Previous Section: Results
Next Section: None

Discussion (TOC)

As explained in section 3.1 a radiation source was successfully constructed. Not only was the source functional, but it was able to meet all the power constraints of the MISI, as demonstrated in section 4.1. As for the actual construction, one might question the geometric placement of the thermoelectric heat pumps. This partially gets into a uniformity issue in that optimum spacing may control uniformity. The heat pumps were placed in such a way so as to have equal distance from one pump to another. This distance was not evaluated in that there were no experiments performed to optimize the dimensions. It was assumed that the current layout would be acceptable and other possible layouts could be implemented in the near future.

The control feedback system was also successfully constructed as described in section 3.2. It was noticed, however, that the operational amplifier in the control system was not able to scale (amplify) the temperature signal to its full dynamic range. This was mostly due to noise issues. Even at an amplification of 40, the signal was plenty large enough to work with. The signal did, however, required some filtering but for the most part the transducer component supplied an adequate signal for the rest of the system to work with. The sensors response was not that much of a concern because the mass it was sensing had a very long response as described in section 4.3. Though its response was not an issue, its placement was. The sensor was placed at three locations across the plate. Namely, the edge, center, and opposite edge. All three points formed a diagonal line across the copper plate. It was found that the plate was able to be controlled with the same tolerance in all three positions.

As for the sampling, the RTI board had a 12-bit resolution which seemed to be sufficient. The voltage it was sampling was only changing by six volts or so. This delta in voltage came from the amount of amplification. Had there been more amplification, there would have been a larger dynamic range. The full dynamic range was set up such that the analog inputs could receive a voltage from 0 to 10v. On the other end was the task of converting the digital signal back to an analog one. This was performed sufficiently with the use of one 8-bit A/D chip as described in section 3.2.4. The characteristics of this chip also seemed to be adequate with a resolution on the order of 0.039 volts/digital count. It is believed that the systems response was not capable of detecting such a change in voltage. In that light, 8-bits proves to be more than enough for the time being. Finally the system was "closed" with the addition of the programmable power supply. This behaved with a fast response time so concerns about it slowing the overall response of the control system were dismissed.

It was noticed that the blackbody was able to reach a temperature (45°C) beyond the usable range with a small input voltage. On the other hand, with the same input voltage reversed biased (cooling mode), the blackbody could only reach a value of 10°C. It is desired to cool the source below the current value it can reach in cooling mode. To do this may require additional voltage which may jeopardize the constraints of the system.

The uniformity of the blackbody was assessed qualitatively as well as quantitatively. The Inframetrics images proved to be useful in detecting variations across the surface due to changes in grey scale. With a little image processing, these variation could be enhanced and additional conclusion could be drawn. Generally the discrete pattern of the heat pumps found at start up time disappeared as the plate came to equilibrium with its surround. The blackbody constructed in this research had a very similar thermal pattern to that of the laboratory standard. To match the profile of the two more closely, the constructed radiation source was insulated. The results of this were best analyzed when 1D slices were taken across the plates surface. Basically it is hard to determine if insulation really improved the overall uniformity because of the noise factor. When the over all profile was looked at across the entire surface, it looked like there was improvement in uniformity. When the center was looked at, results indicated that the constructed blackbody was better in some areas than the laboratory standard. At first this is hard to swallow considering the laboratory source used in all these experiments costs about $15k and the constructed source cost is under $1k. This may be the case though. Non-the-less, the constructed blackbody seems to be very similar to the laboratory standard in that they both had an average temperature variation across the surface of about 0.5°C.

The control algorithm for this research proved to be satisfactory, though not conclusive. The algorithm implemented was very elementary in its design and approach, as shown in Figure 4.18. It did, however, regulate the source to with in 0.3°C. This is actually high compared to microprocessor controlled systems, which can regulated to 0.01°C, but not that bad for the on/off type of algorithm it was. This on/off switching does pose a problem for other parts of the MISI system however. The on/off nature of the algorithm can cause high RF noise that other system components will detect. Still, the regulation of the plate was not the thrust of the research, knowing the temperature of the plate accurately was. So any accurate regulation results were an added dividend.

 

 

Conclusions and Recommendations (TOC)

In summary this research showed that a blackbody radiation source can be constructed and controlled with a feedback control type system. Further more, the uniformity of such a source was shown to be around 0.5°C across its surface, which was comparable to a laboratory standard. Temperature knowledge of the source was believed to be close to 0.2°C which was as good as the device performing the evaluation (the Inframetrics IR camera). As for the controlling of the source, it was determined that with a simple thermostat-type algorithm, the source could be controlled to at least ±0.15°C.

Two of the main suggestions for future work are to apply multiple sensors across the surface of the plate to improve temperature knowledge and use a PID type control algorithm that functions for any ambient and user defined set point. The first suggestion is a common sense one in that the more sensors you have sensing the plate (at various locations), the more believable your results will be. To optimize this a series of thermistors, placed across the surface, should be compared to a solid state(s) temperature sensor with an IR camera evaluating the temperature. The second suggestion is not as important as the first in that absolute regulation is not essential. The regulation is not that crucial in that the MISI only needs to know what the temperature is at any given time. The overall temperature of the source only needs to fall with in the MISI’s dynamic range. For that, an on/off algorithm proves to be sufficient. As a final note, only heating mode data was used to assess the blackbody. Future experimentation would entail collecting and evaluating the data for the cooling mode case.

 

References (TOC)

7)Gong, J., Hornak, J., "A Fast T₁ Algorithm," Magnetic Resonance Imaging, Vol. 10, pp. 623-6, (1992).

 

 

Previous Section: Results
Next Section: None