Sometimes this issue is a make or break issue for a device.
A good example is suppose I want to use an RTLSDR as a direct sampler, a use for which its not designed. To do this I have to make a connection to the unused pins 4 and 5 on the RTL2832 chip which is a differential input. Elsewhere in the circuit there are bypass caps, some are polarized but other caps are not. because of the small size and presence on the same board of a DC-DC converter, noise is a big problem and hard to get rid of. Bypass caps are the logical way to approach it but the optimal size and placement of those caps makes a big difference, i found, and one could likely spend a lot of time experimenting trying to reduce the noise by another db or two.
Visually, however, if you can see the noise in your received signal you can do it interactively.
This reminds me of another tip which I will put at the bottom..
Is this noise something tangible, that is worth to consider?
In that case isn't necessary a conductive cover for the area in risk?
Sometimes, a conductive cover would/does help, definitely. But usually just doing small things is enough.
In many RF and audio applications you definitely want to reduce the tendency for anything attached to the high side of the circuit to increase noise, which it will definitely do if it can act like an antenna. Depending on the application, using the side of any capacitor that is more at ground potential as the ground even if its not marked as such can make anything from a small to a fairly substantial difference.
(This only applies to through hole parts, pretty much.)
The adoption of SMT reduced the problem substantially by shrinking the size of the parts and most importantly, reducing the size of the potential 'antenna'/ground loop.
Actually, if you want an extremely sensitive noise detector, for minimizing ESD pickup on a board, an RTL2932 chip in direct sampling mode, connected to a loop probe is extremely useful, and the cost is almost nothing, under $10. Somewhere in my tons of junk I have an earlier RTLSDR that I had dedicated to direct sampling but set up as a loop probe.
But - arrgh, I cant find it.
Its just a very small loop, a short length of coax and some ferrites between loop and dongle nothing more. The VHF/UHF tuner is disabled.
Its really a useful thing to have for finding broadband noise. I strongly encourage people who are designing PCBs to check this out, its much more informative than a scope probe. It works more like a spectrum analyzer.
The cheap dongles from ebay might cost $10 or even less and be a snap to modify to make a signal probe thats going to pay for itself many times over in hunting down ESD sources on boards like one would do with a spectrum analyzer.
it just is basically a sensitive USB probe that picks up the lower 20 MHz or so of spectrum.