We've got a bunch of APs deployed that use 433s and a panel antenna for the backhaul and two 120 degree 12 db sector antennas. Problem is we need 360 degrees of coverage, and a lot of the surrounding buildings have metals roofs.

I'm thinking of upgrading a couple of them to 435s and using the same panel antenna for the backhaul - but using four 90 degree 17 db sector antennas. That would get me more power and 360 degrees of coverage. I think we'd build a mast to separate the antennas by at least 4' and I would use a mix of half vertical and half horizontal polarization.

Now I know a lot of folks have mentioned all the interference between cards. So any thoughts would be appreciated.

And does anyone know of a decent enclosure for the 435s?

Thanks!

P.S. And I suppose an alternative take on this would be using two 433s instead of one 435 - but I really am trying to stick to the KISS principle.
Mac

Well...no answers or opinions from the peanut gallery. From most of my browsing of the forums it seems that installing a lot of R52 cards in close proximity - all on the 2.4 GHz band - would not be good. We're currently using 433s with a 2.4 on wlan1, 5 on wlan2 (backhaul) and 2.4 again on wlan3...but perhaps we should even reconsider that.

So my current thought is using Grooves or Metals. Bottom line we need to construct a wagon wheel of APs in a 360 degree arc with a backhaul. Need a lot of power...planning on using 90 degree 17 db sector antennas...4 of them. What are folks using these days?

Biggest issue I see with the Grooves and Metals is connecting all of them. Guess we can use a UBNT toughswitch (until Mikrotik makes a similar product...hint, hint) but as far as I can tell the toughswitch needs an AC adapter and won't run off PoE. Trying to avoid running anything more than an ethernet cable to the array.

-Mac

Mikrotik already has a product that suits your POE needs. Check out RB750UP.

http://routerboard.com/RB750UP

If u listen to me avoid using multiply wlans on 2.4ghz on same board (2 max)
Avoid using something like metal - its too much powerful, go for radio with less power and bigger antenna

Don't really know why Mikrotik and UBQ are making/selling this high power 2.4Ghz units (Metal, Titanium...) except if u need them to make a 100km link

Just because it has a 1W tx power doesn't mean you can't turn it down.

Most of my links these days use a high power card turned down. They are usually MMCX, usually higher quality and certainly easier for the climber to deal with.

This sounds like a good fit for the Metal, but I haven't tried it yet myself.

Metal 2.4Ghz is 1.6W and I don't want someone to use this near my towers
by the way where on earth is allowed to use 1.6W on 2.4Ghz? not adding antenna dBi

Technically, point to point in the US is as long as you have a crappy antenna, which makes no sense for their reason for a power limit in the first place...

However, as I said above, you don't have to turn it all the way up, as a matter of fact I recommend against it. The Groove has virtually no shielding, so any tower I might use the Groove will now get a Metal turned down to around 20 or 22dB.

A 1.6W amp should last quite a while driving around 1/16 of the maximum output.

Mikrotik already has a product that suits your POE needs. Check out RB750UP.

http://routerboard.com/RB750UP

Fair enough...but again it runs off a power adapter and not PoE itself, right? We can't run 120 or 240v lines up our towers - most of them are temporary.

Which would mean we would need to run five ethernet cables up to the tower...a lot of cable.

Most of our stuff, by the way, is carried in on our backs. Smaller and lighter is a better...to a point.

-Mac

Why don't you try some sulution like this as for a 360* antenna?

http://antarlangit.com/products/Antena- ... .4GHz.html
http://www.easylink.lv/en/products/117

If you search around a bit, you'll find a quite few of them.

I didn't quite understand what exactly you need to do, but consider a solution of ONE AP as "central" (repeater) with ONLY ONE omni high gain antenna.

rb750UP can be powered by PoE. However i would power it vie 48V with converted to 24V just before the router to reduce losses on the cable.

PoE converted is about 1.1A @24V, if that is fine, you are good to go.

