1. Maxim Tuchin, Anton Biryukov, Mikhail Prokhorov, Andrey Zakharov Space Project Laboratory, Sternberg Astronomical Institute, Lomonosov Moscow State University

2. Comparing real and estimated random errors Dω NNNN  x,y BOKZ, IKI (Russia) SED26, Sodern (France) S3S, (Canada) ST-200, BST (Germany) 2

3. Random error sources 3

4. Systematic error sources IImage pixelization LLimitation of the analyzed area while measuring coordinates DDifferent sensitivity of the sensor elements (PRNU) IInhomogeneous sensitivity within a sensor element DDark current non-uniformity (DCNU) EEffect of "hot" pixels DDisplacing effect of electron traps VVarious aberrations DDifferences in the spectral sensitivity of elements (SRNU) NNon-uniformity of gains, bias DDegradation of sensor and optics EEtc. 4

5. Pixelization and bounded area errors  Using integrated signal from pixels rather then signal distribution function across focal plane  Using limited area instead of whole focal plane  i.e. using instead of 5

6. Error of star image position depending on image size 6

7. Shift of measured coordinates depending on position within pixel 7 Small size star images Large size star images Pixelization effect Bounded area effect

8. Dark current non-uniformity  Sensor elements have different levels of thermo generation rate of electrons  Non-uniformity is high (up to 20-30%)  It leads to increased errors 8

9. Effect of DCNU on the position error 9 Note: SNR is signal-to-noise ratio with DCNU=0 and reflects the signal’s strength

10. Effect of DCNU on the star image detection 10 Note: SNR is signal-to-noise ratio with DCNU=0 and reflects signal’s strength

11. DCNU changes with Total Dose Irradiation and after annealing 11 High Accuracy Star Tracker (HAS) Version 2 CMOS Active Pixel image Sensor (CMOS APS) Source: HAS2 Detailed Specification - ICD

12. Solutions to DCNU problem  Cooling the sensor + We can forget about DS (except “hot” pixels) ─ Additional elements to design ─ More power consumption ─ Heat dumping problem ─ Increased effects of electron traps  Dark signal estimation ─ Increasing requirements to calculation capabilities ─ Less accuracy due to additional noise ─ Recalibrations are highly recommended 12

13. Conclusions  To achieve the high accuracy one should Do ground-based calibrations Store fair-sized calibration data (like bias or sensitivity maps) onboard Provide for various corrections during data reduction Cool the sensor or calculate dark signal Do some space-based calibrations 13