Excellent presentation. Would it be possible to expand upon the relationship between time resolution, frequency resolution and sampling rate?
Thank you for the positive feedback. We'll discuss your proposal internally. In case you'd need more detailed training, feel free to contact our Academy team directly via email at firstname.lastname@example.org or by phone +4961518038061.
Best regards ,
Head of HBK Academy
Thanks for this demonstration. It is very well done and gives a common sense understanding.
Thank you for the positive feedback.
Very good presentation on the basics of both subjects.
Thank you for your positive feedback.
I think the Nyquist/Shannon theorem is incorrectly referred to in the video. It is claimed that only the frequency information is retained. However, the signal can actually be completely reconstructed with an appropriate reconstruction filter. It's just that you didn't perform this reconstruction. Also, you mention selecting fs=200 Hz for a 100 Hz signal, but fs should be higher than 200 Hz according to the common understanding of the theorem. Moreover, at the end of the video, it is claimed that it is wise to capture without any low-pass filter in order to retain all the information. Well, that is dangerous unless you are certain that the signal does not contain any energy above ~fs/2. It was however better explained earlier in the video, when selecting a very high sampling rate. Even then, using an appropriate anti-aliasing filter for the high fs is good practice as you can never know if there are alias components from some kind of interference, etc. I think the video boils down to "for practical measurements, if you want to look at the shape of the time signal as captured by individual sample values and without applying signal reconstruction, then select a high sampling frequency". The second part of the video showing trade-offs between Bessel and Butterworth were nice examples.
Thank you very much for your interest in our ExpertTalk. Your comment shows me that you have a great deal of knowledge about digitization. Unfortunately, this is not the case with the "normal" measurement technician. As you have correctly noted, after digitization the signal must be sent through a reconstruction filter. But in practice this is hardly ever done.
Most measurement engineers have heard of the Shannon Nyquist sampling theorem, but without knowing the details. Therefore, I have very often experienced that it is applied incorrectly.
Therefore, my philosophy is to give the metrologist a recommendation that will help him to get a correct digital result. This is the oversampling factor 10 to 30 above the highest relevant frequency contained in the signal. This is the reason why the presentation in the video is so strikingly chosen.
The frequency I had chosen in the experiment was 98Hz and therefore the sampling rate of 200Hz is correct according to the sampling theorem. In the original version I had also talked about a little less than 100Hz. This is lost in the voice over.
Measuring without Anti alias filter is always a risk. To measure with the "without filter" variant I suggested, you have to know the signal well or be able to estimate it well. For getting to know completely unknown signals I have made a suggestion in the video.
When talking about digitizing, it is important to me that the measurement technician knows the properties of the filters in order to be able to adjust them in the best possible way.
I am looking forward to discussions of this kind, where you always get to know the point of view of the other.
Hi Thomas. Thanks for your reply. I understand some things were lost in translation for the voice over, that's a tough one to handle for you as presenter. I seems we are in agreement that "for practical measurements, if you want to look at the shape of the time signal as captured by individual sample values and without applying signal reconstruction, then select a high sampling frequency". /P
Great , Helpful Lecture, with practical + Handy + Clear Demonstration...Thanks for this for everyone.