If I can add a couple of things I've noticed:

For accuracy, I also always add a DateofElements to the .sgo file with the julian ephemeris date for the elements. I'm not sure what Uniview does otherwise (assumes 1/1/2000 12:00?).

Also of note, the "position.conf" file will often contain a number after the location for the SGO or BSP file, like "1495.9"--which is to scale the orbit from AU to 100,000km (the usual default unit for solar system coordinate spaces). For satellites orbiting planets, this value is usually 1.0 and the semimajor axis should be given in 10s of KMs (ie, if the semimajor axis is 2000km, it should be put in the SGO as "200"). For .bsp files using AUs in the Solar System scale of 0.00001 is assigned, and for those orbiting planets, 0.0014959

BSP files for some active missions can often be found at NAIF: http://naif.jpl.nasa.gov/naif/data_archived.html

However, SPICE kernels are restricted by ITAR, so the most current versions aren't always released (the available New Horizons one, for example, only runs from launch to 2010 or so). It's possible (but extremely complicated) to construct them from Vector data obtained from JPL horizons (which itself pulls from SPICE kernels, including those not available publicly--just the vector data and reconstructing kernels from it is not considered an ITAR issue). This is a REALLY complicated process though, but if there's interest I'd be happy to make a write-up eventually.

For Earth-orbiting satellites it's also possible to use a TLE file, which contains all the keplerian parameters: http://www.celestrak.com/NORAD/elements/