Actions
Engagement
  • Author Author: tariq.ahmad
  • Date Created: Date Created
  • Last Updated Last Updated:
  • Views 13078 views
  • Likes 6 likes
  • Comments 7 comments
Related
Recommended

How to Avoid Oscilloscope Aliasing Pitfalls for Accurate Measurements - Workbench Wednesdays 78

In this episode of Workbench Wednesdays, James dives into the world of oscilloscopes and the Nyquist theorem, shedding light on the often-misunderstood concept of aliasing. He begins by explaining the Nyquist-Shannon theorem, which states that to accurately reconstruct a signal, you need to sample it at least twice as fast as its fundamental frequency. However, James points out a crucial detail – the theorem requires sampling faster than two times the signal frequency, not exactly two times. 

James demonstrates this with a sine wave example, showing how slight shifts in the sample points can lead to incorrect amplitude in the reconstructed waveform. He then introduces aliasing, where the sample rate is much slower than the signal, resulting in a fake, lower-frequency waveform. This is an important concept for makers and engineers to grasp when working with digital oscilloscopes. 

 image

Moving on to practical examples, James illustrates how changing the sample rate affects measurements. He shows how a 100 MHz sine wave can appear as 20 MHz on the scope when the sample rate is insufficient, but is accurately represented when the sample rate is increased to 400 Msps. He emphasizes that understanding the sample rate is vital for accurate measurements and troubleshooting. 

 image

James doesn't stop at sine waves. He explores more complex shapes, such as square waves, and demonstrates how they can lead to aliasing when the sample rate drops below the Nyquist frequency. He also briefly discusses FFTs on digital oscilloscopes, highlighting limitations in frequency span. 

 image

To avoid aliasing, James delves into the oscilloscope's memory depth, explaining how it's connected to sample rate and time-based settings. He highlights how some oscilloscopes automatically adjust memory depth and sample rate to keep the screen filled. James also points out that interleaved ADCs may alter the sample rate when channels are enabled or disabled, underlining the importance of knowing the sample rate. 

image 

In conclusion, James provides practical advice on determining the required sample rate for different signals. He recommends sampling at least 2.5 times faster than the fastest frequency in your signal, with square waves requiring additional considerations. He advises keeping an eye on the sample rate when adjusting the time base, as significant changes can indicate aliasing issues. Finally, he encourages viewers to seek help on the element14 Community for any oscilloscope-related questions and provides a list of his oscilloscope-related videos for further exploration. 

Bonus Content:

Bill of Material:

Product Name Manufacturer Quantity Buy Kit
MXO4 - 4 Channel Oscilloscope 200 MHz Rohde & Schwarz 1 Buy Now
Bench Oscilloscope - 200 MHz, 1 GSa/s, 14-bit, AFG, DMM Multicomp Pro 1 Buy Now
MP720782 Oscilloscope, 2 Channel, 70 MHz, 250 MSPS, 6 kpts, 5 ns Multicomp Pro 1 Buy Now
PicoScope 2204A - USB 2 Channel Oscilloscope PicoScope 1 Buy Now
MDO3014 - 100 MHz 2.5 GSa/p Oscilloscope Tektronix 1 Buy Now
Arbitrary Waveform Generator - 2 Channel, 160 MHz Multicomp Pro 1 Buy Now
 

  • in reply to mharpaopao

    Power supplies are generally measured from DC to 20 MHz.

    The switching frequency is almost irrelevant. The concern is the speed the switch turns on and off. The switcher's edge energizes the inductor which creates all the high-frequency content (and EMI).

    That said, 1.25 Megasamples/s at 500 kHz of bandwidth is 2.4 times oversampling. Personally, I'd prefer to sample faster than that, but it should be enough for that bandwidth. However, I do not think that's enough bandwidth to evaluate an SMPS properly.

    Are you really using a 100 to 1 probe? That's often a poor choice for SMPS since you're heavily attenuating the noise.

  • imageHello. I would like to ask what bandwidth and sampling rate should I use when meausing a Vrms and Vp-p if I have a display like this one. It is an output from a SMPS having a frequency of around 25kHz. My measured Vrms in the photo at a bandwidth of 500kHz is 4.06259V which increases  to around 4.35V to 4.6V everytime I iincrease the bandwidth. Hope you can help me. 

  • in reply to DAB

    I can imagine!

    I've been involved with high-bandwidth digital oscilloscopes (in some way, shape or form) since 2001. I'm not sure if there is any other test tool that has evolved the way scopes have.

    Sure spectrum analyzers have gotten slightly wider bandwidths and power supplies reduced their noise floors and stuff like that.

    But if you look at a digital oscilloscope from 1980 to 2000 to 2020. They are completely different instruments to effectively make the same measurement.

  • Another great episode.

    I worked with a pre-digital oscilloscope where I had 32Kbytes of memory and an 8 bit A/D.

    We captured radar signatures.

    I wrote the software and displayed various information about the data collected.

    Needless to say, things have gone much further during the four decades when I was doing these things.

  • By the way,   , you mentioned this topic when I posted the list of oscilloscope tutorials.

    Thanks again for the suggestion!

  • This topic is one I have wanted to cover for a long time.

    When I was an oscilloscope FAE (and still had hair), I remember one of my mentors, Art Porter, telling me about Shannon. Art always made it a point to talk about the Nyquist-Shannon theorem. Later, when I was looking for their actual papers, that is when I discovered Whittaker had laid the foundation for their work.

    It is crazy to me to think about how these ideas were so well developed in the first half of the 1900s--a time when the US was still working to put electricity in every household.