From: Paul
Date: Fri, 29 Mar 2002 00:35:51 +0000
Subject: [TSSP] Tesla Coil Mode Analyser
Tcma can now be said to be working, as it is achieving the maximum theoretical accuracy on test data, eg on a noise-free set of 4 frequencies: PK FREQ kHz (Error +/-) Q FACTOR (Error +/-) LEVEL 1 82.002 (0.01%,9Hz) 229.83 (0.09%, 0.2) -0.1dB 2 330.006 (0.01%,37Hz) 166.03 (0.02%, 0.0) -17.9dB 3 520.000 (0.01%,58Hz) 120.02 (0.04%, 0.0) -24.8dB 4 749.994 (0.01%,87Hz) 103.06 (0.09%, 0.1) -33.0dB Accounted for 100.00% of input signal and on real trace data averaged over 16 shots, eg 3-26 TEK00008.CSV PK FREQ kHz (Error +/-) Q FACTOR (Error +/-) LEVEL 1 229.951 (0.02%,36Hz) 268.92 (0.07%, 0.2) -0.1dB 2 578.068 (0.09%,501Hz) 138.21 (0.70%, 1.0) -16.2dB 3 903.912 (0.49%,4402Hz) 105.65 (6.04%, 6.4) -22.1dB Accounted for 99.73% of input signal I'm fairly sure that the residual Q errors are due to trace noise, Most of it is due to quantisation noise: if I quantise the noise- free test data so that it has the same dynamic range as the scope vertical resolution, we get: PK FREQ kHz (Error +/-) Q FACTOR (Error +/-) LEVEL 1 82.002 (0.01%,9Hz) 228.43 (0.17%, 0.4) -0.1dB 2 330.006 (0.06%,214Hz) 165.93 (0.81%, 1.3) -17.9dB 3 519.989 (1.54%,8007Hz) 118.25 (6.98%, 8.3) -24.8dB Accounted for 99.73% of input signal and tcma is unable to discern the 4th component reliably. The pattern of Q and F errors is almost identical to that of the real avg-16 trace data. This strongly suggests that averaging over 16 shots is sufficient to achieve max accuracy, and that further accuracy is limited by the vertical resolution (512 steps) of the scope. If anyone wants to have a go at compiling this, there is a draft copy of the code in http://www.abelian.demon.co.uk/tssp/tmp/tcma.c Compile with cc -o tcma tcma.c -lm Run by piping the trace data file into the program, eg ./tcma < TEK00000.CSV If you just want an FT of the trace, do ./tcma -ft < TEK00000.CSV > tek00000.ft.csv The program does not yet handle closely coupled resonators. -- Paul Nicholson, --
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.