Here's the design I've implemented -- www.stubbylatheusa.com/pdf/webench_design_3519623_8.pdf. This is the supply for a low volume commercial product. I'm not really an engineer but I've had a pretty good run of successes designing working boards so...
I've used it as shown except that I have added a 100pf cap in parallel with Cout (physically close) and have a few more of them scattered around the board. I've populated several boards and the results are always the same. When I look at the 5v output, I see a spike and ringing at the switching frequency. The spike is about 200mV and the ringing settles down in about 100 nsecs. While the design says 150 ma, the actual load is probably about 35 ma.
Here's the part where I get confused. I've jerry-rigged a 100 uF cap in parallel with the 10 uF Cout and NOTHING CHANGES. The board layout is not identical to the suggested layout but it is compact, reasonable, (a 2 layer board) and there are ground planes top and bottom in the power supply area.
Everything works, everything is stable --the microprocessor, the com chip, the LCD display -- but I would certainly like to feel more confident before I commit to a production board run. I'm looking at the results with an ancient Tek 2465A scope (13pf 10x probe) but for its time, it was a good one, it works fine and I don't think it is the problem. Should I fix something, should I pretend I never saw it, ... Any help would be appreciated.
Bill