If you don't see changes in uBITX freq error after you change the master cal in MM, then you need to fix the process problem first.

I just did the math, and for every 1 hz change you desire in the uBITX vfo display, you change the MM master cal by +/- 87.  

Is that coor or what?

  I have not tested or validated this notion, but it seems right based on my poor memory of what I did last year when trying to minimize my beloved uBITX freq display error via successive MM changes.