I'm having problems with the link shown on the attached .jpg and would welcome any advice.

The link is 2.79 miles (4.5km) over water all the way. The southern end is about 15m above the water, the northern end around 5m. The Southern end is running ap-bridge with a Senao EMP-8602+S putting out 17dB at 2472GHz, band B, into a 12dB mini-directional with a 35" beam, horizontal polarity, pretty much centered on the northern end.

The northern end is also a Senao EMP-8602+S, on a 24dB grid antenna (also horizontal polarity, of course).

At night the link is stable with tx/rx readings in the -75/-80 range but during the day, particularly hot sunny ones, the link falls to pieces with tx/rx readings that fall through the floor.

I'm guessing it's interference from the water, or a heat-haze rising from it, and suspect the only solution is going to be to gain some more height at the northern end but that isn't going to be easy, and any other thoughts as to stabilizing the signal would be appreciated.

The RB at both ends is a 433, and as the link is primarily to an adsl gateway it could operate at only 5MBps and still be acceptable. A switch to 5GHz would involve changing the CPEs for a couple of other clients on the northern end as well, but would be justified if it would make a difference. In both cases they are using 15dB rootennas with no problems, but in both cases are around 30m above the water.

Try changing the antenna polarity to VP. I found sometimes that helps.
I use, as a rule-of-thumb, 10 feet antenna height for each mile on both ends. Higher on one end can compensate for lack of height on the other, but it is not linear. It will require more than just subtracting/adding the same amount. It is the fresnel effect that really causes the problem.

you should try to model this in Radio mobile and see if the fresnel zone is not in the water

Oh, yes. I did forget one thing. Being a surfer, I know this because I see it before I head out for the beach with my surfboard. When the surf is up, the radio reception across the water is worse.

ADD: Afternoon sun sparkling off a sea breeze ripple will kill a weak signal.

My suggestion is to replace that 12db antenna with a 24db, like the other end. Unlike transmitter gain, antenna gain is bi-directional. Also decreases the size of your fresnel area.

Thanks Normis. I'll download the Radio mobile thing and learn to fly it, tho' it looks a bit hairy.

The mid-point Fresnel zone radius on this link is 12m so it will be pretty close to the water if not touching it. According to MT's link possibility calculator I should be getting tx/rx readings in the mid -60's with this gear, but -76/-85 is the best I've seen and by lunchtime on a sunny day the link is getting noticeably unstable with tx CCQ readings (reception at the southern end on the 12dB mini) dropping into single figures.

Hi Surfer Tim. No swell here, I'm afraid. Just a nasty chop when the wind gets up.

I'd read somewhere, maybe elsewhere in this forum, that horizontal polarity worked better over water which was why I adopted it. Trying Vp is an option, but the 24dB directional didn't get a hint of an 8dB omni at the southern end when I have tried it vertically.

If it is just the Fresnel zone hitting the water switching to 5GHz would help as that would take the Fresnel zone down to 8m from 12, but it would be an expensive option as there'd be five sets of gear to change. Might just aiming the 24dB directional a few degrees higher help?

Changing the antenna at the southern end isn't an option - we need the spread of the 12dB mini to connect a few other clients in the same vicinity of the northern client. Even the beam spread of a 16dB flat panel wouldn't suffice. We could possibly go to a 17dB sector, but they're pricey items, too, and I'm not sure even another 5dB is going to help.

I went for B instead of G in this situation again hoping it would help. Anyone think it does? Are there any other tricks I might try before hauling the equipment up a nearbytree, like having/not having nstream, turning adaptive noise immunity off/on, or generally tinkering with bits of the wireless configuration I usually leave well alone (what is WMM support, anyway?)?

For the outdoor environment, as our experience, you will facing a lot of problem if running on the G.

and for the 5KM, you are advice to setup the point to point link with 24dbi gain antenna on both end with vertical polarization.

The horizontal only suitable for the town area. you may have a try.
And, if an AP service point to multipoint, the transmit power will be reduce, 1 ap serve 2 cpe power may reduce 30% or more.

please be careful on the RF planning.

Dump the 8602's and use either a SR5/XR5/R5H - 5ghz has a much tighter (smaller) fresnal than 2.4 and you should have better luck. If you insist on using 2.4 try using the SR2 or XR2 - 8602's are junk.