Of course, powering RB750UP via plug would be more stable solution. You can lok around the forums to find how some users solved voltage drops over long cable powering RB750UP.

Thanks for the info on the 750UP being capable of being powered by PoE - and the 48v stepped down to 24v idea is good. I do need to do some research on the amperage of the Grooves and Metals and make sure I can power 4 or 5 of them with only 1.1 A - plus the 750UP.

Forgive my ignorance on dbi and antennas...but isn't a 360 degree 17 dbi antenna going to only provide a quarter of the power 4 90 degree 17 dbi antennas would provide? I completely agree than a single antenna single RB dual radio - backhaul/ap approach would be ideal - but we're talking about metal roofs and concrete construction and we're trying to avoid wiring every building individually.

-Mac

With an Omni 360*, you will get 17dBi in a "pure" circle transmission (antenna dependable) - This means that you will transmit your power (1600mW) at 360* with 17dBi gain.

With 4 x 90* panels joined them together to get a 360* transmission, you will have:

a. 4 x money that you will spend to do the EXACT SAME THING with only ONE AP and ONE Omni antenna

b. 90* on EACH direction with – let’s say 1600mW – So OVERAL you will have 1600mW on 90* x 4 = 1600mW at 360*.

c. NOT a "real 360*" transmission because of the interferences which will encode EACH antenna's side lobes, right and left to it so it will NOT be a "real 360*" transmission. PLUS that you will have "black holes" at the points when two 90* panels "see" each other.

d. You think that you need POWER.
I agree with you, BUT.......
BUT afterwards you MUST provide MORE power to ALL of your clients (if this will be an AP). Why? Simply because IF you leave your clients as they are, you will encode the following:

When you have power - let's say 500 Watts - on your AP, the client will "sees" it and shows you a signal strength - let's say - -50dB.
Then the client transmits its power to "talk" to the AP and say to it: "I received you, do you hear me?"
Then the AP responds to it: "I hear you BUT not load/clear", because the signal received by the AP is VERY low considered to the receiving signal of the client (because of its lower output power).
Then AP "locks" its data throughput at the MINIMUM signal it reads after it has a “conversation” and an “agreement” with the client at what Mbps will lock.
SO, EVEN IF you have a LOT of power on AP, you get NOTHING if the clients can’t “reach” the AP with a “workable” signal.

George.

With an Omni 360*, you will get 17dBi in a "pure" circle transmission (antenna dependable) - This means that you will transmit your power (1600mW) at 360* with 17dBi gain. With 4 x 90* panels joined them together to get a 360* transmission, you will have:

a. 4 x money that you will spend to do the EXACT SAME THING with only ONE AP and ONE Omni antenna

With a single omni antenna, one radio will receive all client signals--and noise--from 360*. With four 90* sectors, each radio will receive 1/4 of the client signals and noise.

Whether that outweighs interference from having multiple radios and antennas operating near each other, depends on the expected number of clients, and the existence of any localized noise sources (that might be reduced by the use of directional antennas).

b. 90* on EACH direction with – let’s say 1600mW – So OVERAL you will have 1600mW on 90* x 4 = 1600mW at 360*.

Actually, the Tx advantage is 1600mW / 90* vs. 1600mW / 360*.

And there is a similar benefit on the Rx side, because a 17dBi sector has roughly 4x the vertical coverage of a 17dBi omni. Four 90* sectors at 17dBi will be able to effectively amplify return signals that are coming from within an area four times as large as that of a single 17dBi omni.

With an Omni 360*, you will get 17dBi in a "pure" circle transmission (antenna dependable) - This means that you will transmit your power (1600mW) at 360* with 17dBi gain. With 4 x 90* panels joined them together to get a 360* transmission, you will have:

a. 4 x money that you will spend to do the EXACT SAME THING with only ONE AP and ONE Omni antenna

1. With a single omni antenna, one radio will receive all client signals--and noise--from 360*. With four 90* sectors, each radio will receive 1/4 of the client signals and noise.

