Ok, I'm working (also) on building a spectrum analyzer. I have an electronics instructor that once told me that an LO signal doesn't necessarily have to be a sine wave at all. Last I checked, a spectral analysis of a square wave is pretty nasty....Has anyone here tried something like this especially in the VHF region.
Most mixers will generate harmonics of the LO and that is why receivers are designed with narrow band filters on the RF and IF ports so that the mixing products not wanted do not affect the system. IF = (M x RF) +- (N x LO) where M an N are all integers.
If you drive a diode mixer with a square wave is the best solution because the diodes act as a switch.
But this drives to mix of the armonics of the LO signal with the spurius that could be into the spectrum at the RF port of the mixer and the product could fall inside the IF chain corrupting the received signal.
For example i have build the spectrum analyzer described by Matiaz Vidmar but into the IF chain i have changed the crystal at 60 MHz with a clock block for TTL drive with excellent results ( That day was not available into my lab a crystal with the proper frequency !!!!! )