Here’s screenshot of one of my stations logging brief (<1 sec.) disconnects every few seconds.
In between, as you can see, signal strength is fine. The board is a RB433 with a R52Hn card. The other end of a point-to-point link of 18km is an RB 532 with (I think) an SR5. nstream is running
Any more information required, please ask. Any hints, pointers, explanations gratefully received.
Just my thoughts on this - a link with different boards+radio cards can give unpredicable results, I would first have identical hardware on both ends, second what type/gain antenna is being used, CCQ is low at 74 and overall ccq of 33% explains why disconnecting, and was traffic being sent while you took screenshot. even your rx/tx rate at 6Mbps cannot give you a solid connection maybe radio card(s) faulty or antenna off alignment 
Thanks for the thoughts.
Here’s a shot from this morning - the problem stopped at 21 minutes past midnight and the link has been rock steady for eight hours. CCQ is up to 66% and jumps to >90%.
Antenna are 22dB parabolic. Misalignment is possible - we had 160kph winds through Wednesday night - but then why would it come right?
Identical hardware would be great but any time I have to replace a card/board in the system I find Mikrotik has replaced a perfectly good item with something ‘new and improved’ which means I’d have to purchase two of them and scrap a perfectly good one - also known as the Toyota marketing system.
Could this be interference? The beam passes over a harbour with ferries coming and going at intervals - tho’ there would have been a 4.00 am one.
Could this be interference? The beam passes over a harbour with ferries coming and going at intervals - tho’ there would have been a 4.00 am one.
Change frequency and see if CCQ improves, you have i assume no fresnel zone issues http://www.afar.net/fresnel-zone-calculator/
Your link from screen shot is 54/18Mbps which way too low (54/36 with snow buildup OK) and still too low ccq of 81%
As your using Ver3.30 i would recommend 2 X Ubiquiti XR5 radio cards there are the business I have a link running at 96-100% CCQ,
Ha! I presently have a collection of 7 (seven) Ubiquity SR5 cards that have failed on this link in the course of two years and am half expecting the current problem to be down to the pending failure of the eighth after an impressive three months service. Forgive me if I have some reservation in trying an XR5. That’s one of the reasons I tried a R52Hn, tho’ I won’t be trying any more as even the small print neglected to point out that I needed a minimum of RouterOS v4.03 to run the thing. Many of my RBs aren’t upgradeable to v4 and I’m damned if I’m going to fork out US$45 for another licence just for the privilege - Toyota marketing again.
XR5 cards are the most stable cards out there I think. 100% ccq on every single one of ours, same links with anything else and its horrible.
Routerboards always can be upgraded (so far). If it says you can’t just go as high as it lets you, and then it will let you upgrade again to the next higher. Even the 1xx series boards can be upgraded to 5.x if I remember right.
Not my understanding, and not what the Wiki section on licensing implies, vis:
"# In regards to RouterOS upgrades, RouterOS can be Licensed with two types of keys, L3/L4, and also L5/L6
[snip]
So the math is:
eg. L5/6 = v3 + 2 = v5.21 you can use
Examples:
If current version is ROS v3, L3 and L4 will work with v3.1, v3.20, v4,1, v4.20 but NOT v5.0 and beyond
If current version is ROS v3, L5 and L6 will work with v3.1, v3.20, v4.1, v4.20 and also v5beta1 but NOT v6.0 and beyond
If current version would be ROS v4, L5 and L6 will work with v4.1, v4.20, v5.1, v5.20 and also v6beta to v6.99 but NOT v7"
I’ve several boards with a level 4 licence which I’ve upgraded from v2 to v3, and cannot upgrade to v4.
Damn well ought to be at the price - Mikrotik R52Hn=NZ$104. Ubiquity XR5=NZ$179.
Ha! I presently have a collection of 7 (seven) Ubiquity SR5 cards that have failed on this link in the course of two years and am half expecting the current problem to be down to the pending failure of the eighth after an impressive three months service. Forgive me if I have some reservation in trying an XR5.
I have never had a failure of a XR5 this probably due to I being extra cautious when installing them (yeah I agree they are costly) and as you have mentioned 7 SR5 failed on just one link?
