Very well explained. Your point that all the energy is still there in the leakage is really important and something I learned years ago, by summing the leakage terms. Very important when trying to get a power estimate. I thought the graphic of why you get the leakage when you piece together data which is not synchronized with the sampling was wonderful, illustrating the high frequency components when you piece them together. Thanks.
Hello First, thank you for the content. I really learned a lot from the video! I have a question. Considering industrial applications (machinery in general, on the field), we gonna be obligated to always use the window in our signals right? It should be quite impossible to control exactly the start and end point when we take our sample, mainly when the system is triggered by some kind of sensor data, like the driven motor current.
It is not necessary to use a window if the signal is measured in a way that does not require it. For example, some devices might output a trigger in addition to the signal (for a example a trigger in the same spot in a rotation) and that might make it so the signal can be measured in a periodic manner. Impact tests and modal tests (where a known excitation is applied that can be made to coincide with the measurement start) can also be periodic. That being said, if the signal cannot be measured periodically (probably the vast majority of signals) than a window is the safest alternative. Hope this helps!
Really helpful video - although I do have one question! If the edges of the non-periodic waves cause high frequency components when stitched together, and these high-freq components cause leakage of higher frequencies into the spectra, how does this explain leakage from low frequencies?
Hello, may I ask if the frequency resolution is 2.5hz, and if the frequency is tested at 100Hz, there will be leakage (, in this case, windows need to be added to reduce the leakage. Is that what you mean??? Thank you and look forward to your reply
Yes, that is absolutely correct. If the resolution does not match the frequency being tested, there is leakage. Windows will confine leakage to a narrower frequency range so a RMS summation can be performed to recover the amplitude.
It's weird cuz in our textbook, it says window methods can reduce spectral leakage but as a result amplify the smearing effect. I can't really find the definition of spectral smearing.
For a signal measured non-periodically, smearing is reduced by applying a window. However, a periodic signal, which would have no smearing and not need a window, applying a window actually creates smearing.
@@SimcenterTesting I realized that you used 'smeared' a lot in the video when it comes to the leakage at the demo. So smearing and leakage are effectively the same thing. I'll double check with my tutor for that. Cheers
@@SimcenterTesting If they are on spectral lines, to begin with: "applying a window actually creates smearing". Is that understood correctly? This concept is actually interesting can you reference some sources? I have no idea where to look, but this is precisely what I need to expand on.
Very well explained. Your point that all the energy is still there in the leakage is really important and something I learned years ago, by summing the leakage terms. Very important when trying to get a power estimate.
I thought the graphic of why you get the leakage when you piece together data which is not synchronized with the sampling was wonderful, illustrating the high frequency components when you piece them together. Thanks.
Very helpful. Thanks.
Thank you! Looking forward for more videos like that. Could you possibly make one on data acquisition for real time analysis?
Great Video 😊
great video! Thank you!!
Hello
First, thank you for the content. I really learned a lot from the video!
I have a question. Considering industrial applications (machinery in general, on the field), we gonna be obligated to always use the window in our signals right?
It should be quite impossible to control exactly the start and end point when we take our sample, mainly when the system is triggered by some kind of sensor data, like the driven motor current.
It is not necessary to use a window if the signal is measured in a way that does not require it. For example, some devices might output a trigger in addition to the signal (for a example a trigger in the same spot in a rotation) and that might make it so the signal can be measured in a periodic manner. Impact tests and modal tests (where a known excitation is applied that can be made to coincide with the measurement start) can also be periodic. That being said, if the signal cannot be measured periodically (probably the vast majority of signals) than a window is the safest alternative. Hope this helps!
Really helpful video - although I do have one question! If the edges of the non-periodic waves cause high frequency components when stitched together, and these high-freq components cause leakage of higher frequencies into the spectra, how does this explain leakage from low frequencies?
The Fourier transform of a sharp transient creates a broadband signal that covers all frequencies from low to high.
Hello, may I ask if the frequency resolution is 2.5hz, and if the frequency is tested at 100Hz, there will be leakage (, in this case, windows need to be added to reduce the leakage. Is that what you mean??? Thank you and look forward to your reply
Yes, that is absolutely correct. If the resolution does not match the frequency being tested, there is leakage. Windows will confine leakage to a narrower frequency range so a RMS summation can be performed to recover the amplitude.
@@SimcenterTesting
I think I see. Thank you very much
Hi could you explain how the window function affects the spectral smearing? thx
It's weird cuz in our textbook, it says window methods can reduce spectral leakage but as a result amplify the smearing effect. I can't really find the definition of spectral smearing.
For a signal measured non-periodically, smearing is reduced by applying a window. However, a periodic signal, which would have no smearing and not need a window, applying a window actually creates smearing.
Smearing And leakage are being used interchangeably here
@@SimcenterTesting I realized that you used 'smeared' a lot in the video when it comes to the leakage at the demo. So smearing and leakage are effectively the same thing. I'll double check with my tutor for that. Cheers
@@SimcenterTesting If they are on spectral lines, to begin with: "applying a window actually creates smearing". Is that understood correctly?
This concept is actually interesting can you reference some sources? I have no idea where to look, but this is precisely what I need to expand on.