One trouble with this LF narrow band stuff is that what has served for years as an excellent absolutely calibrated noise source for LF to HF, a 31 stage linear feedback shift register pseudo noise generator clocked at 64MHz, has its lines too far apart.

I had to build another one using a 39 bit long sequence clocked at 40MHz to get narrow enough

The attached file (hoping it makes it though) shows a waterfall plot of the old one, with its drifty clock and the resulting sidebands at 30mHz apart (33.6s repeat)

Just below that in the waterfall plot is the new one which has sidebands 73uHz apart (3.8 hours)

And waiting for those to become visible would need an FFT with a bin size of around one reciprocal-day

'JNT