I wouldn't calibrate to the UA5 peak light, as doing so assumes an unconfirmed value at which the UA5 peak light triggers. I'd feed the UA5 an analog signal of white noise (as already described), with the SVU hooked up wherever you plan on connecting it. I'd then send the UA5 digital output to my soundcard and monitor the actual levels through recording software. This is the only way to ensure the meters are calibrated to an actual level, and not an assumed level.
When I ran an SVU-2, I switched the L/R channels. Makes more sense to me to have the L on the top, but really I did it because I ran the SVU vertically (like the V3 meters), which - without swapping L/R - puts the R on the L side and vice versa). I believe the SVU-1 is already oriented vertically, with L and R properly assigned.
I also ignored the dB values on the SVU and assigned my own values. So on the vertically oriented SVU-1, I would calibrate the top red LED to 0 dBFS. Each LED down then corresponds to -3 dB relative to the LED above it. This scale matches the meters on the V3 (0 dBFS > -27 dBFS). And the coloring scheme is easy to monitor from a distance: amber = -6 dBFS, i.e. approaching 0 dBFS with reasonably comfortable headroom (depending on what one's recording), first red = -3 dBFS, i.e. getting quite close to 0 dBFS, better watch those levels closely, and second red = 0 dBFS, i.e. turn those levels down!