I've been getting reminded of just how awful old-time RF signal processing stuff is.
I have a system with two VCOs, f_x and f_y, whose output is at 2f_x +
2f_y. f_x and f_y both vary over a full octave. I need to do a vector receiver for this.So, I needed to construct 2 f_x + 2 f_y + f_IF.
After several tries, I wound up with
f_x -> LPF -> 20 dB coupler -> pad -> doubler -> pad -> LPF -> HPF (the order matters) -> amp -> LPF = 2f_x
f_x -> LLPF->20 dB coupler-> pad-> PLL offset by +12.5 MHz -> LPF -> pad
-> doubler -> pad -> LPF -> HPF (the order matters) -> amp -> LPF -> pad = 2f_y+25 MHz
these go into
mixer (+7dBm LO, -6 dBm RF) -> pad -> LPF -> HPF = forest of spurs.
I'm currently throwing together a 1:1 PLL to clean this mess up.
If I'd had enough of the right filters, I could have done all of this by double conversion and got rid of most of this crap, but oh, well.
The real version is going to use DDS/PLLs to do all of this, of course.
Cheers
Phil Hobbs