One day I got test failure from older tester colleague. He reported to me:
„Power converter is not stable. Look at fluctuations of output ripple”
He was not aware that I selected buck converter with Duall Random Spread Spectrum to minimize electromagnetic noise peaks and to relax filter specificaiton for externaly connected device.
Take more measurement and look closer
Oscillogram
Comments
Zooming out to see envelope. There is possible barely see triangular changing envelope.
Zooming further out and adding FFT we can see low frequency “noise” component frequencies. We can see about 9, 18kHz components.
Tweaking point of triggering and time basis we can see modulating signal.
From theory or operation that certain DCDC converter we can clearly see triangular waveform with random square modulation at the top of it. This exactly matches with previous oscilloscope observations.
Rationale
Typical low frequency conducted noise measurements are performed in range 150KHz up to 30MHz with RBW=9KHz. What we see is about 9KHz triangular modulation aming to ‘smear’ energy captured by 9kHZ RBW what finally reduces its peak value.
From previous design notes we know that in acusstic range capacitors can generate audible noise. The same 9kHz modulation frequency is also not sufficient at higher frequencies above 30MHz where RBW is increased up to 120kHz. Solution to both probles is second modulating signal. Its pseudorandom digital wave.
Magnetic field spectrum above converter
Capturing magnetic field spectrum above converter we can see that energy peaks are wider than for typical continous condution mode converter.
Conclusions
There are much more specific DCDC converter on the market which aim into different problems. For example this converter should not be used directly for powering SDR radio modules.
Stability of those converters can be quickly guessed with load step response.
Not allways what you would classify in typical way as faulty really faulty is.