Lessons learnt from the SiREUS MEMS detector evaluation 6th ESA Workshop on Avionics Data, Control and Software Systems (ADCSS-2012) 24 th of October 2012 Steeve Kowaltschek (ESA/ESTEC) Consortium involved :
Context 1. Overview of the MEMS used in SiREUS 2. History of the product development 3. The Detector evaluation test programme 4. SiREUS qualification and analysis 5. Short term future activities 6. Lessons Learnt 7. Conclusions
Introduction to the MEMS gyro SiREUS is a coarse rate sensor using a MEMS resonating ring gyro Transition of automotive technology to Space A planar ring structure, etched from silicon, held in a resonating state by a primary drive voltage 10 mm The qualification of MEMS based technologies in Space equipment is critical to market and flight acceptance. No ECSS or alternative recognized standard exists for the qualification of MEMS A bespoke evaluation approach was developed based on ESCC standard
What is SiREUS? And link with miniaturisation SiREUS is the 3-axis MEMS Rate Sensor (MRS) developed by a UK consortium of SELEX Galileo (Edinburgh), SEA (Bristol) and AIS (Plymouth). Two ESA contracts: C18800 covering Phases 1 and 2, C22534 covering Phase 3. Main SiREUS specs 750g 5 W power supply Available in 28 V unregulated bus, soon 50V Rad-hard, using ESCC/QMLV grade 1 parts Analog output, RS422 UART Serial Interface Spacewire capability in the ASIC Scalable output rates (2-20 Hz for UART, up to 14 khz for SpW) Operating range 20 to +20 deg/s 70mm SiREUS gyro detector (MEMS Ring + JFETs) was submitted to a full evaluation program during the Phase 3 to approve its use for high-reliability space application and propose it for entry in the European Preferred Part List (ESA PPL). SiREUS is by far the smaller, lighter and less power consuming space rad-hard gyro in the world.
Long history of SiREUS from initial investigation to FMs Existing Terrestrial Design Date 1999 2004 2005 2007 2007 2010 2009 2013 Milestone BNSC sponsored SEA to study Vibrating Structure Gyros (VSG). Feasibility studies completed by Astrium UK and SEA, development consortium discussions initiated with ESA. SEA under contract for ESA Phase 1 development with Goodrich (now UTC) and SELEX Galileo as consortium partners ESA Phase 2 development led by SEA commences, Flight Experiment model delivered to Cryosat-2 in 14 Months from start of Phase 2 Contract awards for Sentinel-3 Flight Models, Orbital Sciences FMs and ESA Phase 3 development led by SELEX Galileo. Unit qualification. EM deliveries and phase 3 activities. Significant market interest in SiREUS. Final Detector Evaluation and unit Qualification. Electronics design, breadboard manufacturing and testing EQM Manufacturing and Testing DC/DC Power Supply redesign ASIC Development Detector redesign, prototype manufacturing and testing FM Generic Design Activities Flight Test (Cryosat 2) MEMS Detector Full Evaluation FM Concept PFM Manufacturing and Testing MRS delta qualification Development Phase 1 Development Phase 2 Development Phase 3
How to evaluate a MEMS detector? MEMS Evaluation test flow derived from ESCC 2269000 TEC-Q (Quality departement) and TEC-E (Electrical Engineering) were involved to derive this test plan. Adaptation required and may differ for other MEMS devices: Different functional requirements Widely varying manufacturing processes Often unique failure mechanisms Diverse end-customer applications Evaluation purpose it to find the design limits. Many tests are destructive in order to set the threshold for future lot acceptance tests. A large amount of money was required to produce and test the required 106 evaluation detectors. Additional quantities procured for FM and to support analysis programmes.
