The PLL factor is 390. This means you have to be only 5kHz off at the crystal frequency to get an error like that. This is not unusual when the crystal cut does not match the circuit you are using it in. E.g. when you order a crystal for series resonance and in reality the circuit is for parallel resonance, or the loading capacitance is different.
Fortunately it is always easier to pull crystals down than to pull them up.
So you could attempt to correct (part of) the error by using a small parallel capacitance or a small series inductor.
However, such pulling will reduce the stability of the oscillator, so it may be worse than what you have now.
Depending on your local bandplan this may be a quiet portion of the band.
(not here, it is the band segment where the FM repeater outputs are)
It looks like you have quiet weather there. Then, my freerunning LNB is quite stable as well. But the past days it has been rainy and windy, and then it is much less stable. The beacon locking helps a lot. You could try that too.
(or wait for a more ready-to-use solution)