Estimation of cross coupling of receiver noise between the EoR fat-dipole antennas

Transcription

1 Estimation of cross coupling of receiver noise between the EoR fat-dipole antennas Due to the proximity of the fat dipoles in the EoR receiver configuration, the receiver noise of individual antennas may get coupled to the other resulting in a correlated noise component at the receiver output. The antennas also receive the thermal noise emitted by both the resistive screen placed in between them and the ferrite tile absorbers kept below them. Thus the system operating in the interferometer mode produces a complex output P o proportional to the sum contribution of the correlated components of sky background (T sky ), thermal emission (T screen ) from the resistive screen, ground radiation from the tiles (Tgnd) and the amplifier noise of individual antennas (T Rx ), all of which are complex responses. Thus the system response in interferometer mode may be written as Po = G ( T sky + T screen + Tgnd + C T Rx ), where G is the complex system gain and C is the complex coefficient of coupling of receiver noise between the antennas 1 and 2. The response to each of these sources is a complex quantity that may vary with frequency. It is necessary to know the phase response of individual components, so that the cumulative response may be estimated. One of the ways of knowing the magnitude of individual contributions is to independently vary each source, one at a time, and observe the variation in the total response. Since all the components except the one that is being varied are kept constant, contribution from individual components may be identified. In order to know the noise contribution from each of the receivers through cross coupling, the noise temperatures of both of them were varied. The first stage of the receiver is a low noise amplifier (BX from Spectrum Microwave) having a noise temperature of about 130 K (NF of 1.8 db). A post amplifier (QB 300 from Spectrum Microwave) that has a noise temperature of about 330 K (NF about 3.2 db) follows it. These two amplifiers were swapped in the receiver units associated with the two antennas in order to significantly change the noise temperature of the receiver (by about 200 K). Let Tr 1 and T r1 represent the noise temperature of receiver 1 and P o1 and P o1 represent the receiver 1 output under swapped and unswapped conditions respectively. Thus the receiver output can be written as and hence P o1 α ( T sky + T screen + Tgnd + T r1 ) P o1 α ( T sky + T screen + Tgnd + T r1 ) and (P o1 - P o1 ) α ( T r1 - T r1 ) (1)

2 representing the variation of the receiver output when amplifiers are swapped. In a similar way the receiver 2 output can also be written as (P o2 - P o2 ) α ( T r2 - T r2 ) Let C represent the coefficient of cross coupling for power, (C T r1 ) and (C T r1 ) represent the noise temperature of receiver 1 coupled to the receiver 2 under unswapped and swapped condition respectively, Similarly let (C T r2 ) and (C T r2 ) ) represent the noise temperature of receiver 2 coupled to receiver 1 under unswapped and swapped condition and Pc and Pc represent the correlated output power under those conditions. The correlated output power can be written in terms of receiver temperatures as Pc α (T r1 ) ½ (C T r1 ) ½ + (T r2 )½ (C T r2 ) ½ and Pc α (T r1 ) ½ (C T r1 ) ½ + (T r2 ) ½ (C T r2 ) ½ The change in the output due to swapping can be found out by taking the difference between the above equations. Thus we get (Pc - Pc ) α C ½ ( T r1 - T r1 ) + C ½ ( T r2 - T r2 ) If we assume that T r1 is identical to T r2 since similar amplifiers are used in both the receivers, the above equation can be simplified as (Pc - Pc ) α 2 C ½ ( T r1 - T r1 ) (2) Dividing Eq. (2) by Eq.(1) and rearranging the terms to calculate the coefficient of coupling, we get (Pc - Pc ) 2 C ½ ( T r1 - T r1 ) = (P o1 - P o1 ) ( T r1 - T r1 ) (Pc - Pc ) 2 C ½ = (P o1 - P o1 ) (Pc - Pc ) 2 C = (3)

3 4 (P o1 - P o1 ) 2 Tsky Tsky Tscreen Ant-1 T rx Ant-2 Tgnd Balun + Ph.Sw Balun Cal αcal LNA (NF=1.8 db) LNA (NF=1.8 db) G1 G2 QB300 (NF=3.2 db) QB300 (NF=3.2 db) Digital Receiver Auto and Cross Spectra out Fig.1 Schematic showing the various components of the input signal to the receiver

4 The measurement sequence was as follows: first the amplifiers in ant 2 were swapped raising its receiver noise, then this was done in ant 1 so that both antennas had high Trec, then the amplifier sequence in ant 2 was restored and finally ant 1 was also restored. In each case the interferometer data was acquired for about two hours. Reference data was obtained in identical conditions he following night over the same LST range. In the following figure is shown plot of Tsys versus LST; the reference Tsys for both antennas obtained during the second night is also plotted. When one of the antenna Trec were increased by about 200 K, the Tsys of the other antenna did not noticeably change Tsys; the measurement errors indicate an upper limit of about 2 K (1%) in the change in Tsys in the neighboring antenna. This also shows that the absolute calibration is stable to within 2% on successive days. Shown below are the cross power spectra recorded in the three cases with ant 1 amplifier modules swapped, with ant 2 amplifier modules swapped, and the third curve shows the spectrum for the case where both antennas had their modules swapped. The real and imag components of the three spectra are shown separately, followed by amplitude spectra in the three cases.

5

6 Since the reference spectra are from the next night, the following plot was created of spectra on the two successive nights made in identical system configurations. Difference spectra are shown, of real and imaginary components, between spectra recorded with same receiver configurations. Clearly, the interferometer spectra do change when the receiver noise increases. The magnitude of the change varies across the band and is of order 5 K. However, the spectra recorded when both antennas had higher Trec are not simply the sum of the spectra recorded when the individual antennas had their Trec raised this requires further careful measurements. Given that the antennas are about 0.8 m above the receivers, and the spacing between the antennas is 1.5 m, the relative path difference between the propagations of receiver noise from any of the antenna LNAs to the correlator is about 2x = 3.1 m. Over the octave band, this results in the receiver noise component having a phase change of about one cycle. This frequency structure in the above plots is roughly consistent with such an expectation. 5K correlation product arising from 200 K Trec noise corresponds to a power coupling coefficient of -32 db. This may be reduced by the insertion of a broad band isolator in the signal path between the directional coupler and the LNA; however, that will degrade the system sensitivity owing to ohmic loss.