[time-nuts] Re: Noise down-converter project

[ed breya]

After more experimenting and measuring, I think the RF is done enough. I 
found that the main 20 MHz and the 300 MHz LPFs were letting a lot of HF 
above 1200 MHz straight through. I believe the choke I used in the 300 
MHz LPF has a self-resonance in that range - it's a wire coil molded in 
plastic type, which may have a lot of inter-turn capacitance, spoiling 
the HF. I had used the same kind in the diplexer, and also as a helper 
choke at the 20 MHz LPF input.

It looked pretty bad above 1200 MHz, but some quick experiments with the 
HP8568B and its TG, let me figure it out up to 1800 MHz and correct it 
without having to change a bunch of parts, and possible complications. 
The diplexer is now a different HPF that looks like a short (instead of 
a termination), reflecting most back to the mixer, and an added 475  R 
damping resistor across the helper choke took out the overall filter 
resonance above 1200 MHz - all the way down and flat as the MF portion 
of the stop-band, up to 1800 MHz. It turned out that using the HPF as a 
short is beneficial - the minimum signal levels appear at the LPF, and 
it cut the DC offset nearly in half..The possible cable resonances do 
not seem to be a factor, and it looks like most of the >1200 MHz 
spurious content was going straight through the filter system, and not 
around it.

All of the LO harmonic spurs are now -100 dBm or less, basically in the 
noise floor, and only discernible by zooming in around the spot 
frequencies. I finally settled on 20 nF for the CM capacitance, so 
ground loop CMR should be very good. Now that the RF is "done," I can 
wrap up the DC offset control - unless I discover some new tangent to go 
off on.

Ed