LabOSat as a versatile payload for small satellites: first 100days in LEO orbit
G. A. Sanca, M. Barella, F. Gomez Marlasca, G. Rodríguez, D. Martelliti, L.Patrone, P. Levy, F. Golmar
LLabOSat as a versatile payload for small satellites: first 100days in LEO orbit
G. A. Sanca (1), M. Barella (2,3,4), F. Gomez Marlasca (4), G. Rodríguez (2), D. Martelliti (2), L.Patrone (2), P. Levy (3,4), F. Golmar (1,2,3) (1) ECyT-UNSAM, Martín de Irigoyen 3100, B1650JKA, San Martín, Bs As, Argentina(2) CMNB-INTI, Avda. Gral. Paz 5445, B1650KNA, San Martín, Bs. As., Argentina(3) Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Bs. As., Argentina(4) DMC, GAIANN, CAC-CNEA, Avda. Gral. Paz 1499, 1650, San Martín, Argentina*[email protected]: In this work the first results obtained by LabOSat-01 platform are presented. This platform wasdesigned for testing custom devices on board of small satellites. Two LabOSat-01 type boards were launchedand placed into Low Earth Orbit (LEO) on May 30, 2016. We present here an analysis of data collected byone of these boards during the first days of mission. Total Ionization Dose results are compared with dataacquired by LabOSat-01’s predecessor board, MeMOSat-01, launched in 2014.
1. Introduction
LabOSat (acronym for “Laboratory On a Satellite”) is a universal electronic platform designed to performexperiments in harsh environments, such as outer space [1]. LabOSat-01, the first version of the platform,was designed for testing custom devices on board of small satellites. Two LabOSat-01 type boards werelaunched and placed into Low Earth Orbit (LEO) on May 30, 2016 in two different nano-satellites, Ñusat-1and Ñusat-2, developed by the Argentinian company Satellogic [2]. The LabOSat-01 board is divided into different subsystems. The “MeMO” subsystem is dedicated toexperiments on custom made ReRAM devices, a novel technology based on the Resistive Switchingphenomena on metal-oxide-metal capacitor like devices. Another subsystem, called “xFET”, was speciallydesigned to test any transistor device. The goal of these two subsystems is to test devices’ degradation underhostile conditions. Stimuli applied to Devices Under Test (DUTs) can be either sourcing voltage or current.The MeMO subsystem allows operation over 30 devices while the xFET subsystem allows over 7. Both ofthem work as Source-Measure units (SMUs). Although the satellites have been in orbit since June 2016, first experiments began in November aftersatellite’s boot-up routines. After more than 100 days of operation in LEO, telemetry data and results fromthe experiments running inside LabOSat-01 are analyzed and discussed.
2. Background: MeMOSat-01
LabOSat’s predecessor, MeMOSat-01, is a CubeSat-like platform which was specially designed to measureReRAM devices in LEO. It was installed inside BugSat-1 and is working since June 2014. Besidesmeasuring the I-V curves of the DUTs, MeMOSat-01 senses Total Ionization Dose (TID) using COTS pMOStransistors. The principle of operation and the design of those subsystems in LabOSat were inherited fromMeMOSat-01, nonetheless they were improved to cope with new requirements.
3. Brief description of the platform3.1 Hardware abOSat-01’s communication and experiment modules are commanded by the ultra-low-powermicrocontroller (MCU) MSP430F1612, frequently used in LEO CubeSat-like missions. Guertin et al . [3]showed that this MCU presented no failures up to 20 krad of Total Ionization Dose. Calculations performedby Samwel et al . [4], [5] suggest that expected TID for LEO must be at maximum 10 krad/yr and stronglydecreasing depending on the Al shielding. At the present time, measurements realized by MeMOSat-01 inBugSat-1 [also presented in this proceeding book – Lipovetzy et al.] show that TID is approximately 1.2krad/yr, at a 480 rad/yr rate, which implies that the MCU is suitable for LabOSat’s 3-year mission.LabOSat’s SMUs are capable of exciting the DUTs with voltage or current. MCU’s 12-bit DACs outputs areused to control voltage or current sources that force on-board devices. The 12-bit ADCs inputs are used tomeasure the response of the DUTs and other parameters that are periodically checked to guarantee the correctbehaviour of the board. The DACs and ADCs are configured to use the internal 2.5 V voltage reference. Toincrease the dynamic range in measurements and avoid out-of-range ADC readings, amplifier or attenuatorstages are used. GPIO ports are used to switch between subsystems or modes of operation.The MeMO subsystem was designed to perform experiments over memory devices, particularly ReRAMdevices, a technology believed to be intrinsically resistant to radiation [6]. In general, any 2-terminal devicecan be excited and measured. Another experiment is running aboard LabOSat, designed to study theperformance of transistors. This subsystem is named xFET and basically is able to test any 3-terminal device,commercial or custom made.A more detailed description of the LabOSat’s subsystems, like ReRAM and xFET modules, and condition-signal measurement can be found in a previous work made by Barella et al. [1].
