Frequency offset cause...
The LO in the LNBFs is usually +/- 1 to 3 MHz in frequency. The receiver assumes 5150, 9750, 10600 or 11250 MH LO.
The (transponder freq) - (LO freq) = IF freq. The receiver determines the IF frequency of best signal strength/lowest BER and ASSUMES the LO number to be one of those listed above (depending on the receiver's LO setting)
the ACTUAL LO freq can be that value +/- 1 to 3 MHz for most LNBFs. This will yield a +/- 1 to 3 MHz offset in your scanned TP frequency.
The LO freq can change with temperature, voltage or mechanical vibration.
In most cases the tolerance/drift is not a problem, but does increase the BER (reduces quality)
If you have a very weak signal and/or narrow transponder signal, plan ahead, let everything run for 20 minutes before trying to get it (so the LO is stabilized) and scan for it. You can also try to manually enter the TP freq at (nominal) (+ 1 MHz) (- 1 MHz) (+ 2 MHz) (- 2 MHz) etc until you hit it. More than +/- 4 MHz is not necessary. [I assume the receiver IF is +/- 1 MHz tolerance at most, it's probably less than that]
There are some commercial LNBs which have very accurate LOs, down to 5 kHz. They are higher noise, very expensive and require a polarization selector (or 2 LNBFs at 90 degrees to each other) plus a feedhorn for each, one such brand name is Norsat. (they don't care about low NF because they use huge dishes to guarantee a stron signal)
Unfortunately such stable LOs do not appear to be available in the low noise consumer LNBFs. Putting a good temperature control and tight ripple filter on the LNBF will help but these are tricky to build. I'm an experienced RF engineer but I'm not in a hurry to do this.
However, I'll try this trick with SatMex 5 (the 12080 transponder is zip for me) to see if I get anything.
I assume the blind scan locks onto the center freq of the TP.