Averaging in an oscilloscope for measuring very low level ripple in a DC/DC converter

[alimjoco]

Hi,

Just want to confirm if the purpose of averaging in acquisition mode in a oscilloscpe is to average the absolute signal and remove the apparent and random noise. Is it applicable or right to use averaging in an oscilloscope in measuring very low level ripple in a DC/DC converter?

 

[FvM]

Re: oscilloscope average

Averaging works only for periodical signals. If the DC/DC converter isn't operating with strictly constant duty cycle or you don't manage to trigger synchronous to it's switching frequency, the measurement will be incorrect.

 

[alimjoco]

Re: oscilloscope average

I think ripple is solely based from capacitor ESR function and its switching frequency(the inductor current ON/OFF that flows on the capacitor). Thus results could be correct since ripple is periodic.

 

[dick_freebird]

I find that you need to look at ripple across a range of timescales,
and if you want reality you won't average or bandwidth-limit the
measurements. Now, if you're preparing marketing material this
may be part of "hair and makeup". But you should look at the
part as its load sees it, not how makes it the prettiest.

Ripple is periodic -if- the design is fully stable, etc. Which is a
thing to be proven rather than assumed. The best performance
to load-step comes at the edge of stability. Subharmonic
oscillations might average nicely to the target DC voltage on
the 'scope but produce significant fsw/2 ripple.

If you want the bad news, try envelope mode and then run
down the extrema.

 

[mtwieg]

If your converter is stable and the source and loads are both steady, then yes averaging mode should work okay. But of course if that really is the case then there shouldn't actually be much of a difference in how the waveform looks between normal and averaging modes (neglecting for the effects of actual thermal noise in the scope or the load). Averaging mode is specifically meant to help filter out things like noise and switching interference, so you have to be careful not to get an overly optimistic result.

 

[alimjoco]

thanks.. i think averaging is ok especially if i want to correlate the ripple spectral energy in a frequency domain like using a spectrum analyzer(i guess spectrum analyzer also averages energy per freq bin)..it really depends on what I want to measure but switching noise should not be filtered if I want to consider those things

 

[FvM]

averaging is ok especially if i want to correlate the ripple spectral energy in a frequency domain(i guess spectrum analyzer also averages energy per freq bin)

Unfortunately averaging in time domain is not necessarily averaging energy in frequency domain, because signals of arbitrary phase are cancelled out. I agree, that in the present case, it's reasonable to expect a constant phase for ripple related to a particular switching edge. But as already explained in detail, the assumption may be wrong under certain circumstances.