Optical Driver Comparison

Data was obtained using double position switching between NGC 7027 and 2202+422. See GBT01A_004 Observation Checkout Report -- March 31, 2004 for information on double position switching and its data reduction. If everything is behaving linearly - there are no gain changes and bandpasses are constant - then the resulting data should just be the ratio of the source fluxes. For example, if the target source has a spectrum given by $S^{target}(\nu) = A \nu^\alpha$ and the calibration source has a spectrum given by $S^{calibration}(\nu) = B \nu^\beta$ then the double position switching would result in a spectrum that goes as

$${S^{target}(\nu) \over S^{calibration}(\nu)} = {A \over B} \nu^{\alpha-\beta}.$$ (1)

Figure 1: Double position switch observation of NGC 7027 and 2202+422 using different Optical Drivers with data taken at the same time.

The resulting data are shown in Figure 1. The plot consists of four spectral windows with two polarizations in each window. The plot is color coded as to which signals used common Optical Drivers. As can be seen from Figure 1 a constant flux ratio is not achieved. The flux ratio depends on which set of Optical Drivers was used. The variation in flux ratio is greater than could be expected due to measurement noise. Furthermore it was found that the flux ratio varies in time. This means that you can not convert GBT double position switch results onto a scale with known flux units.

Given the previous results that the non-linearities arise between LO1 and LO2, the results of Figure 1 suggest that the non-linearities arise in components associated with a particular Optical Driver path. Data coming down the same Optical Driver has about the same flux ratio so this suggests that the largest non-linearities are before the LO2 mixes and after the IF splitter in the receiver.

Figure 2: Double position switch observation of NGC 7027 and 2202+422 using ${\rm T_{cal}(off)}$ only data in red and ${\rm T_{cal}(on)}$ only data in blue. Data from two polarizations are shown. It is easily seen that a power change of $\sim 2.6 {\rm Kelvins}$ from firing the low noise diode is enough to see a non-linearity in the results.

In Figure 2 we compare the flux ratios found from double position switching using data only when the noise diodes are off compared to using data only when the noise diodes are on. It is easily seen that the $\sim 2.6 {\rm Kelvins}$ that the noise diodes add to the total power in the system are enough to excite a non-linear response. This suggests that the problem may be such that the IF system always has some degree of non-linearity.