Do I need to specify the standard calibrations as being a calibration curve set? I'll have another look but I couldn't see how to calibrate without having to specify a background of some sort.
So you can use any background type but you should use the same background type for your standards and your unknowns.
For instance generally you would just do off-peak bgds on both, and if you do you get an extra benefit in that you can (from the Analysis Options dialog) specify to add a 0,0 (synthetic) intensity versus concentration point to help the fit.
If you decide to save time you can specify using both MAN backgrounds for the stds and unks, but then you cannot add a synthetic 0,0 point into the fit, because then zero concentration won't have a zero intensity.
Either way, once you select multi-standard calibration curve it will figure out the rest automatically- no need to even assign standards!