(click to enlarge) |
(click to enlarge) |
|
(click to enlarge) |
(click to enlarge) |
The temperature and gain plots above are for a single
descent from 35 kft to sea level and back to 35 kft for a DC8 PEM Tropics
B flight on 19990320. G_raw is the gain derived using (Ctgt-Csky)/(Ttgt-Tnav)
for cases where Ttgt-Tnav > 20 K. Since this conditions is
only met for a short time on descent, most of the raw gain data is for
the ascent where target stays warmer that the OAT by >20 K. Note that since
the temperature measurements are not noisy, the raw gain variation is entirely
due to "counts" noise, or equivalently, radiometer thermal noise (and it
is of the correct magnitude, about 0.3 K rms). In the past we have used
the IF Amplifier/Mixer temperature (Tifa) to monitor the
temperature dependence of the gain. The gain resulting from a linear regression
of the raw gains to Tifa is indicated by the curves denoted:
G_ifa. It is clear from the right-hand end of the raw gain curves
that G_ifa does not do well here. This is because the gain is sensitive
to other temperatures. In this particular case, there is a clear correlation
between the gain drop and a temperature increase in Tlo1, the temperature
of the synthesizer. It is also clear from the temperature curves that Tlo2
(the LO amplifier temperature) shows a peak-to-peak temperature variation
of ~2 K, while Tifa and Tlo1 have a peak-to-peak variation
of ~5 K. The gains derived by performing a linear multiple regression to
the the raw gains using Tifa, Tlo1, and Tlo2 are indicated
G_lmr. Clearly this gain is the best fit to the raw data. Finally,
the gains denoted G_old are the gains from a linear fit using Tifa
BEFORE the quad multiplier on the LO was replaced in March 1999. As a final
comment, we note that it should not be suprizing that a gain fit using
several temperatures does better than only Tifa. This is because
the new sensor units use synthesized LO units which are not heat sunk to
the same temperature as the IF Amplifier. Also, in the case of the DC8,
cold air from a Mission Manger's "eyeball" is blown directly on to the
synthesizer. We would expect that it should show more gain changes, and
indeed it does.
(click to enlarge) |
(click to enlarge) |
This set of plots is for the entire 10 hour flight on April 10, 1999. The same information is plotted as for the previous pair of plots. Although the regression coefficients and offset are different for this fit compared to the previous fit, a careful comparision of the resulting gains for this and the previous flight are in good agreement.