At the SETIcon 2004 raffle, I won an N1BWT "weak signal source".
It has served me well since then, but sometimes I wished it
would produce a really weak signal. It's signal is
weak compared to normal HAM transmitters (EME...), but compard to
received levels in radio astronomy, it is still extremely powerful!
Sometimes one needs a REALLY weak signal, below -100 dBm, for example -140 dBm can still be heard by ear on a good SSB receiver, and with DSP one can see even lower levels.
If you start around 0 dBm, it is very hard to go that low with attenuators - in most cases the cables and boxes will leak much more than that! A few times I was forced to put the generator and half of the attenuators in one room, and run a cable into a second room, where the other half of the attenuators and the tested receiver were.
To avoid such "decibel limbo dancing" (how low can you go?) it would be nice to have a source that doesn't generate a high level of a signal, not even "inside the box". This way, there is less need for huge external attenuators, and the leaks aren't such a problem anymore.
The documentation of the WSS suggests to leave out the last stage of amplification to reduce the output power. By replacing the last MAR-3 stage with a copper strip link (the bias components R11,C20 and L6 can be left out, too) the output power is somewhere between -5 and 0 dBm.
Next I tried to cut the signal at the source, and removed the multiplier diode D1. The signal went down only cca 40dB - obviously, the xtal oscillator itself produces quite a bit of 12th harmonic, which gets then amplified by the long MAR chain.
I wanted to go lower than that, so I switched off the supply to IC3 (MAR-6) by lifting R10. I left the MAR-6 in the circuit. It has more attenuation that a copper strip would have! This reduced the power by about 30..35 dB, to cca -40dBm with D1 installed or cca -80dBm with D1 removed.