I know this would require a major change, but have you thought of going with circular polarization. I have not tried it, but have heard this may be an answer over water. Multipath is minimized.

If you insist on using 2.4 try using the SR2 or XR2 - 8602's are junk.

I'll ultimately do whatever is necessary but tho' it's the Subscriber's money I don't want to spend any more than is necessary.

I have been harbouring dark suspicions about the 8602 as I've had to pull a couple out of working installations because they seemed to be playing up, yet they were as meek as lambs back here on the test-bench. But I will give a change of card a try, thanks.

And I'll try vertical polarisation as that seems to be becoming a consensus.

Making this a point-to-point would mean upgrading from a 433 to a 600 as there are already 3 cards in the box, and an upgrade of antenna support too as what's there probably wouldn't carry a 24dB parabolic in a gale, along with everything else! And with three antenna's already on the pole I'm running out of 2.4 frequencies, tho' fortunately I have 13 at my disposal.

Angbootiong's comment about multipoints reducing the transmit power of the AP is a puzzler. How would that work?

I don't see a problem with wind loading with the right grid antenna. A 19db grid parabolic has:
100mph = 20 lbs
130mph = 31 lbs
and over 4 times the gain of the 12db.
reference: http://www.l-com.com/item.aspx?id=21706
And I have the same wind challenge. I use this formula:
Florida + Summer = Hurricanes

If this is a point-to-multipoint setup, it would be illegal here in the U.S. Too much EIRP on the 24db end.

I just wonder know, in the US, is there somebody will really check on the every wireless setup?

You are correct. The Federal Communications Commission does not check every transmitter. That is why I did not mention the 12db side. The minor infractions will probably go unnoticed. But a major infraction would probably be noticed, but not immediately. It would need to interfere with someone else, and they complain. But considering that almost half of my antennas directly face Eglin Air Force Base, if I messed up their stuff, that would be very costly!

You mentioned you had 13 frequencies. You actually have 3, maybe 4 that are not overlapping each other.
Why not use a 5 gig frequency for the link? Smaller antennas too. Consider circular polarization for the link.

All your suggestions are valid and useful, I'm just not 100% sure the problem lies with the effect of the water. In fact I'm head-bangingly baffled.

Earlier this am I was logged on to the AP watching the registration table, and this particular client was appearing on the table for a few seconds and then disappearing, appearing and disappearing. According to the 'status' page of the ap interface this client was registered but not authenticated - I have WPA2 encryption on the link. Now, two hours later with me having done nothing, the link is up and as solid as a rock at a tx/rx -76/-85 at the client end.

Moreover at the time the link was having problems earlier I noticed that an 8 dB omni at the AP end was picking up an 8dB omni at the client end - both are holiday lodges running a hotspot for clients. Admittedly the signal strength was only -89, but nevertheless two 8dB omnis (vertical p) with low-powered cards were hearing each other when a 12dB<>24dB link with more powerful cards (horizontal p) weren't.

I can't think polarity could make that much difference, could it?

If this is a point-to-multipoint setup, it would be illegal here in the U.S. Too much EIRP on the 24db end.

It's likely illegal here, too, but we're far from the madding crowd and have the aether to ourselves while hopefully the authorities are too busy sorting out the squabbles over access to the airwaves in the cities to worry about us.

You probably need to get further back from the shore for the link endpoints. When radios are near the water, tidal reflection are serious. Move both ends back 1/4 mile from the shore, and things will not fluctuate as much. Go for all the antenna gain you can get as well.

see the attached file for my home's 2.4ghz radio link. One end is only 50 yards from the water (solar powered small island repeater), so the closeness to the water causes the signal fluctuation. 5ghz is worse. The higher the frequency, the worse the tidal influence. 5ghz would not work for my situation.

You probably need to get further back from the shore for the link endpoints. When radios are near the water, tidal reflection are serious. Move both ends back 1/4 mile from the shore, and things will not fluctuate as much. Go for all the antenna gain you can get as well.

see the attached file for my home's 2.4ghz radio link. One end is only 50 yards from the water (solar powered small island repeater), so the closeness to the water causes the signal fluctuation. 5ghz is worse. The higher the frequency, the worse the tidal influence. 5ghz would not work for my situation.

Pic of the northern end of the link attached. It's only 60 yards from the water's edge but does get some shelter from the roof - the 'horizon' of the roof is probably several hundred yards beyond the shore. But there again might the roof - a steel one - be a problem in its own right?

