SWEA Sporadic Low-Energy Suppression

Starting in December 2018, a small fraction (0.006 %) of spectra exhibited
reduced signal at energies below ~20 eV.  This occurred sporadically for
individual spectra surrounded by normal spectra before and after, indicating
that the instrument recovers quickly.  The anomaly rate increased to 0.04 % in
January 2019, and then to a maximum of ~1 % during aerobraking in
February/March 2019.  By analyzing instrument housekeeping, we found that the
occurrence of the anomaly is highly correlated with the analyzer temperature,
as measured by a thermistor mounted on the anode board.

The anode board is located between the microchannel plate (MCP) and the high
voltage power supply (HVPS).  This part of the instrument is farther away from
operational heaters at the base of the electronics box (which houses the
digital board and LVPS), and thus experiences larger temperature variations.
Most of the time, the spacecraft points at the Sun, and the SWEA analyzer is
in the shadow of its own electronics box, with the analyzer temperature near 
-4 C.  The anomaly is not observed in this configuration.  When the spacecraft
points away from the Sun (e.g., during communications with Earth), the
analyzer is outside the shadow of its electronics box and is thus warmed by
sunlight.  The anomaly begins to appear when the analyzer temperature rises
above 0 C and reaches its maximum occurrence rate at or above +2 C.  The
anomalous behavior continues to occur after the analyzer is back in shadow, so
long as the analyzer temperature remains above 0 C.

The anomaly energy dependence points to an issue with the analyzer voltages.
A small voltage offset on one of the interior surfaces could "detune" the
analyzer, resulting in signal suppression at low energies, when analyzer
voltages are small (~1 V or less).  This same voltage offset becomes
negligible at high energies, when analyzer voltages are large. The temperature
dependence suggests that the issue could be associated with the high voltage
power supply (HVPS).  On April 20, 2019, the instrument was commanded to
generate high time resolution housekeeping data, so that the analyzer and
deflector sweep waveforms generated by the HVPS could be captured during both
normal and anomalous operation.  However, the sweep waveforms were identical,
indicating that the HVPS is operating normally.

An alternative explanation is that there is a marginal electrical connection
between the HVPS and one of the surfaces it drives, most likely the top cap,
which is close to the entrance of the hemispherical analyzer.  This could
account for the sporadic occurrence as well as the temperature dependence
through thermal expansion and contraction.  The cumulative effects of thermal
cycling over time could also explain why the anomaly only began to appear
after four years of operation.

The temperature dependence points to a mitigation: reduce the operating
temperature.  SWEA's operational heater set points were lowered by 5 C on May
1, 2019, and the anomaly rate dropped to nearly zero.  The peak occurrence
rate from May to August is ~0.001 % (less than one per 60,000).

Software was designed to automatically identify anomalous spectra by comparing
every spectrum to those measured before and after.  Anomalous behavior has a
characteristic energy dependence, which helps in the identification.  False
positives, which tend to occur in the sheath, where there are large
fluctuations, are removed by visual inspection.  Spectra affected by the
anomaly are masked with NaN's at energies below 28 eV (to be conservative).
This applies to all three data types: SPEC, PAD, 3D.