Hi all, I have a signal of the following form - S(t) = A*cos(B*cos(2*PI*FM*t) + C*cos(2*PI*SF*t)) FM is the modulation frequency, and SF the signal frequency. I have written an algorithm to extract C*cos(2*PI*SF) from S(t) above. The algorithm is implemented on the C6711, and I use a 16 bit ADC to digitalize S(t). I'd like to carry out an analysis as to how accurate my algorithm is. For example, I use 3 FIR filters, 1024 in length, and when A and B above are 1, I can detect signals of amplitude C=0.01 accurately (a linear FFT shows no harmonics other than at SF). But when C=0.001, I get a slight ripple added to my demodulated signal because the filters are not suppressing the carrier harmonics enough. I would therefore like a way of expressing how accurate my demodulation scheme is for various values of C above. How might I do this? I could create a 16bit version of - cos(cos(2*PI*FM*t) + C*cos(2*PI*SF*t)) and compare it's demodulation with a floating point version of C*cos(2*PI*SF*t), over the period 1/SF. But how should I do the comparison exactly? Any other ideas for establishing the performance of my system? How might I establish my quanization noise? Again I could perform the demodulation technique on a floating point version of - cos(cos(2*PI*FM*t) + C*cos(2*PI*SF*t)) and it's 16-bit fixed bit version. And compare the two. But compare how exactly? Thanks very much for your help, Barry.
Newbie: System Performance
Started by ●June 4, 2004
Reply by ●June 6, 20042004-06-06
If I correlate - PI/200.0*cos(2*PI/80) with itself over one period, I get a peak value of - 0.009869604 If I correlate it with my signal after demodulation I get a peak of - 0.009924131 Based on this, how can I express the accuracy of my demodulated signal when D = PI/200? Their ratio is 99.45%, but I'm guessing its not correct to describe the accuracy this way. When D = PI/2000, autocorrelation gives a peak of - 9.8696E-05 and cross correlation with my demodulated signal gives - 9.65007E-05 Thanks for your help. These are my two signals, D=PI/200 and D=PI/2000 respectfully - 0.01543 0.015416 0.015309 0.015105 0.014802 0.014339 0.013782 0.013137 0.012414 0.01168 0.010877 0.010005 0.009065 0.008 0.006879 0.005715 0.004517 0.003357 0.002179 0.000986 -0.000219 -0.001492 -0.002762 -0.004017 -0.005245 -0.006374 -0.007462 -0.008506 -0.009503 -0.010511 -0.01146 -0.01234 -0.013142 -0.013797 -0.014365 -0.014846 -0.015241 -0.015612 -0.015892 -0.016076 -0.016159 -0.016076 -0.015892 -0.015612 -0.015241 -0.014846 -0.014365 -0.013798 -0.013143 -0.012341 -0.011461 -0.010512 -0.009504 -0.008507 -0.007463 -0.006375 -0.005245 -0.004018 -0.002763 -0.001493 -0.00022 0.000985 0.002179 0.003357 0.004516 0.005714 0.006879 0.007999 0.009064 0.010004 0.010876 0.01168 0.012414 0.013137 0.013781 0.014339 0.014802 0.015105 0.015308 0.015416 0.001302 0.001303 0.001226 0.001135 0.001036 0.000934 0.000891 0.000845 0.000792 0.000729 0.000592 0.000445 0.000294 0.000144 0.000058 -0.000025 -0.00011 -0.0002 -0.000358 -0.000519 -0.000679 -0.000832 -0.000915 -0.000989 -0.001059 -0.001128 -0.001259 -0.001389 -0.001511 -0.001622 -0.001657 -0.001679 -0.001693 -0.001702 -0.001769 -0.001832 -0.001885 -0.001924 -0.001885 -0.001832 -0.00177 -0.001702 -0.001693 -0.001679 -0.001657 -0.001622 -0.001512 -0.001389 -0.001259 -0.001128 -0.001059 -0.000989 -0.000915 -0.000832 -0.000679 -0.000519 -0.000358 -0.0002 -0.00011 -0.000025 0.000058 0.000144 0.000294 0.000445 0.000592 0.000729 0.000792 0.000845 0.000891 0.000934 0.001036 0.001135 0.001226 0.001303 0.001302 0.001288 0.001264 0.001235 0.001264 0.001288