Hello.
Still cant figure what is causing this issue. I have dozens of long distance connections running perfect. This one is 10km away with good LOS. AP 90° 17dBi sectoral antenna, CPE side A (that who has low PThroughput) 30dBi parabolic dish (Made by MT with new type feed). CPE side B 22dBi flat panel. both A and B CPEs are connected to same AP and runs same settings and R52 radios.
Bearing in mind high gain dish’s alignment is far more critical than a panel antenna, you have rechecked alignment, also have you tried another AP frequency and see if there is any improvement, the signal of 80 using this high gain dish is low have you tried a panel and what result, CPE R52 - what radio card is used at the AP
I did alingment, there shouldnt be problem. AP uses R52 as well. What u mean by using panel? Panel is connected to same AP only distance is ~5km. Only one im not quite sure is FZ. 
AP ir 43m high, 30dBi parabolic dish ~7m.
Panel - I mean have you tried in place of the dish another antenna and compared results,
For comparsion with my network we are using AP - 433+XR5+16dbi intergrated antenna and for CPE -Rb411+UB5 radio card or RB711 with builtin in radio cards both use 23dbi integrated panel antenna’s, distances of 5 to 7km signal would between high 50 and low 60 dbi range,I have one 7km Ptp link using 433+xr5 + 24dbi grids both ends and it’s signal is 48dbi,99-100% CCQ,44db SNR from your screen shot using R52 83 for panel at 5kms to 80dbi for 30dbi dish for 7kms?
My guess is if not alignment or interference (have tried another AP frequency ) is the radio cards R52 is just low powered and this is causing low signal + low SNR + low CCQ ( from screen shot you are using reduced data rates which should result in near 100% CCQ at that distance )
currently cant set panel instead of dish. Current -75dBi and P Throughput is 13000Kbps (above). thinking to order XR5.
Sorry if I missed it but what ROS version and wireless package are you running?
Dish setup runs RB532A with ROS 4.15 and
Flags: X - disabled
# NAME VERSION SCHEDULED
0 dhcp 4.15
1 system 4.15
2 wireless 4.15
3 advanced-tools 4.15
4 calea 4.15
5 routerboard 4.15
Flat setup runs RB411 with ROS 4.5 and
Flags: X - disabled
# NAME VERSION SCHEDULED
0 routeros-mipsbe 4.5
1 system 4.5
2 X ipv6 4.5
3 routerboard 4.5
4 routing 4.5
5 dhcp 4.5
6 wireless 4.5
7 hotspot 4.5
8 mpls 4.5
9 security 4.5
10 advanced-tools 4.5
11 ppp 4.5
Main problem is signal. You don’t want a point-to-point link to be in the 80s. With that distance and a 30db dish, signal should be in the -60s or better. I would bet on alignment. 30db dish antenna need to be in almost perfect alignment. There is very little room for error. If not, the only other issue would be swapping out the coax jumpers/cable. The R52s should easily handle that distance with a 30db dish.
Eric
Free Space Loss at 5500MHz at:
5km=121,2dB;
10km=127,3bB. see here: http://www.terabeam.com/support/calculations/free-space-loss.php#calc
So 6,2dB weaker signal at 10km compared to 5 km if antennas are equal.
But since you have 30dB compared to 22dB antenna gain, so at side A you have 8dB more antenna gain than at side B. So at side A you should have 1,8dB stronger signal compared to side B. And your screenshots confirm that, this means antennas at side A and B are ok.
So I suggest you to search the problem at AP side.
could it be low FZ? There is another CPE connected to same AP - runs cheap 29dBi e-zy.net diecast grid, only that CPE have ~30000kbps p throughput and its ~7km from AP. Signal ~-67/-67 Tx/Rx and 100% CCQ. But if i start download something with max speed (~0.9MB/s), that 29dBi CPE P Throughput goes down to ~14000-16000kbps and CCQ ~55-65%.
I agree with kkolev. But I have still seen these types of illogical signals with improper alignment. Particularly when you are just catching the side lobe of the signal.