Strange attenuator behaviour

Thread Starter


Joined Apr 7, 2022
I'm doing an RF test where the output of a 2.4GHz 2W PA is connected to a SA through a few attenuators. The signal is QPSK modulated data.

When I put 3x 30dB attenuators in series, I get the expected power measurement of -57dBm (33-90) at the SA. But when I put a 4th attenuator of 20dB, I still get the same power level of -57dB at the SA rather than -77dBm. The noise floor of the SA is -93dBm with the selected RBW. What could be the reason for this strange observation? Some kind of leakage?


Joined Jan 27, 2019
It's certainly a mystery. The fact that it doesn't change at all is the most curious thing. If it didn't change as much as expected leakage might be a reasonable guess, but to stay the same is either an amazing coincidence or... something...


Joined May 10, 2009
Yes, I have checked the attenuators on a VNA, and it's as expected. And the first 2x 30dB attenuators are rated for high power (20W).
Then not all the signal is going through the attenuators. Getting 110dB of attenuation on the bench is certainly doable, but it does need care. Shielding, good cables, good connectors, good attenuators, well shielded spectrum analyzer, etc... What happens if you terminate the last attenuator and terminate the input to the spec an, keep the terms and connect the cable shields etc.. Connect the 2W PA to an antenna and simply terminate the input to the spec an - does it read anything?