I think that the lines we see in the LSC_DARM spectrogram are likely to be produced by two "beating" square waves with frequencies very close to each other. In particular, this could be the case of one square wave with frequency f0 = 27.7778 Hz which couples to a second one with frequency f1 = f0 + 6.4*10^-5 Hz. This frequency difference produces the very slow amplitude modulation, of the order of some hours, visible in the LSC_DARM spectrogram, which also depends on the frequency of the line.
Alternatively, the same effect can be produced by the superposition of two square waves with similar amplitudes, one with 55.5556 Hz and the other with 6.4*10^-5 Hz of frequency.
Further details can be found in this ipython notebook: http://nbviewer.jupyter.org/urls/gitlab.com/fdirenzo/spectrogram-tests/raw/master/on-off_lines.ipynb.
I attached one plot representing the beating of two square waves with frequencies differing by just 2% (exagerating the described effect) , and the spectrogram of two beating square waves with f0 = 27.77 Hz and |f1 - f0| = 6.4*10^-3 Hz, that is two orders of magnitude "faster" than what we see in the LSC_DARM spectrogram. As you can notice, the spectrogram I reproduced is very close to the LSC_DARM one (besides the 100x time squeezing).
Therefore, I wonder if there exist two digital signals with approximately the same frequency, 27.7778 Hz, that could couple and affect the DARM signal?