Good article, and also would like to see more.
Years ago I ran into a problem with noise from an isolated 48V to 5V converter into a fiber optic receiver, but only when the card was placed into a metal card cage. On the bench alone it worked fine. The local ground noise was common mode in the optical receiver PIN diode to the external transimpedance amplifier connection on the bench, but the metal optical receiver housing was physically installed on the card front faceplate. Placing the front faceplate at earth (chassis) ground capacitively quieted the local ground, which then caused a 3dB worse bit error rate. We ended up having to mount the optical receiver diode on the pcb and run an optical pigtail to the front faceplate connector.
On a subsequent optical receiver I anticipated the problem and used a pigtail in the initial design as well as including mounting holes for a shield for the the optical receiver diode. The shield was needed, the card was stand-alone in a metal box and installing into the box caused similar converter noise problems. All that was needed was a square inch of metal on a couple standoffs at pcb ground to hide the photodiode and transimpedance input from the box.
Very interesting - but the enlarged drawing links appear to be broken on this page. Second the request for more similar articles, this is first-principle stuff so has wide applicability beyond DC-DC converters.
The inductor can be made with minimum parasitic capacitance. The bobin of the inductor to have corrugated sections and the wire is wound in these sections will reduce the parasitic capacitance very much. In turn can minimize the ground noises.
@PierreBTOL Loved this article! This other one was also useful in tips for common mode noise:
Measurement and Filtering of
Output Noise of DC/DC
Great article, I would like to get a PDF copy of it, but the print link does not seem to work.
To add to comment of agk re: winding inductor with low parasitic C, another trick is to have the turns at the ends of the winding (near the connection terminals) spaced further apart compared to turns in the middle of the winding. This is an old trick used by ham-radio builders, and can result is a smaller inductor than would result from spacing all turns the same distance apart, but of course the parasitic C is a bit higher.
There is some system bug which is preventing the "enlarge" of Figures 5 and 6 from appearing properly. Here are the direct links for those figures:
http://www.eetimes.com/ContentEETimes/Documents/Schweber/C0893/C0893-Figure5.pdf AND http://www.eetimes.com/ContentEETimes/Documents/Schweber/C0893/C0893-Figure6.pdf