Understanding Oscilloscopes - ENOB

[@bellkidsbasketball8889]

Another great video. Simplifying a complicated topic.

[@pauldenisowski]

Thanks!

[@terrysteven3528]

Doesn't the viewing screen affect the ENOB. A 1920x1080 tft screen for example can only be sliced 1920 times so wouldn't this limit the number of effective bits to around 11 ( 2 ^11 = 2048).

[@zynthos9]

How would one estimate noise and distortion power? I expect noise power would mainly be a function of ADC noise figure and additive noise and of VGA noise figure? And distortion power would be a function of mainly VGA's IP3 and IP2 ratings (assuming the level isn't high enough to drive into compression/saturation)? To me this only makes sense if you assume multiple signals present

[@pauldenisowski]

Those are all great questions, and I could probably do another entire video answering them. :) Although some components can be measured in isolation, ENOB in oscilloscopes is very much a "system" measurement, and the whitepaper linked to in the video description goes into detail about how ENOB is measured. The IEEE also defines (1241-2010) methods and definitions for measuring or estimating ENOB. You're absolutely correct that the NF of the ADC and VGA play an important role in ENOB, and even things such as signal routing will have an impact on overall noise performance. Clock signal jitter also leads to reduced SNR in the conversion result, so the quality of the oscilloscope clock or time base will have a (potentially significant) impact on ENOB. With regards to distortion: ENOB is often strongly a function of spurious free dynamic range (SFDR). You might be surprised by how many spurs some ADCs produce. Avoiding compression (IP3) is a lot easier than removing spurs from an ADC. :) And keep in mind that unless the input to the oscilloscope is a very pure sinewave, the measured signal will always have multiple frequency components -- the classic "square wave is the sum of odd harmonics" is a good example of this. Hope those help. Please take a look at the application note linked to in the video description if you'd like more information about how ENOB is calculated. And thanks again for the great questions!

[@pauldenisowski]

There's a (very brief) discussion about ADC's and noise / distortion - including a graph of SNR as a function of frequency and jitter - in another one of my presentations: kzbin.info/www/bejne/o6GvZ5ycrd2qr80

[@joelw781]

really well done

[@pauldenisowski]

Thanks!

[@paulosanchesgoncalves5567]

GOOD JOB

[@pauldenisowski]

Thanks!

[@paulosanchesgoncalves5567]

GOOD VIDEO

[@pauldenisowski]

Thanks! Hope it was helpful!

[@lixiang7349]

For god sake, people need to stop saying "and/or". If you do a logic "or" between "and" and "or", you effectively get the logic "or", assuming "/" means the logic "or". So "and/or" is the same as a simple logic "or". Try analyzing this with venn diagrams and you'll know what I mean, please.

[@pauldenisowski]

(laughs) Point taken. Usually when I say "A and/or B" what I mean is the logical equivalent of (A ∧ B) ∨ (A ⊕ B). For example, when I say in the video that noise from the VGA "may be frequency and/or level dependent", what I mean is that in some cases, the noise is a function of both frequency AND level (A ∧ B), while in other cases (∨) the noise is caused only by frequency or level but NOT both of them (A ⊕ B). Frankly, I would say that the vast majority of VGAs meet the (A ∧ B) condition, but it is possible that VGA noise is entirely a function of frequency, particularly if it is never pushed into compression. But as someone who studied languages before becoming an engineer, I can very much appreciate the desire for precision in both fields :) Thanks for the comment!

[@lixiang7349]

@@pauldenisowski hmm actually I’m speaking from a pure discrete math perspective instead of which language the logic is conveyed in.