(1) I forgot to ask at first was this link working OK and just deteriorated over a period of time or just not good from the start?
(2) Has the SR5 failure happened just at one or both locations of this link,
(3) Any radio card (and more so higher power cards) are very intolerant to poor RF load/connections and will blow, have you replaced pigtails, redone N connectors, replaced any link lead to antenna, checked antenna/leads for moisture ingress , checked connectors.
Note - I had a 2 X AP’s using XR5 but had low signal turned out to be caused by pigtails mmcx connector not making a good connection onto the XR5 replaced the pigtails and signal increased , i was lucky the XR5’s were not damaged.
(4) Is the voltage good and clean at both locations, have you UPS,
To be blunt you don’t simply replace a costly component the second time and let it go faulty in that configuration, you load the radio card with a dummy load to ensure it’s working ok then replace with the cheapest radio card at hand, and start a process of elimination-note signal at other end (if you cannot get a signal from the other end at 18kms then setup a temp link a short distance away) and start replacing leads (pigtail, link lead, even antenna) if there is increase in signal level that speaks for itself?
For the most part the failures were sudden - the link just stopped working. Part of my problem is that one end of the link is at the end of a difficult forty-minute quad-bike wrestle up a mountain and the other end a two-and-a-half hour drive away, and I’m not aware of any easy way of telling which end has failed. So I replace the mountain-top end first and if the link stays down replace the other end and assume the moutaintop one was OK and can be re-installed at a later date.
Difficult to say, as I’m beginning to suspect the problem was interference all along.
In the absence of obvious damage I tend to leave well alone - and we’re a very small, not-for-profit community organisation which can’t afford to carry much in the way of spares, and certainly can’t afford to replace things for the sake of it.
The mountaintop is solar, with voltage drawn directly from the batteries. The far end has 18v pushed 60m up a hill from a UPS.
To be equally blunt we have neither the resources nor the facilities to pursue a policy of perfection - anything better than dial-up at 2kbps is a bonus. However I take the point that assuming pigtails, cables, etc. will do properly what they are supposed to do and what we paid good money for them to do might be foolish.
When the link was playing up last evening I changed the frequency of the link to 5200Mhz and it’s been rock solid with a CCQ of 90 - 95% ever since. Which suggests interference, probably from the shipping. Doing a Frequency test from the other end cuts me off from it (and takes our main backhaul down!) so I can only do it when I’m there, which isn’t very often, and I’ve never seen much use when I have done one. So perhaps it’s one particular ferry.
Where in NZ are you? what are the altitudes of each point? You could be seeing ducting - we see this a lot in Northland when we have very pronounced temperature changes (hot day, clear cold night) and also when humidity is high (over summer)
Also R532’s are getting a bit old now - we have one or two in production but both of them give us weird results from time to time (especially with ROS4 and newer radio cards)
I would have thought there were clues enough in the screenshots to tell you that.
Ducting?
The station end is 500m, the AP 100m. The beam crosses a lot of water.
The R532s were Mikrotik’s latest and greatest when we began the network - and none of MT’s replacements for the 532 fit the enclosures we bought for them so replacing a 532 with eg a 433 means replacing an enclosure as well which is just throwing away good money. So we’ll run the 532s for as long as they last, and if we can’t upgrade them because of licencing problems it will be because Mikrotik don’t put enough thought into their clients needs.
I’ve had no problems since I changed frequency, so interference from one of the ferry’s seems most likely to be behind the issue.
When crossing water you can / will see major variations in signal levels do to reflections - especially if the AP end is only 100m ASL - you will also see ducting. There are heaps of topics on here about dutcting - very common, especially in the summer months over water.
Putting the frequency lower means a longer wavelength - depending on waves / tide heights it may be that this is why the issue is fixed - however at 5200mhz, legally you’re only allowed to run 1Watt - and there’s no way your link would stay stable over water at 1Watt
BUT - if its stable, its stable 
Great! If it becomes unstable again see if you can lift the AP end any higher - maybe try an N card ($100 give or take from NZ)