LabOSat-01 has COTS pMOS transistors, characterized before integration to measure TID. Its exposure toionizing radiation causes, among other effects, a shift of the threshold voltage (Vth). This effect is explainedby charge accumulation in the traps of an oxide. Electron-hole pairs are created when a charged particlepasses through the gate oxide of MOS transistors. In silicon oxide, electrons has a higher mobility than holesso they are swept out of the oxide, typically in a picosecond or less. However, in that first picosecond, afraction of the electrons and holes will recombine. The holes, which escape initial recombination, arerelatively immobile and remain near their point of generation, where they cause a negative threshold voltageshift in a MOS transistor [9].Hence, as Vth is affected by TID, its shift can be used to estimate the dose absorbed by the board. Thisparameter is measured indirectly by biasing the source of the pMOS transistor with both known current andvoltage (IS and VS) while the drain is grounded. Then, reading the gate voltage allows to observe variationsin VGS, which will be traduced to Vth variations [7], [8].Figure 1 shows a simplified schematic of the dosimeter’s configuration. LabOSat has four on-boarddosimeters and the possibility to connect other four external dosimeters, to be located in strategic spots. Inorder to measure all of them, multiplexers are used. In Ñusat missions, only two on-board dosimeters wereplaced in each board. Also, a resistor was welded as control.As shown in the schematic, the voltage is measured at the gate of the transistor. Since it is a pMOS transistor,drain is grounded and VGS voltage which is used to measure TID must be negative. In this configuration, thesource voltage value is determined by the voltage divider formed by R1 and R2 resistors. Thus, the referencevoltage, VREF, is a fraction of the supply voltage. igura 1 . Simplified schematic of the dosimetry module.
As the gate voltage must exceed 2.5V, it is out of the ADC reading range. To solve this, an attenuator stage isused, as shown in the Figure 2. Finally, VGS is determined by V GS = V G − V S = V ADC − α V DD ( ) here α is a fraction given by the voltage divider. The variation in time of the voltage VGS will beproportional to the incident TID, according to the sensitivity of each dosimeter. To improve accuracy in themeasurement, corrections using measured temperature and power supply are possible. Firmware was designed to perform three major tasks: communicate with the satellite’s main computer,control the experiments and process data acquired during the experiments. Communication with the satellite,physically implemented with a 200 kHz 3-wire SPI interface, allows LabOSat-01 to receive command listsand to send data packages. LabOSat-01 is either waiting for a new command list (as an idle state) orexecuting one. There are two types of command lists: Arbitrary Lists (generated on the ground, and thenuploaded) and Standard Test Requests (automatically generated by the satellite).Arbitrary Lists are meant to command custom experiments and/or firmware maintenance chores. Thefirmware embedded in LabOSat-01 is capable of processing these Arbitrary Lists, which can be put togetherto perform a wide variety of tasks; e.g., updating/fixing firmware, perform different experiments on theDUTs, ask for telemetry data, run self-integrity tests, etc.Standard Tests are run, at most, one time a day. The satellite is programmed to send Standard Test Requestsperiodically to LabOSat-01; if one day has passed since the last Standard Test execution, LabOSat-01 allowsthe request and executes a new Standard Test. Standard Tests consist of a set of preconfigured experimentsand activities that must be sequentially executed.Among the scheduled activities there are battery voltage measurements, temperature and dosimetry readings,firmware self-integrity checks (using a 16-bit CRC routine), etc. Firmware in LabOSat-01 was tailored toexecute different kinds of standard experiments in every type of DUT onboard. The single-bit ReRAMmemory banks are characterized by performing I-V curves measurements and endurance tests. xFETtransistors are characterized with different families of I-V curves. . Devices Under Test4.1 Single-bit ReRAMs
