Report

Deutsches Insitut für Lebensmitteltechnik
Edward Carley
5/18/11

This report documents the process through which I went to find a usable setup for a pulse signal and the subsequent tests I set up in preparation for calibrating the probe to the analysis of meat. 

Initially, I thought the TGP110 10MHz Pulse Generator would supply the pulse, but the setup was somewhat complex and the analog knob for the amplitude was libel to change accidentally.  Also, the pulse it generated had secondary peaks besides the desired reflected pulse, which were probably pulse echoes from the apparatus.  Since extra signals are undesirable, it being unclear where they come from, I continued to look for other ways to generate a pulse. 

The Handyscope HS3-100 has a function generator feature, but no pulse function.  However, by setting the square function to the max 12.0V DC Offset, max 12.0 Volt amplitude, 0.2MHz Frequency, and 0.1% Symmetry, I achieved what was effectively a pulse.  The Out cable was attached to a coaxial T adaptor at the CH1 In port, leaving a long cable out the other side attached to the probe.  The Handyscope displays several measurements a second, resulting in turbulent appearing data.  To reduce random error, I set the program to perform an averaging of 128 measurements before wiping the results and I record the waveforms generated by the probe only when they are over 115 measurements averaged. 

A long cable for the probe is needed to separate the pulse from the reflected pulse and the accompanying signal.  At first I connected several smaller cables together with female to female coaxial adaptors, but concerns were raised about possible error due to cord arrangements.  It was supposed the close contact of coiled cables could create a small magnetic field, affecting the pulse.  So, as the experiment in the Excel spreadsheet ‘Cord Arrangement’ in the ‘rn’ folder of ‘Eddie Data’ shows, I set up an experiment to compare positions of the cable and the order of the individual cables. 

To determine if these differences were significant beyond random error, I created the experiment which can be seen in the nearby Excel spreadsheet ‘Random Noise with Air.’  I took measurements of air with the probe, isolating temperature and cord position, approximately every thirty seconds to see what error could be attributed to random noise.  What I found was that the probe in the most consistent environment I could create has differences on the order of 0.1% of the amplitude.  Comparatively, the cord arrangements have differences greater than 1% of the amplitude, over an order of magnitude increase. 

When I switched to the new cable, which is one line of cable with no adapters, the differences between the arrangements shrank dramatically.  While the doubled over and quadrupled over cable arrangements had differences with the straight line greater than 1%, the other arrangements had much smaller differences.  The differences between the straight and jumbled cables were on the order of 0.1%.  I used the jumbled arrangement for subsequent experiments.

I began using the probe on solutions of salt in water, beginning with the Open Office spreadsheets ‘Salt Test Range,’ ‘Salt Test Range 2,’ and 3.  I wanted to examine the reflected signal waveforms generated by high and low concentrations of salt in these experiments.  Later, I also tested how the waveforms responded to changes in temperature at different concentrations, with tapwater as a base of comparison.  I have yet to mathematically analyze these results, but it currently appears there is no interaction between the variables of concentration and temperature. 

I note here that it matters where in the liquid solution where you hold the probe, but this is not a problem as long as you are consistent.  The probe senses less salt, and the waveform rises, when it is in contact with the sides or bottom of the container.  Holding it in the middle of the solution seems to give a consistent reading, but this is conjecture and will be tested.