0.1 Design of an in-vivo radiation measurement scheme using a  (Page 6/7)

To store the results in a spreadsheet file,‘Save data’knob needs to be turned on. By default the data is stored in a file‘radiation_rfdata.xls. After collecting enough of data into the spreadsheet, process can be halted by clicking on the same knob again. Green LED indicates that the data can be saved to a file and a red LED indicates that the saving is in progress. The spreadsheet file can be opened to examine the contents. There should be two columns of data, with a measurement point on each row. The column values are Time (sec) and output voltage read from the RF receiver DATA line (V).

The data is collected for different input signal frequencies (50Hz, 80Hz and 100Hz) and different radiation sensors with varied dosages (100CGy, 1000CGy, 3000CGy and 5000CGy). Each time after saving the data to a file, it is renamed to radiation_rfdata_<freq>_<dosage>.xls’to further use them in calibration.

7.2 Data Analysis:

After saving the data to a spreadsheet file, it is analyzed to understand the radiation effects on MOS transistors. As explained before, due to the changes in sensor (inverter) characteristics with radiation, the On-time of the output signal should decrease and the Off-time of it should increase with a corresponding raise in radiation dosage. The lower the input frequency, the higher the difference in on and off times of the output signal (with increase in radiation dosage). So, the input frequencies are chosen as 50Hz, 80Hz&100Hz. This frequency cannot be lowered further because of the limitation of data slicer in RF receiver which works better for creating the squared waveforms from 100Hz to 28 kHz. So, this introduces noise for the signal frequencies 50Hz and 80Hz.

Using Matlab the signal noise based on its spectral characteristics was eliminated. First, the data is read from a file which is obtained from the LabVIEW program. The 512 point FFT is performed on the signal to get its power spectral plots. The plots show that there is a DC component and few unwanted high frequency signals in the spectra. So, a notch filter centered at 0Hz frequency (with a very low bandwidth) is used to eliminate the DC component. Also, a low pass filter with a cut-off frequency of 1000Hz is used to remove all the undesired high frequency components. This cannot be reduced below 1000Hz, as it causes distortion to the signal. The actual and filtered signal plots are illustrated in Fig. 13. Also, a sample power spectrum plot for a 100Hz input signal is shown in Fig. 14.

Fig. 14

After filtering, few spikes still exist in the low frequency signals. The filtered signal is converted to a square wave (from 0 to 5V) based on minimum and maximum limits of 1V and 4V respectively. In this square wave, the times of first zero-to-one and one-to-zero transitions are measured in the non-spike regions. The averages of On and Off times are taken over the entire signal region to get a valid estimate of these values. An approximated square waveform is shown in the Fig 15.