The final result of the belt modification is here.
Thanks to Bert (Avandonk) who kindly offered me his spare hand paddle (because the processor in mine stopped working, together with the processor on the main board...The main one was easy (Element14, or formerly Farnell helped) but the one in hand paddle would have been more difficult problem to solve without Bert's help).
Interestingly, there is a low frequency signal present (green line) , it must be the flaw associated with worm gear (I can't imagine it is due to fluctuation of crystal frequency).
But, overall the PE curve is smoother than before and definitely correctable by autoguider.
PE amplitude is around 9arcsec (5.6 RMS)- all original bearings and no additional grease.
Now, I only need time to start using it... well, retirement is hopefully getting closer