JEM/SMILES AOPT EM, Part 2 Bandpass Characteristic and Beam Pattern after Thermal Cycling

Transcription

1 JEM/SMILES AOPT EM, Part 2 Bandpass Characteristic and Beam Pattern after Thermal Cycling Axel Murk Research Report No March 2001 Institute of Applied Physics Dept. of Microwave Physics Sidlerstr. 5 Tel. : CH-3012 Bern Fax : Switzerland iap @iap.unibe.ch

2 Contents 1 Introduction 2 2 Thermal Cycling 3 3 Bandpass Characteristic at the SLO Frequency 10 4 FSP Tuning 12 5 Absolute Beam Position 17 6 Conclusions 22 1

3 1 Introduction Part 2 of the AOPT EM test report documents the bandpass and antenna pattern measurements after thermal cycling. The test procedure and data analysis are the same as described in part 1. The measurement setup was modified from the one described in section 2 of part 1 in two minor details. First, the submillimeter source was mounted in a new fixture which allowed to change the polarization angle. With this fixture no additional grid was needed between the source and the AOPT. Second, all measurements were done with the horizontal grid in the COPT simulator to reduce the reflections at the detector (see section 7.8 in part 1). For the bandpass measurements the COPT simulator now consists of SMX horn, CM2, one horizontal grid, one 45 degree grid in transmission and a second 45 degree coupling grid in reflection. For the beam pattern measurements the coupling grid was replaced by a flat coupling mirror because they are sensitive to the COPT alignment, but not to the cross-polar leakage. Section 2 contains measurements of the AOPT in the same state as during the measurements in part 1. For the measurements in section 4 the following changes were made: 1. FSP is tuned by replacing one of the Invar mirrors 2. SLO mass- and reflection-dummy is replaced by TK-RAM 3. SMI grid is activated Section 5 discusses the absolute position of the TRN beam and the connected errors in more detail. 2

4 2 Thermal Cycling A programmable thermal chamber was used to cycle the AOPT in air between temperatures of -40 C and +60 C. The chamber controls the temperature of the air flow as well as the humidity in the chamber. Initial test cycles have shown that the humidity control is not good enough to guarantee that no ice is formed on the test object at low temperatures. For that reason a constant flow of dry nitrogen was injected into the AOPT through the TRN BBH. During cycling the air temperature and humidity were monitored with sensors independent from the control sensors of the chamber. An additional sensor measured the actual temperature of the AOPT close to the COPT port. Figure 1 displays the measured air and AOPT temperatures during the cycling. The nominal qualification sequence for thermal cycling was C, 1 60 C and six cycles between 0 C and +50 C. Since the chamber only controls the air temperature, but not the actual temperature of the test object, these requirements were not fully accomplished. The achieved AOPT temperatures are 1 h C and 1 h 59.2 C with peak temperatures of 59.6 C and C. In addition, only four of the six 0 to 40 C cycles could be finished due to a failure of the thermal chamber. The beam pattern in figure 2 has to be compared with Fig. 9 of part 1. It seems that the symmetry of the first sidelobe has improved significantly after thermal cycling. This is difficult to explain because, according to section 7.6 of part 1, the asymmetry is at least partly generated by the COPT simulator which was not thermally cycled or modified between the measurements. Another important change is that the phase of the beam pattern is more tilted after thermal cycling, which corresponds to a larger lateral offset of the phase center from the rotational axis. This will be discussed in more detail in section 5. The bandpass characteristic is similar to the one measured before thermal cycling. 3

5 T [C] Air AOPT Time [h] Figure 1: Measured temperatures of the AOPT (red) and air (blue) during thermal cycling. Missing data points are an artifact of the limited storage capabilities of the different data loggers used for the temperature monitoring. The temperature variations at the end of the cycle are caused by a failure of the thermal chamber cooling circuit. 4

6 trn_h_21.dat, # 9 10, f= GHz 0 10 Gaussian Fit hpbw= 2.08 Θ = 0.10 Relative Amplitude [db] Θ [deg] Phase [deg] Phase Fit = λ axial = 1.00 λ lateral Θ [deg] pattern_trn_h_lsb_2,a. Murk, IAP, 29 Nov 2001, 19:51 Figure 2: TRN LSB antenna pattern after thermal cycling, AOPT horizontal 5

