| Electronics > Projects, Designs, and Technical Stuff |
| Low frequencies Vector Network Analyzer, arduino based. |
| << < (13/20) > >> |
| Bud:
What i think is fft or not is the least of the worries so to speak. I think some the device architecture needs a few clarifications. What calibrations do you perform to remove systematic errors? |
| Mechatrommer:
--- Quote from: MasterT on September 23, 2018, 03:32:18 pm ---...me the Bandwidth in Hz for single bin DFT. There is no issue with pure math, consider ratio: "Bandwidth / Clock_cycles_uCPU".... I'm not argue, that single bin (6-8 math operation code) is faster than 2048 bins, it's all about bandwidth (second page in a row). For single bin Bandwidth is 250 kHz. --- End quote --- everyday i learn new thing.. today is Goertzel Algorithm if you just dont care to read here is the snip from Goertzel... --- Quote ---For covering a full spectrum, the Goertzel algorithm has a higher order of complexity than fast Fourier transform (FFT) algorithms, but for computing a small number of selected frequency components, it is more numerically efficient --- End quote --- when Goertzel said computation efficient for single or few bins, it means the same bin as we speak when we calculate DFT or FFT, the resolution for each bin will be = sample rate / number of points / 2. whether DFT, FFT, or Goertzel algorithm they are the same. Goertzel was not fooling around when formulating the single bin algorithm. nobody want to argue with Goertzel that his bin's resolution will be = (sample rate / 2) aka BW. i think the confusion happened (vocabulary) is between DFT and Goertzel. the competition of FFT is not between DFT (FFT is derivative or DFT we know that), but the competition should be between FFT and Goertzel (single bin DFT) and again, the resolution of the single bin of Goertzel will be just the same as bins for DFT or FFT. fwiw... |
| radioactive:
--- Quote from: Mechatrommer on September 24, 2018, 02:12:50 am --- --- Quote ---...and again, the resolution of the single bin of Goertzel will be just the same as bins for DFT or FFT. fwiw... --- End quote --- --- End quote --- True, I showed that in an example a couple of posts ago, but the issue in this instance is that the Goertzel algorithm is good for real data input. I think the OP needs complex input for his application (feel free to correct me if I'm wrong here). The single-bin DFT, as others have pointed out many times, should be more efficient and identical to FFT in this case. Goertzel is good for applications where you want to optimally detect a single frequency in magnitude (e.g. DTMF detection). The computational savings would be very good for N=2048... |
| Mechatrommer:
--- Quote from: radioactive on September 24, 2018, 02:26:39 am ---True, I showed that in an example a couple of posts ago, but the issue in this instance is that the Goertzel algorithm is good for real data input. I think the OP needs complex input for his application (feel free to correct me if I'm wrong here). The single-bin DFT, as others have pointed out many times, should be more efficient and identical to FFT in this case. Goertzel is good for applications where you want to optimally detect a single frequency in magnitude (i.e. DTMF detection). The computational savings would be very good for N=2048... --- End quote --- there is no mentioning that Goertzel is limited to real part only, (istr there is even faster algortihm than FFT when calculating real part only) so Goertzel should be capable of calculating both the magnitude and phase, as others linked the Impedance Measurement Techique by Keysight (i think Goertzel algorithm is used in there but not sure didnt read) its just a matter of the OP reading carefully and understanding how Goertzel formulation works, and start coding it in complex number, he got ee degree aye? i dont. he should be able to implement it in C/C++ ;) |
| radioactive:
--- Quote from: Mechatrommer on September 24, 2018, 02:32:32 am --- --- Quote ---there is no mentioning that Goertzel is limited to real part only, (istr there is even faster algortihm than FFT when calculating real part only) so Goertzel should be capable of calculating both the magnitude and phase, --- End quote --- --- End quote --- I think at the point that you implement that, it probably won't have any gains over the single-bin DFT (again correct me if I'm wrong). I used Goertzel in an FPGA application where I didn't have enough cells for FFT in the past, but I was only interested in magnitude. It worked quite well. [edit] fixed quote block |
| Navigation |
| Message Index |
| Next page |
| Previous page |