Cross pol issues in stabilized antenna

Dear community,
First message here. To be honest I found the forum during research regarding a problem with an antenna.
My name is Michael and I am marine electronics engineer.
I sending here because I have faced a really strange problem while commissioning a VSAT service.
First things first antenna is a seatel 4006 and modem an Idirect x5.

During line up provider informed me that cross polarization was not good (at about 20 db) and need to be improved. After that routine procedure followed with adjustment of skew in order to improve it, but with no result(!). My first thought was that skew motor was not working. I have disassembly the antenna check all systems but is seems that all worked OK! Motor is working good, I could even manage to track satellite with different Rx polarity and different skew settings with good receive signal and without too much effort. This problem confirmed with three different satellites, operator always report bad isolation, so surely it is an antenna problem.
Any ideas?
What component could affect isolation? Tx chain consists of the Buc wave guides connected to the feeder with wave guide rotary joint. No circulator, there are some bends though(according to drawings at least) . Operator suggest to try to clean the feeder but to be honest I don't believe this will work.
Have you ever experienced anything like this?
Any ideas?

Thanks in advance
Best regards,
Michael

I suggest you tell the manufacturer and ask for help. If your equipment differs from that previously made which tested OK then that might point to the problem.

Here are some thoughts from me:

Is the polarisation motor setting the correct angle? Most satellites have their linear polarisation aligned with the earth's axis and equator, but not all. The tracking controller may not allow for different satellites having different offsets. Eutelsat and Astra are know satellites whose polarisation is not aligned with the earth axis/equator.

Is the polarisation motor mount aligned in its mount?. If you point due south/north the motor should go to zero or 90 deg. Does the readout correspond to actual measurement?

Just in front of the OMT the waveguide becomes circular or square. From this point on to the horn it must be straight. Flanges must be tightened evenly so that they are exactly square. If there is a long waveguide tube to the horn then it must be straight and not bent at all.

No loose nuts, bolts or washers (or water!) should be inside the feed system.

It is unlikely, but if the OMT is a 4 port device then the unused ports need to be terminated.

In front of the LNB will be a transmit reject filter. The distance from the filter to the OMT junction must be correct.
Some filters are not symmetrical and one particular end is supposed to go to the OMT, the other to the LNB.

Trying to align polarisation using receive quality is difficult as angle errors may have little effect on receive quality. You could try turning a long way each way, and measuring the exact degradations. If you can do this reliably and get two exact same degradations and then halve the angle between them that will work. It helps if the cross-pol traffic is steady carriers (not TDMA). With a spectrum analyser and narrow resolution bandwidth, alignment on the cross pol beacon null is a good method. Even then you need to swing either way, a couple of degrees, and then halve the difference.

Short F/D dishes with offset feeds have poor polarisation performance and designers should use mode matched feeds as available from Prodelin.

If anyone can add further comment here please do so. If you are not a member register at the extreme bottom left of the forum home page.

Best regards, Eric.