7 trn_h_20.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = 0.00 db trn_h_20.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = db bandpass_trn_h_lsb_2,a. Murk, IAP, 29 Nov 2001, 9:55 Figure 3: TRN LSB bandpass after thermal cycling 6

8 trn_h_24.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = db trn_h_24.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = 0.00 db bandpass_trn_h_usb_2,a. Murk, IAP, 04 Dec 2001, 11:36 Figure 4: TRN USB bandpass after thermal cycling 7

9 trn_h_20.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = db x 10 3 trn_h_20.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = 0.00 db bandpass_cst_h_lsb_2,a. Murk, IAP, 29 Nov 2001, 10:2 Figure 5: CST LSB bandpass after thermal cycling 8

10 x 10 3 trn_h_25.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = 0.00 db trn_h_25.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = db bandpass_cst_h_usb_2,a. Murk, IAP, 04 Dec 2001, 12:26 Figure 6: CST USB bandpass after thermal cycling 9

11 3 Bandpass Characteristic at the SLO Frequency Besides the sideband rejection the FSP filter of the AOPT also has to split the SLO power equally between the two SIS mixers. For that reason another critical test is to determine the bandpass characteristic of the AOPT around the SLO frequency. Because of the availability of gunn oscillators these measurements were possible for the first time after the thermal cycling. In section 7.10 of part 1 it was shown that the transmission characteristic of the FSP filter is affected by the cross-polar leakage of the polarizing grid RG1 and the analyzing grid of the COPT simulator. The latter can be minimized by using two grids in tandem. This comes close to the true situation in the COPT of SMILES where the analyzing grid is co-polar to the Rx horn antennas whereas in the COPT simulator the angle of that grid has to be set to +/-45 degrees with respect to the receiver polarization. The leakage of RG 1 can be minimized for the TRN path by injecting the right polarization into the BBH. For the CST path, however, the different angles of RG1 and LG1 always leads to a cross-polar component at RG1. A small fraction of it can leak into the signal path an cause the observed shifts of the rejection frequencies between TRN and CST. In the case of the SLO path the cross-polar component on RG1 is much larger than the co-polar, and thus the frequency shift can be expected to be much stronger. For that reason bandpass measurements around the SLO frequency through the TRN or CST path, as shown in this section, can only be used to understand the FSP theory. The true SLO power split between the two mixers has to be determined after the SLO integration. Since the unbalance from the SLO cross-polar leakage is expected to be to large to be acceptable an additional grid was manufactured to be placed between RG1 and LG1. In this configuration the SLO split should be very similar to the measurements through the TRN path with the adequate polarization of the source. 10

12 2 2.5 trn_h_30.dat, # , # deg +45 deg dB GHz Relative Amplitude [db] bandpass_trn_h_slo_3,a. Murk, IAP, 26 Mar 2002, 22:39 Figure 7: TRN bandpass characteristic around the SLO frequency trn_h_30.dat, # 53 54, # deg +45 deg dB GHz 2.8 Relative Amplitude [db] bandpass_cst_h_slo_3,a. Murk, IAP, 26 Mar 2002, 22:39 Figure 8: CST bandpass characteristic around the SLO frequency 11

13 4 FSP Tuning The FSP filter was manufactured with a positive bias of 2 m on each mirror to have the chance to tune the filter by removing this bias. All measurements in part 1 and section 2 were made with these initial FSP spacings. The distances between the mirror and the center of the grid wires were measured as d = mm and d = mm. After the initial measurements the FSP was tuned by replacing one of the backing mirrors with a thinner one. The nominal FSP distances are now d = mm d = mm. The figures in this section display the bandpass characteristic after the FSP tuning. In addition the SMI grid in the TRN path was activated and acts as a linear to circular polarization converter. Since no submillimeter source with the appropriate circular polarization was available the the linear polarization of the source was set to 0 degrees for these measurements. Tabular 1 summarizes the FSP rejection and -3 db cross-over frequencies at different stages of the test program. The bandpass characteristic after the tuning is almost identical with the design values of , and GHz. This is a very remarkable result if one takes into account that a 1 m change of the FSP spacing leads to a frequency shift of 0.2 GHz. However, the true bandpass characteristic of the AOPT in SMILES will differ from these measurements because of the following reasons. First, the measurements were made under atmospheric pressure of about 950 hpa, while SMILES has to operate in vacuum. The refractive index of air leads to a reduction of the rejection frequencies by a factor of Second, the measurements were made with a linearly polarized source, while the atmospheric signal is unpolarized. Together with the cross-polar leakage of the grids the latter might lead to a frequency shift which is not analyzed yet. TRN TRN TRN CST CST CST LSB USB -3 db LSB USB -3 db before vibration after vibration after thermal cycling after FSP tuning Table 1: Summary of the rejection and -3 db cross-over frequencies of the AOPT 12

