Search the Community

There is a bug in the export of data from the impedance analyzer from Waveforms (specifically from Waveforms 3.8.2 using the Analog Discovery 2, but I have seen it with earlier releases of Waveforms 3 also). The problem is that the frequencies are reported as if they were from a linear sweep, even though a logarithmic frequency sweep was done: #Date Time: 2018-07-01 16:50:47.258 #Trace: Trace #Start: 100 Hz #Stop: 1e+06 Hz #Steps: 101 #Resistor: 10000 Ω #Average: 10 #ProbeCapacitance: 1.04e+06 Ω #ProbeResistance: 4.3e-11 F #Compensation: Open Short #Wavegen: Wavegen1 #Amplification: 1 X #Settle: 20 ms #MinPeriods: 32 #Channel: Channel 1 #Range: 5.49577 V #Offset: -0.000318847 V #Attenuation: 1 X #Channel: Channel 2 #Range: 5.51181 V #Offset: -0.000280998 V #Attenuation: 1 X Frequency (Hz) Trace θ (°) Trace |Z| (Ω) Trace Ls (H) Trace Cs (F) 100 0.00067832392258167652 9971.9407134030535 0.00018789460945304077 -0.013481119008321029 10099 0.0018260127183053441 9972.4554790954317 0.00046132118972998595 -0.0045670615004842902 20098 0.0026165423173400044 9972.2098918719494 0.00060286028271848858 -0.0029068564833067186 Many of my students had problems with this bug last quarter, and we had to create a work-around in our gnuplot plotting scripts: #if the frequencies in your data file don't follow log increments use the code below # Written by Ali Fallahi #example: use plot 'data' u (correct($1)):3 instead of plot 'data' u 1:3 freq_u=1e6 #last frequency freq_l=10 #first frequency NoOfSamples=101 #number of samples indexno(f)=(NoOfSamples-1)*(f-freq_l)/(freq_u-freq_l) correct(f)=10**(log10(freq_l)+indexno(f)/(NoOfSamples-1)*log10(freq_u/freq_l)) ## plot 'measuring-10kohm-buggy-freq.txt' u (correct($1)):3 title '10k{/Symbol W}±0.1%' It would be nice not to need to correct the data files.