Sources of Non-Ideal Instrument Performance
(from McFadden et al., 2008b)
There are several performance issues with the plasma data of which data users should be aware (sources of background contamination, non-ideal response of the instrument, limitations due to missing information, and telemetry formatting problems). Below we provide a description of these potential problems. For more information, see McFadden et al., 2008a, 2008b.
- Scattered solar UV to the detectors is nearly negligible with peak rates <20/s summed over all anodes and confined to ~10 degrees of rotation.
- Photoelectrons from Langmuir sensors produce a spectral peak in the eESA at energies just below the spacecraft potential, Φsc. During some non-ideal Electric Field Instrument modes, photoelectrons were produced at energies greater than eΦsc.
- Photoelectrons from spacecraft surfaces prodcue a broad, low-energy, anisotropic distribution. Eliminating these photoelectrons from moment calculations is often difficult, with no clear spectral break to allow determination of Φsc.
- Photoelectrons are prodcued on internal analyzer surfaces over about 20° of spacecraft rotation centered on the sunward direction. These low-energy (<10eV) photoelectrons do not generally introduce errors unless the spacecraft potential is small.
- Internal scattering of energetic electrons, including secondary electrons produced by these primaries, results in a low-energy, out-of-band-pass response (rate~E-5/3) in the eESA, which can result in significant (~20%) errors in density when Te>10 keV.
- Bremsstrahlung x-rays from ~100 keV electrons striking s/c surfaces produce background rates in both ESAs. These rates are lower than those described in caveat #5 for eESA, however iESA rates can introduce large (factor 2-10) errors in ion density.
- Penetrating radiation from the radiation belts becomes important at ~6RE geocentric. This background produces flat spectra in both iESA and eESA, with increasing count rate that peaks at ~4RE.
- THEMIS has no composition information and data analysis assumes the measured ions are protons. A higher-mass ion's contribution to number density will be underestimated by the factor (m/q)1/2. Similar errors are incurred for higher order moments.
- Spacecraft charging may prevent the measurement of portions of the plasma distribution. Since spacecraft gnerally charge positive to attract photoelectrons, cold ions are often missed.
- Spacecraft charging can result in distortions of the electron distribution that are difficult to correct. In particular, a cold (few eV) electron component may be difficult to separate from spacecraft photoelectrons even when spacecraft potential is measured.
- The upper limit to the ESA's energy range can result in significant plamsa being missed and therefore errors when calculating moments from ESA-only data. Combining ESA and SST data is recommended for moment computations when the plasma is hot.
- Since ESA has a limited FOV, time aliasing during a spin can skew the measurement. This is especially true for the electron velocity moment where a few percent density variation during a spin can result in a large (~100 km/s) error in the calculated flow.
- Since the ESA FOV is limited to ~6° perpendicular to its 180° field-of-view, and since nominial energy sweeps occur during 11.25° of rotation, narrow beams can be missed. This affects the iESA during solar wind measurements in nominal modes.
- Lost counts due to instrument dead-time result from electronic and detector dead-time. Significant dead-time corrections (>10%) are required fro high fluxes seen by the electron sensor in >50 cm-3 magnetosheath, and by the iESA throughout the solar wind.
- Periods where EFI data is unavailable occur, especially during the first year. Since knowledge of Φsc is essential for the correct transformation of measured counts to plasma distribution function, EFI data should be checked for availability and quality.
- During an eclipse, when sun-sensor data is unavailable to organize spin-synchronous plasma data, small drifts in spin phase due to thermal contraction of the antenna will result in distortions of particle distributions and computed vector moments.
- Spacecraft charging during an eclipse can result in large, negative, kilovolt spacecraft potentials which distort measured particle distributions and generally exclude low-energy electrons. EFI measurements of Φsc are not valid in shadow.
- Data formatting problems resulted in the misdirection of small amounts of data in ESA and SST packets prior to April 27, 2007. Care should be taken in interpreting these early commissioning-phase data, especially any moment computations on these data.
- Data formatting errors occur at the transition between insturment modes. Depending upon the data product, these transitions can result in a data loss, or incorrectly formatted data, that lasts for a few seconds to a few minutes.
- The data formatting algorithm can cause counter saturation (flat-topped spectra) in data products that are formed by summing counts into a small number of solid angle bins. It primarily impacts high count rate data, such as electron data in the magnetosheath.
- Significant errors in onboard computed moments were present until corrections to the flight software were implemented on August 6, 2007. Onboard moment computations prior to this date should be ignored.
- Onboard moment computations, especially electron moments, are generally invalid until EFI antenna deployments allow computations with Φsc. For THB and THA these deployments were compelted on November 18, 2007 and January 13, 2008, respectively.
- In-flight calibrations are an ongoing process with final calibrations generally lagging data collection by about one year. Data are only spot checked for accuracy as they are collected and made available to the community. We appreciate any reports of problems or unusual data to the ESA team leads.
- Electron ESA sensors have experienced unexpected degradation in sensitivity at low energies (<40 eV). This is probably due to intense spacecraft photoelectrons being focused onto small portions of the MCP detectors by leakage fields at the analyzer exit. These intense fluxes have caused localized scrubbing of the MCP, and associated gain degradation. In-flight calibration corrections will be developed and should eventually provide corrected data, but such corrections are not available at this time. The amount of degradation varies with s/c and detector look direction. This will primarily impact measurements in the solar wind.
- Ion ESA sensors, especially THB and THC, have experienced sensitivity degradation over a small portion of the MCP detector centered on the ecliptic plane. The degradation is due to intense solar wind ions causing localized scrubbing of the MCP. This will cause underestimates of density and overestimates of temperature for solar wind measurements.
- During the second tail campaign, "full" (peef) distributions of electrons, with 15 energy x 88 solid angles (15Ex88A), were captured at spin resolution during fast survey, rather than the previous 32Ex88A distributions at 32 spin cadence. Although this mode provides provides much better time resolution for full distributions, the reduced energy resolution does impact moment calculations since the break between plasma and spacecraft photoelectrons is poorly resolved. Ground computed moments from these data products should checked for photoelectron contamination. This problem can be partly overcome by combining peer (32Ex1A) and peef (15Ex88A) data to create a higher resolution (32Ex88A) data product using the get_th?_peec.pro routine. Preliminary testing shows this routine reproduces moments very similar to 32Ex88A burst data. However, problems with this routine may arise within the sheath where peer data are often saturated.