Evaluation groups for MEMS Gyro Detector Control Group (Group 1: 5 detectors) Destructive Test Group (Group 2: 61 detectors) Step stress tests (temperature & power step stress - Electrical and Mechanical) Radiation tests (TID, SEU, SET) Construction analysis (eg, internal visual inspection, SEM inspection, bond strength, die shear) Package tests, split into Thermal tests (temperature cycling & thermal shock) Mechanical tests (shock & vibration) Soldering test (resistance to soldering heat) Pin to pin isolation Electrical tests (ESD testing & Characterization) Endurance Test Group (Group 3: 35 detectors) High temperature reverse bias Accelerated electrical endurance test Extended burn-in test Reserve Group (Group 4: 5 detectors)
Group 2B: Radiation Testing (1/3) All types of radiation tests have been applied to the detector: Total Dose up to 100 krads, (Co-60) Heavy Ions for SEU/SET Protons for SEU/SET Protons for Displacement Damage Configuration of the tests: Detector Lids removed for testing. Bias Board to power up detectors for TID and Displacement damage. JFETs were targeted during Heavy Ion & Proton SEU tests. MEMS + JFET electrical characteristics were monitored throughout the tests. ESTEC TID Radiation facility Representative electronics (EM SiREUS) for SET testing. Detector outputs and SiREUS outputs recorded.
FM Detector screening testing at SELEX A long term drift component is identified on some detectors from the early builds. Additional detectors procured with the Phase 3 evaluation batch for FM programmes and to investigate detector characteristics. Screening introduced on representative electronics as a short term measure to assess detector performance ahead of system integration The test is performed at 60degC switched on after at least a 1 hour unpowered bake. Log sampling lasts 24 hours or greater from rig/detector switch on. Important: a measured long term drift of 0.1 deg/h per day it is possible to: Update the bias offset every 100 days for instance to remain within a + or 10 deg/h envelope. A simple TC can update SiREUS EEPROM with updated offset. Similar approach is used on Star Trackers to update regularly the Julian Date in order to improve performance (relativistic aberration mitigation) No need to store history in the OBC. Next steps For future builds, consortium is discussing on how to improve this screening.
Charge Trapping Update Out of the screening carried out by SELEX Galileo to characterize the FM batch, here are the key results: Charge Trapping (CT), at initial (first virgin ) switch ON at 60C: Average CT peak: 68 deg/h (temperature effect included) Less than 15% of the samples peak above 100 deg/hr Average CT settling time: 2 hr Standard Deviation CT Settling time: 2.65 hr For the follow on switch-ons, the time constant to reach performance is drastically reduced. Here is the example of a 11-day and 20-day off-period with the unit left at room temperature. time to settle from previous bias value before switch off (11 days) time to settle from previous bias value before switch off (20 days) @30 degc - Jan 2012 @30 degc - Oct 2012 10 deg/h 20 deg/h 30 deg/h 10 deg/h 20 deg/h 30 deg/h X 00:07 immediate immediate X 05:30 immediate immediate Y 00:38 00:14 immediate Y immediate immediate immediate Z 02:16 00:48 immediate Z 01:21 immediate immediate Settling time in hh:mm Recommendation: to power the unit ON for a few days before launch to improve the performance of initial acquisition following separation with the launcher.