Shifting the end back to the house upper left is possible and give a few more meters of height but the hotspot has to stay where it is to be able to hear tx from guest's laptops through the roof, so that would mean another set of gear and a radio or ethernet link between the two - which I don't want to do if it ain't needed.

Here's what it logged tonight:

21:33:05 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:33:07 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:40:23 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:40:25 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:40:46 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:40:47 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:41:02 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:41:03 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:41:13 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:41:14 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:41:54 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:41:56 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:42:17 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:42:18 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:42:28 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:42:30 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:42:45 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:42:46 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:42:50 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:42:51 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:43:10 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:43:12 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:43:22 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:43:24 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:44:28 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:44:30 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside
21:44:36 wireless,info 00:02:6F:52:DB:CD@wlan1: lost connection, extensive data loss
21:44:38 wireless,info 00:02:6F:52:DB:CD@wlan1 established connection on 2472, SSID Otherside

But then it was up for nearly 10 minutes without a break. tx/rx at the client end -84/-76

I'm beginning to suspect an electrical problem either on the board or the card. Wouldn't water interference be somehow more consistent?

-84db is not good. The Senao card needs about -75db for a 54m connection.
Can you switch to b-mode only for a test? That gives you better sensitivity.

I still recommend increasing the antenna gain on the 12db side. That will significantly reduce the effects of most of this, roof and water included.

ADD: If you need a visual representation of this, get a camera with a zoom lens. Get on the side of the water that faces the sun. Zoom out until the sun and water glare is in the picture. See what happens to the picture quality? Now zoom in until the sun and most of the glare is no longer in the picture. See how much better the picture is? Same with wireless, just a lower frequency. Wave theory is wave theory, even in Malaysia!

BTW, great looking place. You wouldn't provide service to any surf resorts, would you?

-84db is not good. The Senao card needs about -75db for a 54m connection.
Can you switch to b-mode only for a test? That gives you better sensitivity.

We've been using b-mode from the outset and we only really need 5MBps transfer as this is primarily a link to an adsl outlet for internet. The 8602-plus-S is supposed to be able to deliver 11MBps at -92 on b.

I still recommend increasing the antenna gain on the 12db side. That will significantly reduce the effects of most of this, roof and water included.

If it really is the water or roof that's behind it.

ADD: If you need a visual representation of this, get a camera with a zoom lens. Get on the side of the water that faces the sun. Zoom out until the sun and water glare is in the picture. See what happens to the picture quality? Now zoom in until the sun and most of the glare is no longer in the picture. See how much better the picture is? Same with wireless, just a lower frequency. Wave theory is wave theory, even in Malaysia!

But the log shows one of these disconnect/reconnect patterns at 30 past midnight last night so it's clearly not sun-glare. Nor can it be what I originally suspected, which is a heat-haze above the water.

When I first scouted the place I took a reading across the water to the southern AP with a 15dB Rootenna and got a signal in the mid -80s both ways, which was why I said it could be done - the southern AP then was just the 8dB omni, so it was vertical polarity. This was from the front of the property and probably only 3m above water-level.

I'm leaning towards it being a crap card. What would you recommend for a good stable signal at 2.4 using B?

BTW, great looking place. You wouldn't provide service to any surf resorts, would you?

No way. We're a non-profit community trust trying to provide broadband to our loal community by wireless because it's the only way we're ever going to get it. No commercial operation would operate in this area because of the mountains, the water, the remoteness and the kind of headache this link presents. And which is why we don't have the kind of expertise needed to deal with it.

About midnight here we pick up fog. Strange stuff. Causes a lot of diffusion. Not saying that is your problem.

BTW, 15db is twice the signal strength and half the "noise area" of the 12db. Try that in place of the 12. Not sure I have anything else to add, except check your connectors and alignment.

tombrdfrd66 -

I can't tell you exactly what is worng but I can tell you that what you are getting and what you should be getting are far different. According to just about every calc I can do you should be in the neighborhood of -65db at both ends of your connection. 17db power, plus 12db antenna, minus various losses - 3db= 26db effective power. Path loss in the neighboorhood of 110db. 27db-110db= -84db. Now add back the receiving antenna gain (minus losses of course) 24db - 3 db = 21db. So -84db + 21db= -63db Now you can add more loss if you want but what you are getting in the -hi 70/80s db is very wrong....

