[time-nuts] Pitfalls of Digital 'Scopes
jfor at quik.com
Mon Jun 6 18:11:28 UTC 2011
> On 06/06/11 18:40, J. Forster wrote:
>> IMO, the lesson is that digital scopes do not always accurately depict
>> what a circuit is doing. Even a $50 analog 'scope would never have this
> Out of idle curiosity, what sampling mode were you using? (ACQUIRE menu
> on my TDS2024B).
> In SAMPLE mode, you're guaranteed to see this kind of aliasing. It's
> usually best to run digital scopes (especially the cheaper ones with
> limited acquisition RAM) in PEAK DETECT mode.
OK. It was a couple of weeks ago so I can't tell definitively, but
probably SAMPLE, with 4 WF average.
> What this does is keep the acquisition front-end running at maximum
> speed and stores the minimum and maximum values recorded in each 'sample
> bin'. You still only have 2500 data points out of the (say) 2.5 million
> the scope might have analysed, but the min/max will at least tell you
> "there's a signal here" even if the scope doesn't have the memory /
> resolution to tell you what that signal *is*.
> If you decide that the burst is interesting enough to look at (and you
> know what the repetition rate is), you zoom in (TIME/DIV and H POSITION
> controls) as normal.
> I seem to recall someone covering this in a Youtube video. Probably Jeri
> Ellsworth (jeriellsworth on Youtube or Twitter) or Dave Jones (EEVBlog).
> It's a big, nasty beartrap and one that's all too easy to fall into.
Yup. That's my point exactly.
> lists at philpem.me.uk
More information about the time-nuts