14 trn_h_41.dat, # , f= GHz 0 10 Gaussian Fit hpbw= 2.04 Θ = 0.34 Relative Amplitude [db] Θ [deg] Phase [deg] Phase Fit = λ axial = 1.19 λ lateral Θ [deg] pattern_trn_h_lsb_4b,a. Murk, IAP, 27 Mar 2002, 20:7 Figure 9: TRN LSB antenna pattern after FSP tuning, AOPT horizontal 13

15 x 10 3 trn_h_41.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = 0.00 db trn_h_41.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = db bandpass_trn_h_lsb_4,a. Murk, IAP, 26 Mar 2002, 21:53 Figure 10: TRN LSB bandpass after FSP tuning 14

16 20 trn_h_41.dat, # Relative Amplitude 45 deg [db] f0 = GHz a0 = db trn_h_41.dat, # Relative Amplitude +45 deg [db] f0 = GHz a0 = 0.00 db bandpass_trn_h_usb_4,a. Murk, IAP, 26 Mar 2002, 22:8 Figure 11: TRN USB bandpass after FSP tuning 15

17 2 2.5 trn_h_41.dat, # 14 15, # deg +45 deg dB GHz Relative Amplitude [db] bandpass_trn_h_slo_4,a. Murk, IAP, 26 Mar 2002, 22:17 Figure 12: TRN bandpass characteristic around the SLO frequency after FSP tuning 16

18 5 Absolute Beam Position The beam pattern measurements were used to determine the absolute position of the phase center with respect to the rotational axis. Tabular 2 summarizes the AOPT beam parameters taken from the figures in part 1 and 2. The half power beam width and the pointing error are a least squares fit to the amplitude pattern. The boresight position of the rotation stage was not optimized to the same degree for all measurements, which affects the accuracy of. With the currently used technique an alignment of the setup better than 0.1 will be difficult to achieve. It should be noted that the reference plane for the bore-sighting was the BBH aperture, and not the alignment panel. The axial and lateral offsets of the phase center from the rotational axis were derived from the measured phase as described in part 1 of the test report. They are given relative to the wavelength. The axial offsets should not be taken to serious because their values depend on the definition of the phase center and on the fraction of the pattern which is used for the fit. The most important parameters in tabular 2 are the lateral offsets. They include the errors of the rotation stage, of the test fixture holding the AOPT, of the COPT simulator and of the AOPT itself. The errors of the rotation stage may have changed between the measurements because the test fixture can be mounted in two different ways on it. The filenames in the tabular show the BBH under test (TRN or CST), the frequency (USB or LSB) and the plane in which the AOPT was measured (horizontal or vertical). The numbers *0.eps indicate the initial measurements, *1.eps were taken after vibration, *2.eps after thermal cycling, *3.eps after the EMC tests and *4.eps after the FSP tuning. 17

