Hmm, the first video demonstration really does not look abnormal at all, when I replicated it required taking the audio well into the crunch - I had a different track configuration, I recorded into the first track the audio from the second or the cued second, src3 light on the second was not lit when I looked, just froze on the first track ! It’s possible my source audio was internally sampled from ins which applies a 12dB cut to a unity signal, so my levels may have been lower than yours especially if you normalised or imported normalised audio into the feeding tracks the plot thickens !