The values that are relevant are RA_SLT, DEC_SLT, PA_SLT. RA_SLT and DEC_SLT are different for different nod positions while RA_FOV and DEC_FOV remain the same between nods. So let's say your PA_SLT is 0 degrees (that means the long axis of the slit is rotated 0 degs east of north and the top of the slit is due north of the bottom of the slit). If PA_SLT is 90 degs, then the top of the slit is due E of both the slit center and the bottom of the slit. Now to map this onto a BCD.
For SH and SL, the source is first at the slit bottom and then the top (for the two nod positions). The bottom of the slit is mapped to the left of the BCD and the top of the slit is mapped to the right of the BCD. For LL and LH, the source is first at the top of the slit and then at the bottom. The top of the slit is mapped to the right of the BCD and the bottom of the slit to the left.
Now you want to calculate the coordinates of your source. Let's say your source is at pixel x. Then its offset from the slit center is alpha = (x-xcen)*pixscale in arcsec (you can look this up in the wavesamp_offset file or you can read x and xcen off with your cursor)
This has to be projected in RA and DEC.
delta_RA = alpha*sin(PA_SLT) in arcsec
delta_DEC = alpha*cos(PA_SLT) in arcsec
note that alpha is negative to the left of the center of the slit and positive to the right of center of the slit. So for all PA_SLT values between 0 and 90, delta_RA is -ve. For PA_SLT values between 0 and 180, delta_DEC is -ve.
Since cos(DEC) is pretty much constant over the length of even the longest subslit:
true_RA = RA_SLT +/- (abs(delta_RA)/3600.)*sec(DEC_SLT)
true_DEC = DEC_SLT +/- (abs(delta_DEC)/3600.)
where you use +/- depending on which quadrant you are in. Of course, to save you all this trouble, you can just extract a spectrum in SMART and save the extracted spectrum and it will give you the coordinates of the source you extract.