I have 3 links running over water at various heights - one as low as 15 feet above the water at one end and 40 feet at the other - at 8miles - all water.... I get better signal than you are getting now....

What I would look at:
Does your 12db omni have any downtilt? If so that can kill you at that distance and height.
You said you have three 2.4ghz radios (are these at both ends?) - you know you need to have 20mhz (or more) channel separation between each radio - otherwise you will self interfere.
Check your cabling - you could be losing a lot there...You should be using higher number LMR cable for longer runs. I.e. - LMR400 for less than 20 ft runs works great, if you are pushing 50 feet then LMR600 would be a better choice.
Check the remote side - is that cable from the card to the bulkhead ok? (Because you sure are losing a LOT of signal from one end to the other).
You can tilt your remote antenna upwards a few degrees to help with possible reflected signal off of the water which may well arrive 180degrees out of phase with the direct path signal - effectiviely reducing the perceived received power significantly.....

Let us know what you find.......

R/

Thom

Now that Thom mentioned his calculations, I can think of one more thing. It is a bit embarrassing, so I hate to even mention it.

I connected a box with two radios in it, both a-mode backhauls. When I booted it up, it worked, but the signals were bad. Both radios. Drove me crazy for a while, until I figured out that each of the radios were connected to the wrong antennas. I thought I had it down on which radio was which, but after checking the mac addresses, I found that the MikroTik box did not assign wlan1 and wlan2 to the radios I had expected.

I know it is a long shot, but Thom is correct. Your signal strength should be much better.

To Thom: I have been waiting for a chance to thank you for the tutorial on natted nets. I could not do what I do without it. Thanks!

ADD: I also see the Senao card has the potential for my other mess-up. It has two antenna connectors. Might want to check that. I use SR2's and the MMCX connector (the one I use) is the 'aux' antenna, not the 'main'. So antenna-mode=ant-b in /interface wireless for that.

tombrdfrd66 -
SurferTim makes a good point on the radio card connectors - a or b. I didn't research the card type to see if it had two different antenna ports - this is definitly something to look at as you are almost exactly 20db down from what you should be - and that is what you usually get when you have either a 'blown' diversity switch or are on the wrong port..... So check and make sure you have selected the correct antenna port.....


SurferTim -
I had hoped to get in touch with you when I was in FL last year - didn't work out though - I kept having to travel to/from the area..... I figured we'd swap some lies and have a beer.... Maybe next time - or if you get to the Seattle area - drop me a line - I'm a little over an hour away from downtown. You're welcome for the tutorials...all I ever ask is that folks share the wealth (knowledge in this case).....

R/

Thom

Hi guys, thanks for the advice. I'll check all these things, tho' I nowhere use LC400 longer than 1m.

Watching the client radio's advanced status in Winbox this morning (as it was up) the tx/rx was pretty steady around -85/-77 and signal to noise a solid 26/27dB. rx CCQ was constant in the 90/94% range but tx CCQ was all over the shop, from 34 to 66% in moments and changing wildly every update. Last night while I was watching it, tx CCQ went as low as 13% and then up to 60% in a matter of seconds.

I'm going to try changing the polarity of the link when I can get someone on the roof at the southern end while I'm at the northern end - I noticed that the beam-width of the 12dB mini-directional I'm using is 65' in horizontal (as it is - or as it should be, anyway. Perhaps it isn't!) and 34' in vertical. As 34' would be enough to cover the clients on the northern shore and is presumably a more 'concentrated' signal it should help?

The three radios at the southern end are working on 2412, 2442 and 2472GHz, with 2472 for this link. At the northern end there's an 8dB hotspot on 2412, and running a scan from it picks up the 8dB hotspot at the southern end, at -90!

I'm had something of a nightmare time with the southern end too, an RB433 running v3.20, as Winbox had suddenly started refusing connections to it and while I could connect to it by telnet and type into the command prompt, hitting enter just seemed to wipe the command leaving me with the prompt. Before this happened I noticed that the board was down to around 3MB of RAM out of its 61. Fortunately I was able to reboot it and it seems to be behaving itself again, free RAM is back up to 41MB, but I've a nasty feeling there's a memory leak crept into v3 again. I don't see it can be behind the signal problemsm, tho'.