You ask about QCX noise floor as antenna is connected. It is a good test to listen
with a 50 ohm dummy load, then switch to your antenna, and listen for the noise
floor increase - if noise comes up, your receiver passes the test.
Have compared my 30M QCX against a 40M SmallWonders radio. Not really a
fair test, since noise floor increases at longer wavelengths, so the test is really
subjective. A poor antenna connected to the QCX did bring the noise floor up
noticeably. The same antenna brought up the noise floor by a larger amount
on the SW40+ receiver. My 30M QCX measure out with a noise floor around
-121 dBm, similar to most others.
I did do a modification on the 30M QCX. Have run afoul of 7805 regulator chip
noise in a previous receiver. In this one, the mixer bias (via R1, R2) is fed by
the +5V from IC11, 7805. It is bypassed by C6, a 0.1uF capacitor, which is very
adequate @ 10Mhz, but inadequate at 700 Hz audio. Adding a 10uF capacitor
in parallel with C6 drops the noise floor by nearly 2dB. Am assuming that this
audio noise passes through the mixer, entering the audio chain.
7805 voltage regulators are not well specified for noise - I suspect that some
are better than others. It is almost impossible to add a filtering capacitor to its
output in an effort to kill noise (at 700 Hz) because its output impedance is
just a fraction of an ohm. Impedance at R1, R2 is much higher, so a bypass
capacitor is more effective.
2 dB is a piddly improvement in noise floor, hardly worth the effort. But I do
intend to probe further to see where noise comes from. There's 30 dB of
noise to account for. Some of that is RF bandpass and/or lowpass filter losses.
Some is mixer losses. Much is likely audio preamp noise figure.