Weaver SSB Spectrum Analysis

This spectrum plot of a single tone Weaver-generated SSB signal is not good:

It is supposed to be a Single Sideband signal at 3927Khz modulated with a single 600 Hz tone Lower Sideband - which should produce one pure carrier at 3926.4Khz. However we're getting spurs as far down as 3894Khz and as far up as 3958Khz, and they're only 12db down from the main signal! I worried that the unwanted components might be a show-stopper for getting a GNU/Radio signal on the ham bands. However, just a little tweaking cleaned it up nicely. The signals reflected about the Digital UpConverter center freq of 3900Khz were due to too few taps in a Hilbert filter, which changes the float stream to complex just before sending to the hardware USRP. The other unwanted component was due to setting the 2nd, the interpolating audio filter cutoff way too high. Fortunately it didn't need to be any sharper, which would use up more cpu which is running at 80% as it is.

Here is the cleaned up signal:

This plot has the same scale as above, and only the 3927Khz LSB signal is present in any significant amount. This is actually a two tone test, with 440hz and 1400Hz audio modulation, same carrier point of 3927Khz and DUC at 3900Khz. Here it is zoomed in on the carrier:

Now I'm ready for the big amplifiers!