According to the FT8 Operating Tips guide available at
www.physics.princeton.edu:
"If you get your transmit levels badly wrong (such as the FT8 station labelled 0 in pink below), you mayunknowingly be generating ‘ghostly barcodes’ (1-5): despite looking like some obscure new
digimode, these are audio harmonics caused by generating too much audio output from the PC
sound card which overloads the audio input circuits in the radio"
The operating guide is here:
https://www.physics.princeton.edu/pu...ating_Tips.pdf
Secondarily, something that is not mentioned in the OTG:
The Windows Audio Subsystem, along with software like VAC/Voicemeeter/USB audio codecs, do resampling when an audio signal has too much gain. This is to reduce digital clipping. However, while the digital clipping is mitigated by the resampling, the effect is a compression of the audio which can create artifacts as shown in my waterfall picture (and is also observable on a spectrogram). In effect a widening and distortion of the signal.
This can also be seen in single sideband operation when mic inputs are over driven and ALC compression is applied- regardless of power output.
So the issue isn't the accuracy of the waterfall relative to gain but the
evidence of the widening of the signal- especially in consideration of the low power levels used for weak signal modes.
This is often true on the JT modes as well as WSPR. The solution is to back the gain off at the *beginning* of the audio chain and running a DB or two below maximum gain so that your input is -1db into the radio. On more complex audio chains, PC audio might make transit through a few virtual devices before entering the radio so it is important to avoid over driving anything in the chain.
Back to the "Chicken Band"! :)