19 HPBW axial lateral Filename [deg] [deg] [ ] [ ] pattern cst v 0.eps pattern cst v 1.eps pattern cst v 1 650GHz.eps pattern trn v 0.eps pattern trn v 1.eps pattern trn v 1 650GHz.eps pattern trn h 0.eps pattern trn h 1.eps pattern trn h 1 rot normal.eps pattern trn h 1 rot copt.eps pattern trn h 1 rot cm2.eps pattern trn h 1 usb.eps pattern trn h lsb 2.eps pattern trn h lsb 2 flat.eps pattern trn h lsb 2 g1.eps pattern trn h lsb 2 g2.eps pattern trn h lsb 3.eps pattern trn h lsb 4b.eps Table 2: Summary of the AOPT beam parameters from part 1 and 2 of the EM test reports The tabular shows that offsets and pointing are affected by the coupling into the COPT port. The three measurements of pattern trn h 1 rot*.eps were taken with different setups of the COPT simulator as discussed in section 7.10 of part 1. This had a significant influence on the symmetry of the antenna pattern (figure 23 in part 1) and on, but only a smaller influence on the lateral offset. For the measurements pattern trn h 2 g1.eps and pattern trn h 2 g1.eps two different sides of a grid were used instead of a flat mirror for the COPT coupling. As shown in figure 13 this has changed the sidelobes, but also by 0.25 degrees and the lateral offset by It remains an open question why the lateral offset has increased significantly after thermal cycling. However it is unlikely that this was caused by changes within the AOPT itself because the symmetry of the pattern and the internal reflections remained unchanged or have even improved. 18

20 5 0 5 trn_h_23.dat, # , f= GHz coupling mirror, flat coupling grid, position 1 coupling grid, position 2 10 Relative Amplitude [db] Θ [deg] Phase [deg] coupling mirror, flat coupling grid, position 1 coupling grid, position Θ [deg] Figure 13: Subsequent measurements of the TRN antenna pattern where the coupling into the COPT port was done with a flat mirror as usual and with the front and the back side of a wire grid. From these measurements lateal offsets of 1.44, 1.58 and 1.20, respectively, can be computed. 19 pattern_trn_h_lsb_2_misaligned_grid,a. Murk, IAP, 21 Dec 2001, 21:5

21 To investigate whether the observed beam offsets can be explained by artifacts of the test setup the mechanical offset of the AOPT aperture from the rotational axis was measured directly. For that purpose a steel ball was fitted on the BBH with its center at the same position as the center of the BBH aperture. With a digital clock the mechanical offset can be determined by rotating the AOPT (Fig. 5). The results of such measurements with and without the COPT simulator attached to the test fixture are shown in figure 5. The fact that the two measurements differ indicates a significant dynamic effect from the weight of the COPT simulator. It can be expected that the AOPT and the fixture have a similar effect on the test setup because all components have a relative large mass which is not centered over the rotational axis. It is unknown whether the bearings of the rotation stage, the Aluminum table which holds the experiment or any other component is bending under the large torque of the test setup and whether this has changed during the tests. A possible solution for this problem will be to counterweight the equipment on the rotation stage. When the clock measurement of 351 m lateral offset from figure 5 is compared to the electrical offset of 1.19 (571 m) from figure 2 a difference of 220 m remains to be explained. To quantify the misalignment of the TRN beam the following mean absolute values can be summarized from tabular 2. The mean pointing error results in = 0.26 degree and is mainly caused by the bore-sighting errors of the setup. The true value of this error and the accuracy to which the test setup can be aligned should be in the order of 0.1 degree. The mean axial offset of 14.3 has only little significance. To get a reliable value for this property the data has to be analyzed again with an appropriate definition of the phase center. The mean lateral offset is 0.34 and 1.30 before and after thermal cycling, respectively, which includes the errors of the AOPT and the test setup. A reduction of the measurement artifacts might be possible with more elaborated alignment techniques and a more rigid test setup, but the effects of the distortion and alignment errors at the COPT/AOPT interface will remain. 20

22 BBH Figure 14: setup with a digital clock and a steel ball centered at the TRN BBH aperture without COPT Simulator Fit: 149 µm lateral, 294 µm axial with COPT Simulator Fit: 351 µm lateral, 491 µm axial 300 Clock Offset [µm] Angle [deg] Figure 15: of the mechanical offset of the TRN BBH aperture from the rotational axis 21

23 6 Conclusions The AOPT has passed thermal cycling without degradation. Only the lateral offset has changed significantly after the cycling which can partially be explained by measurement artifacts. The tuning of the FSP filter resulted in the designed rejection frequencies and SLO balance. Because of time limitations and technical problems with the SLO and the harmonic mixer the following actions still have to be done: Bandpass and antenna pattern measurements at 40 C Reflection measurements with activated SMI Planar 2D beam pattern measurements SLO integration, vibration and testing 22