Which gyro for AOCS? And how to use it? Low-class accuracy gyros do not bring additional performance to an AOCS Control based on State-of-the-art Star Trackers, and their bias can be continuously monitored using Star Tracker measurement, due to the very low power consumption. Their main role is to support : - The initial acquisition after separation - The safe-mode (whenever magnetic is not feasible due to altitude) - Eventually, short term attitude integration during spacecraft reorientations with STR blindings (whenever singlehead configurations are used). In addition, cold redundancy is sufficient for most of the missions, and charge trapping (bias evolution at switch on) is, in this configuration a second-order issue, for the following reasons : - A short on-period duration few days before launch assures the performance of the initial acquisition. - Once on-station, since the nominal gyro on station is not used for control (but continuously monitored), it cannot be the cause of an anomaly : Safe-Mode can rely on the nominal calibrated and previously-on gyro. - Monitoring shall be activated on the nominal gyro in order to reconfigure on the redundant if needed. However the probability of a gyro failure DURING a safe mode is negligible during a one-day safe mode (0.999988 using 500 FITs)
Role of ESA TEC-EC Control Hardware Lab Constant Technical Management from ESA-ESTEC Control Systems Division since 2005. Hands-on support in TEC-EC Control HW Lab started in 2010 with the procurement of an EM: Independent verification of bias, scale factor, switch-on to-switch on bias Characterization of behavior in temperature Independent re-tuning for optimized performance In-house EGSE development with auto-testing capabilities Characterization of charge trapping (behavior at power ON) and long-term drift Support to projects through real mission profiles stimulation MEMS Gyro as tested in ESTEC Thermal Chamber Outcomes: Active role in the several investigations performed New tuning procedure for FMs based on Lab work
Performance Summary Requirement Specified Measured on EQM ARW Angular Random Walk RBD Bias over temperature SSC Switch on to Switch ON (after power cycling) 0.24 deg/ h 0.009 deg/ h 10 deg/h 40 deg/h (FPGA) 10 deg/h (ASIC) 10 deg/h 2 deg/h Angular Rate Bias 10 deg/h Max constant drift of 0.1 deg/h/day Periodic bi-annual calibration allows to remain within the spec Scale Factor Linearity < 5000 ppm < 1000 ppm Resolution/LSB 0.27 arcsec/s 0.27 arcsec/s Shocks 1500 g 1000 g passed 1500 g survived but bias change Random Vibrations 25.1 g rms 25 g rms survived but bias change Temperature range (Qual) -40 to +70 C -40 to +60 C (FPGA) -40 to +70 C (ASIC)
Short Term future activities ASIC Based-EQM Qualification The ASIC is now developed (ATMEL 0.18 um) and is fully functional. TRB of the ASIC has been passed successfully. It replaces the FPGA RTAX2000 from ACTEL/Microsemi. Opens the door to the ITAR-Free product removing the most critical component Performance improvement: 4th order fit for the thermal calibration Power consumption drastic decrease (improves reliability) An EQM will qualify the ASIC version in 2013. Improve post-shock & vibrations performance through new mechanical design foreseen for GEO (bigger walls covering 18yr + Electrical Orbit Raising) on EDRS Other activities Improve screening and performance assessment of the detectors at detector manufacturer level (AIS UTC). Availability of exportable qualification test report and evaluation report to be finalized. Longer Term: Phase 4 activities for product competitiveness (e.g. Front-End ASIC development, productionisation / industrialisation).
Lessons Learnt Unit cost is not only driven by performance, but also by the required reliability specifications SiREUS is (and has been) a pioneer programme, with a performance improvement between the commercial automotive and space product of 1 to 2 orders of magnitude. At the time, the only solution to meet the required performance was to build a specific MEMS for the space market, but a posteriori, this decision has had two drawbacks : Loss of heritage for the space project Discovery of performance-related issues that were not measurable (and not of interest) on the automotive devices (i.e. charge trapping or long term drift) Today, the improvement in robustness of Star Trackers and more a tailored use of the rate sensor when it is really needed allows worst performance requirements. This trend is shared by many industries and agencies. Therefore, future developments should : Target similar performances for the space unit than existing ground products Re-use the same MEMS component to benefit from mass production Use a similar flow of evaluation program developed for MEMS devices Use representative electronics for detector screening Maintain the strong reliability requirement key for a safe mode gyro
Conclusions On the detector: Detector evaluation completed ESA SOW. Final documentation still being reviewed / finalised with TEC-Q components, radiations experts, and technical officer. A PAD will be derived extracting all the results in an exportable format. Most of the effects being investigated are long term effects (bias drift stability), all requiring very long time scales (weeks) requiring more time being spent on the detector evaluation as originally foreseen. On the equipment: Depending on how the equipment is used on board, the Charge Trapping effect could have not any single impact. Screening process on detectors will be an important axis of improvement. 8 units are under manufacturing (2 ESA FS and 2 non-esa FS) First SIREUS FMs are for Sentinel-3A, with delivery early 2013. Several other (ESA and non-esa) projects are in negotiation Phase