I have a 23 Mile link over water, that is currently V-Pol, and not working too hot. We are slated to swap the antennas to H-Pol, but I am wondering if N would be an optional at all. Radio Mobile puts the link at -60 with 29dbi grids and R5H cards. We have seen as good as -55, and as bad as -90 and lower. Is N ready for prime time? Should I get two SR71-15s to try it out?

R5H or H-pol or V-pol has noting to do with signal strength or maybe u can get +/-2 dBm depend of situation, I have over 35 5Ghz Links I never seen that kind of difference from -55 up to -90 this is a lot.
if the humidity is big, or if it rains my links drops in worst scenario up to -4 or -5 dBm
My opinion is u are losing line of sight in that time if it happens frequently (if u'r shure That Antennas, cables, or ur hardware are ok and isolated good)
but u have to send more info...

Did you just put the link up or did you perform a path analysis beforehand?

Over water links, especially longer ones, normally suffer from multipath and ducting. If you deploy a single pole link, regardless of polarity, you will experience issues at some point unless you have extremely high endpoints relative to the path (in excess of the established 22 deg minimum.)

Carrier grade equipment gets around this by using spatial diversity, which N finally brings to the table. So using a 'real' N card, with at least 2x2, and more importantly, spacing your dishes at least 3m apart vertically, should compensate for multipath. It is important to note that when this compensation occurs, be expecting the throughput to drop as only one stream is being used.

Changing to carrier grade equipment that uses time and spatial coding is another answer.

Did you just put the link up or did you perform a path analysis beforehand?

Over water links, especially longer ones, normally suffer from multipath and ducting. If you deploy a single pole link, regardless of polarity, you will experience issues at some point unless you have extremely high endpoints relative to the path (in excess of the established 22 deg minimum.)

Carrier grade equipment gets around this by using spatial diversity, which N finally brings to the table. So using a 'real' N card, with at least 2x2, and more importantly, spacing your dishes at least 3m apart vertically, should compensate for multipath. It is important to note that when this compensation occurs, be expecting the throughput to drop as only one stream is being used.

Changing to carrier grade equipment that uses time and spatial coding is another answer.

The link was modeled with Radio Mobile, but I don't believe the atmospheric effects over water are taken in to account with this. I have been looking at other options. In my mind I have two options at the moment, trying two SR71-15s, or a Radwin 2000. If N is mature enough on MT, it is worth a try.

I just swapped this link to H-Pol, and it needs to be watched
overnight, but looks good so far. Signal fluctuating between -59 and
-66 on a 20mhz channel, CCQ at 90/90 or better. After flipping to
H-Pol, the channel was still set to 5Mhz, and the same fast start and
slowdown was occurring, the radio would disassociate with poll
timeouts and too many retransmissions. Switching to a 20mhz channel
fixed this.

status: running
duration: 3m59s
tx-current: 15.7Mbps
tx-10-second-average: 18.0Mbps
tx-total-average: 17.4Mbps
rx-current: 16.3Mbps
rx-10-second-average: 17.2Mbps
rx-total-average: 17.2Mbps
lost-packets: 60
random-data: no
direction: both
tx-size: 1500
rx-size: 1500

You cannot model effectively over a predominent water link. You are in the lap of the gods.

Previous post hit the nail on the head... Ducting.

Ducting is more common over sea paths, can happen each day...

Hint... Look at smoke stacks if there are any... Does the smoke rise or go horizontal.

You cannot model effectively over a predominent water link. You are in the lap of the gods.

Previous post hit the nail on the head... Ducting.

Ducting is more common over sea paths, can happen each day...

Hint... Look at smoke stacks if there are any... Does the smoke rise or go horizontal.

A few hours after posting my message the signal fluctuations came back. I am going to try dropping the higher end to around the same height as the lower end (40 meters ASL or so) and see if things stabilize. Swapping frequencys from 5.8 to 5.3 or vice versa seems to help the issue. I'm wondering if two links bonded, one at 5.3 and one a 5.8 would make sense.

Did you read my post and the solution for long links over water?
http://forum.mikrotik.com/viewtopic.php?f=9&t=31644

Did you read my post and the solution for long links over water?
http://forum.mikrotik.com/viewtopic.php?f=9&t=31644

I did read that thread, and had considered the same. At he moment, my plan is to upgrade the 29dbi grids to 32dbi parabolics, drop the higher end of the link to the same height as the remote end to minimize ducting losses, and replace the R5H cards with XR5s for a bit more TX gain. These changes, along with shorter cable runs off of the ODU to the antenna adds 10db to my link budget, which will hopefully put me in a good spot. I'll keep you posted. For the past 24 hours the link has been running at 5.8Ghz, H-Pol, and been relatively stable, but without much sun today I don't think the evaporative ducting effects were out in full force.