Whether that outweighs interference from having multiple radios and antennas operating near each other, depends on the expected number of clients, and the existence of any localized noise sources (that might be reduced by the use of directional antennas).

b. 90* on EACH direction with – let’s say 1600mW – So OVERAL you will have 1600mW on 90* x 4 = 1600mW at 360*.

2. Actually, the Tx advantage is 1600mW / 90* vs. 1600mW / 360*.

And there is a similar benefit on the Rx side, because a 17dBi sector has roughly 4x the vertical coverage of a 17dBi omni. Four 90* sectors at 17dBi will be able to effectively amplify return signals that are coming from within an area four times as large as that of a single 17dBi omni.

1. If he hasn't THAT many cliens to "overload" the AP, he will be OK.

2. When the 4 x Tx powers with 4 x Panels translated at 360* coveradge, HOW many watts will be transmited at 360*??
For your second paragragh, all I have to say is: re-think about it.

I've got a tower with 3 17dBi sectors and 1 15dBi omni as a backup.

There is little difference between the signal of the sectors and the omni, two customers are better on the sectors, one customer is much better on the omni, 15+ don't change.

There is little difference aside from the capacity issue in this situation.

One thing that is hugely important is with the sectors I can prevent the sectors from interfering with other towers easier than with omni.

With an Omni 360*, you will get 17dBi in a "pure" circle transmission (antenna dependable) - This means that you will transmit your power (1600mW) at 360* with 17dBi gain. With 4 x 90* panels joined them together to get a 360* transmission, you will have:
a. 4 x money that you will spend to do the EXACT SAME THING with only ONE AP and ONE Omni antenna

With a single omni antenna, one radio will receive all client signals--and noise--from 360*. With four 90* sectors, each radio will receive 1/4 of the client signals and noise. Whether that outweighs interference from having multiple radios and antennas operating near each other, depends on the expected number of clients, and the existence of any localized noise sources (that might be reduced by the use of directional antennas).

If he hasn't THAT many cliens to "overload" the AP, he will be OK.

Cross-talk from clients increases--and performance decreases--rapidly with the number of clients on one radio. If you have a tower serving 120 rural clients within a one-mile radius, I guarantee you that throughput will be better using four radios with an average of 30 clients each, than with one radio serving all 120 clients, during high-demand periods.
As for noise: if there is a noise source in one direction, it will be attenuated for the directional antennas that are pointed away from it. With an omnidirectional antenna, that noise will degrade all signals.

b. 90* on EACH direction with – let’s say 1600mW – So OVERAL you will have 1600mW on 90* x 4 = 1600mW at 360*.

Actually, the Tx advantage is 1600mW / 90* vs. 1600mW / 360*. And there is a similar benefit on the Rx side, because a 17dBi sector has roughly 4x the vertical coverage of a 17dBi omni. Four 90* sectors at 17dBi will be able to effectively amplify return signals that are coming from within an area four times as large as that of a single 17dBi omni.

When the 4 x Tx powers with 4 x Panels translated at 360* coveradge, HOW many watts will be transmited at 360*??

If you are transmitting 1600mW over each 90* arc, you are transmitting 6400mW over the full 360* arc. Since the gain of each antenna is the same, the advantage is not in the energy received at any given point, but in the number of points at which that signal level is received--because a 17dBi 90* sector has 1/4 the horizontal axis of a 17dBi omni, but 4x the vertical axis. Or, if you want the same narrow coverage pattern as the 17dBi omni, you can use four 23dBi sectors. Now 4x the energy will arrive at any given point within the identical coverage area.

For your second paragragh, all I have to say is: re-think about it.

Bottom line: there are good reasons that towers designed for wide p2mp coverage use multiple radios and sectors, vs. single radios and omnis--and it has nothing to do with spending money unnecessarily.

