The math is simple but the LO is not a direct contributor yet required. I considered 2if-LO as possible
as well as aboth follow the spur. Usually 2if spurs only show up when we hit the 1db compression
point of the mixer and then they stat to grow. This spur is there even when the mixer input (45mhz)
Changing the LO does not help. The spur is 45mhz (IF) minus output frequency (dial frequency)= spur
So for 45mhz and 28mhz out the spur is 17mhz. Change to a low side LO to 45-17mhz(new lo)=28 plus spur plus
mixer leakage and the IF-dial frequency spur. Measured isolation for the mixer was about 40db varied about 1 db
from that with frequencies so the mixer is very good but not perfect.
Remember your line and mine, DBM is a conundrum [headaches] in three ports. Any of the ports are an
input AND output so even if we assume the RF port is the signal out its also a signal in. So the LO generates
28mhz if its 73mhz or 17mhz and the spur will be the IF always 45mhz minus the output frequency and the
lo is required but what frequency is only dependent on creating the output frequency. Happens with
commercial mixers too.
It works up to 20mhz as the spur frequency is more than the lowest low pass. The spur being an image
spur goes down with increasing frequency. So at some point a low pass filter cannot stop it.
Nyquist sampling theory forecasts it. At 22.5mhz is the alias point the spur overlaps the
generated frequency. So does Spurtune[program]. Higher IF greater than 60mhz would
work. Or the traditional solution band pass filters above 20mhz and low pass filter below that.
Also the assumption is when not talking there is no spur (false) as you sill have carrier
leakage so 45mhz is always there.
It does not appear for CW as the mixer has bias on it to make it a switch and the LO is at signal frequency
without any significant mixing action and nothing else to mix with.