.: :.


There are two main parts to our experimental apparatus. One part is the hardware all the physical components in the experiment, for which a diagram can be found in Appendix A. The other part is the software, which is much more complicated since we made the program allow for two methods for analyzing the data, direct correlation and FFT. The software also not only takes all the data, but runs all the fitting and math to produce the result.

The physical apparatus isn't as sophisticated as the software, yet has some important aspects that make the experiment run more efficiently. In previous Dynamic Light Scattering experiments, many students used two mirrors to align the beam. We found this to be a waste of resources and space. Since the laser itself can move up and down there is no need for a mirror to do this. The converging lens is used to very precisely focus the laser beam to a point at the center of the sample cell. There is a black body holder around the sample cell that helps to absorb light that is not coming from the right places. The scattered light then reaches the photo detector at a 90 angle and is further defined using the iris. We placed the photo detector far away from the sample cell, on the contrary to previous groups, to more precisely define the scattering angle. We made the beam travel in a square path as it was easiest to setup without future complications. The sample used in our initial runs is 500nm diameter spheres that are mixed in a 1:1000 ratio with Distilled water. Aaron also put together a wood box that covers the end of the experiment from any external light and absorbs any internal light. We noticed that the black cloth wasn't thick enough to block out a lot of the external light.

The signal from the photo diode is very weak so we must not only amplify the voltage, but the current as well. The signal passes through a low-noise current amplifier first and is amplified by 100 nA/V. The signal then passes through a low-noise voltage amplifier where it is amplified by two orders of magnitude and inverted as we found that the output signal on the oscilloscope was going more negative as the photo detector became active. We did a lot of positioning of our equipment to reduce as much 60Hz noise as possible. We still had a fair amount of 60Hz noise, along with the multiples of 60Hz noise, showing up in our data. We figured that a fair amount of that noise could be coming from the ADC ribbon cable as it is not well shielded like the other coaxial cables. The output of the voltage pre-amp connected to the ADC card. We set it up so that we are actually reading from two separate channels for the intensity readings for the two different (I) values in the correlation function. For our initial trials we only tired the Auto-Correlation method so on the ADC card we just made the signal run to both channels being used. This will make it easier in the future when we decide to modify the experiment for Cross-Correlation. The software mainly just follows the math behind this, also found in the Theory section. For each run we quickly sampled data points from the ADC card at a rate of 200kHz for the two channels so it was actually 100kHz for each channel. We can alter the number of data points per scan to do and we also run multiple trials.

We made the FFT method of the program to take the FFT using the Lab Windows FFT function after failed attempts to get the Numerical Recipes FFT function working. This then exports the FFT data to a file, which we then imported into Excel for analysis.

Since we did have a fair amount of noise in our system we wanted to increase our signal to noise ratio to over come this. We made a late addition to our apparatus of placing a converging lens in the scattering path after the sample cell. By focusing it on the center of the same cell we were able to focus the light, that we were interested in, back on the photo detector to increase the intensity signal. We used this for all of our data. When we performed the varying concentration tests I actually had to reduce the gain to 1 (no gain) as the voltage amplifier was over-loading from the amount of light being scattered back. We are unsure about the detrimental effects that introducing the second lens may have caused. It could be possible that it was more harm than help.

We weren't able to continue with the Auto-Correlation and Cross-Correlation because both of them require the Correlation function method to work correctly, so if one fails so does the other. Prof. Mueller explained to us how the correlation function works and reasons that it may not be giving us the right results. A couple different things may be the reason we are having problems getting this method to work. One, is that since our ADC card can only take data from one channel at a time we can't have the two channels be overlapping with their times. This presents a problem because we need the signals both to come from t=0 to get a perfect starting correlation. Another thing that could present a problem is that if you divide the correlation function by the wrong number it could generate a bias, which is what we may have seen in our results. Our results, strangely enough, produced an exponential, but at the end of our data as a positive exponential instead of a negative exponential at the beginning of the data.