I cheated with the use of a diode probe (series schotky diode for UHF) insert to produce a poor signal when the sig gen is set to 1V output. This caused the RF sine wave is bent enough to see the harmonics on 6MHz on a scope (>20MHz) & the diode output is terminated into 50Ω by the spectrum analyser.
The filter I had is an 8 section M derrived video LPF & is 40dB down @ 8MHz, but only 2dB down @ 6MHz. But any suitable filter will show the benefits of a LPF after a harmonic rich source.