I was trying to process a number of B band images for RS Sge from SRO50. If you process each image individually, you get B mag values of approx 15. If you do a time series over the same images, VPHOT reports mags between 11 and 12. Pretty far off.
Please check that all comparison stars have photometry for the B band. There is an issue in VPHOT that missing photometry is taken as 0 (zero) in the time series ensemble. If so you can disable these comp stars on the time series results page.
(I'm currently on vacation and not able to look at your caes in detail until Sunday).