So back to the original question then - since it seems that four radios/sectors and one radio/panel for the backhaul is the way to go.

Use a 435G and five radio cards or use five Grooves/Metals and a 750UP with a 48v PoE injector stepped down to 24v?

-Mac

@ Hotz1
Not to post a lot, the only thing I aggree with you is ONLY IF the AP has a lot of clients to serve (thing that we don't know since the starter of this thread hasn't told us). In all other cases, ONE AP is better.... :p

Lots and lots and lots of clients. Barracks chock full of people. Why I am leaning towards the sectors. I could imagine as many as 100 clients (or more) connecting to each tower...and breaking that up into smaller chunks would be ideal.

-Mac

Go with the sectors then and more APs....

Totally agree. Lots of radios. Sectors. Check.

So again - faced with using a single 435G and five radio cards all causing potentially disastrous crosstalk and interference because of their close proximity on the RB but allowing me to power the device with a single ethernet run and PoE injector or using five Grooves or Metals and a 750 UP with a 48v stepped down to 24v to provide needed amps.

That's the dilemma. And the question.

-Mac

What I've found, playing with this exact problem recently, if the TX is pretty low (below 20dBm) then self interference isn't as much of an issue. You can have higher TX if you use MMCX cards and good pigtails, but I still recommend going as low as you can on the TX power as possible that works.

I have to make a decision on this soon myself. I had a 433 w/3 2.4GHz sectors start to die, now a 433 to replace it would be the cheapest route, just swap the board, but I could throw together a new AP with better cards, pigtails, etc, or go separate with 3 Metals.

As it is working fine as long as the tx it turned pretty low I may just go the cheap route and spend my $$ on hooking up new customers.

Lots and lots and lots of clients. Barracks chock full of people. Why I am leaning towards the sectors. I could imagine as many as 100 clients (or more) connecting to each tower...and breaking that up into smaller chunks would be ideal.

I hate to say it after the foregoing (mostly theoretical) debate about omnis & sectors, but if:

...we're talking about metal roofs and concrete construction and we're trying to avoid wiring every building individually...

then you'll probably get much better results putting an inexpensive CPE setup in each building--client antenna on the inside, backhaul antenna on the outside--and one central backhaul radio.

I have some experience with metal roofs and concrete construction. From a distant, central location, you can punch through metal and concrete--but what you get back from a room full of underpowered clients, after all the reflections off the walls and roof, is going to "sound" like mud. With high-attenuation construction, you really want the client antenna to be either inside the building, or located just outside a window or other low-attentuation surface where the signal can reach as much of the interior as possible.

If you can find such a location outside the building that also has clean LOS to the central backhaul tower, you can put an RB433 inside a dual-polarity 5GHz panel enclosure for backhaul, hang an omni or two off it for the client radio, and that takes care of that building and its immediate environment--which may include other buildings, depending on layout; e.g., three or four buildings with large windows facing a central compound. Be creative. Otherwise, put an SXT outside the building, a RB951 inside the building, and connect them with Cat5 /PoE.

For the central backhaul radio, since it will only be receiving from one radio for each building (all of which support protocols that minimize cross-talk), use an OmniTik.

I completely agree with you about wiring each building individually. We run into a couple of big stumbling blocks with that approach...namely we don't have a tech on site and since electrical outlets are in such high demand a lot of time our equipment gets accidentally unplugged, the PoE injectors go missing or some other mischief...usually involving competitors who do have onsite staff.

My favorite way to wire individual buildings so far have been SXT - and dare I say this on a Mikrotik Forum - UBNT Picostation 2HPs - combos. Toss the SXT on the apex of the roof and aim it at the BH antennas, toss the Picostation inside. I would love to use Grooves but they have less power and no antenna and cost more.

So this post was mainly an attempt to mangle our way through a less than ideal situation.

Any comments on the 435Gs handling 100 plus client sessions simultaneously?

-Mac