Single Pixel Output Waveguide Return Loss
February 24, 2009. Steven D. White, Program Manger. I have agreed to purchase the isolators and accepted the cost and schedule impacts for the improved return loss associated with the sliding waveguide.
February 24, 2009. Matthew A. Morgan, Project Engineer. I agree that the proposed isolators are the best and safest way to eliminate the amplitude ripple.
Tests and Model
During single-pixel lab tests, a gain ripple with period about 270 MHz with amplitude approximately 3 dB was found to be present in the dewar path. Investigations found that the ripple is caused by mismatch loss between the LNA output port and the dewar output circuitry: the waveguide telescoping section and the vacuum window. The distance between these two elements (~16 inches) corresponds closely to the 270 MHz ripple period, when the waveguide group velocity is taken into account. The file attached below shows results of VNA testing of the system and sub-assemblies. A waveguide isolator was temporarily inserted following the LNA output port for test, and the results show that the isolator virtually eliminates the periodic gain ripple, as expected. Also included in the data are measurements of the vacuum window and sliding waveguide prototypes separately and together. The measurement on the single pixel assembly seems a few dB worse than expected based on the prototype measurements. All the tests shown here were done with the system at room temperature.
To see if simple adjustments to the telescoping waveguide design could improve the return loss, a simplified CST model was built of the sliding waveguide assembly, and the results are shown in the attached file. First, the as-built assembly was modeled, then the lengths of the slide socket cavity and the taper section lengths were swept, and finally an optimization run was done to optimize these two lengths simultaneously. Basically, the model predicts about -13 dB RL for the as-built assembly when cold, and this could be improved by 2-3 dB with slight adjustments of the lengths.
For reference, below is a plot of the theoretical mismatch ripple peak-peak amplitude due to two mismatches connected with a lossless transmission line. The point identified "5 and 10 dB" represents the current case of LNA and as-built output circuitry, and the point identified as "20 and 10 dB" represents the expected ripple if an isolator is inserted close to the LNA output. The third point indicates the possible further improvement if the dewar output return loss could be improved to 20 dB. This is not likely based on the modeling above and the measured IDM input RL.
Summary & Recommendations
Based on the tests and calculations shown above, we recommend that cooled isolators be installed close to the LNA output ports. The test data discussed above shows that this step produces a significant reduction in ripple amplitude. In addition, modeling shows a small improvement in the waveguide telescope section return loss can be obtained with slight adjustments to the design dimensions, and this will be done. It does not seem necessary to put further effort into the vacuum window design.
The CDR committee expressed concern about the stability of the receiver due to the waveguide telescoping section. To address this an improved total power monitoring system is being assembled for further tests and for use in production of the 7-pixel receiver.
R. Norrod and B. Simon -- 12 Feb 2009