtMP package.
For migrating operators just an Mimosa AP in legacy mode already doubles or quadruples your network against relative low investment… (At least, with MT that is the case. I read somewhere that the latest ‘ac’ hardware of ubnt is not backward compatible anymore so no use in a Mimosa network…)

The only other reason to go for a full Mimosa network is when may AP’s have to be sync’d to conserve spectrum and that needs tdma and thus Mimosa CPE’s too… Is a balance between top performance and money available…