Tested custom ReRAM devices are tri-layer structures composed of two metallic electrodes divided by a thinfilm oxide fabricated at our laboratories. The studies are focused on non-stoichiometric titanium oxide (TiO ) and manganite La Ca MnO . The devices’ functionality rely on the Resistive Switching phenomenon.When an electric field is applied cross the dielectric it produces mixed filamentary structures formed byvacancies of the oxide cell and metal atoms from the electrodes. This structure reduces the resistance of thedielectric oxide cell, modeled as a dielectric plus a parallel resistance of lower value. In some cases,depending on the applied electric field, the filamentary structure could short circuit the electrodes and leavethe device in a very low resistance state.The switching between different non-volatile resistance values or “states” is exhibited when this filamentarystructure is disarmed (or even fused) and formed again. This is accomplished by applying voltage or currentpulses of opposite polarity (bipolar Resistive Switching). Into LabOSat-01’s xFET module there are ZnO based Thin Film Transistors (TFTs) developed by E. Lopez et al . [E. Lopez 2016]. TFTs are three terminal field effect devices with a structure assembled by depositionof different types of thin films such as gate, gate insulator, semiconductor channel and source-drain contactsonto a substrate.TFTs operation could be compared with classic MOSFET operation. The threshold voltage concept could beapplied to TFTs too and operation regions are the same: cut-off, linear or triode and saturation. The analysisof the current-voltage characteristics of TFTs are similar to those for conventional MOSFETs. The goal is thein-orbit observation of V th shifts in correlation with TID. The main hypothesis is that the film gate oxide (25nm) will not be able to collect charge so a change in the threshold voltage is unlikely to occur.To determinate the threshold voltage, the saturation region is taken into account. The drain current, I D ,depends quadratically on the gate-source voltage, V GS . By linearizing, the relationship between I D and V GS : I D = k ( V GS − V t h ) (2) √ I D = √ k V GS − √ k V th ( ) where √ k is the slope and − √ k V GS the intercept.
5. Results
Presented results in this work are partial results. The experiments are intentioned to last the 3-years missiontime. The elapsed time is not significant in order to conclude nothing about DUTs performance in LEO.Although, it is important to remark this preliminary results are useful to improve our development process,both with LabOSat systems and with ReRAM devices. n the present work, collected data between February 3, 2016 and February 28, 2017 is analyzed. The firstfour reports (listed from 3 to 6) were generated in our labs, in order to test the correct platform functioningbefore delivering the board for integration into satellite. Reports number 7, 8 and 9 were generated afterintegration and before launching. In Figure 2, report’s chronology is represented. In brackets report numbersare presented corresponding to each milestone, along with a brief description.The lift-off took place on May 30, 2016. After four months, on September 7, LabOSat was switched on forthe first time in LEO at 498km (average) and tests started to run periodically. Data generated by the platformis downloaded within satellite’s telemetry. This information include temperature, GPS coordinates, powersupply, Cyclic Redundancy Check (CRC), dosimetry, response of the DUTs.
Figura 2 .LabOSat’s timeline 2016-2017. Milestones (report number).
Temperature measurements are performed both by LabOSat platform and the satellite. The sensor reportstemperature with a resolution of 1ºC in range between -50ºC and 127ºC. Future missions will improve theresolution. Temperature measurements taken before the end of a Standard Test routine is always higher thanthe temperature at the beginning of it.Satellite’s temperature measurements showed a mean increase of (5.03± 0.14)ºC between values obtainedbefore and after the Standard Test routine. This increase in temperature is explained by the board warming-updue to power consumption and dissipation.
In order to perform corrections of V
REF variations in dosimetry module, ReRAM and xFET subsystems, it isessential to know exactly the supply voltage of the platform. The battery voltage values are plotted againstreport number in Figure 4. The circled red bullets represent the measurement at the start of the standard test,while the squared blue bullets are the measurements in the middle of the routine.Large variations between days/reports are due to power cycles of the satellite. Within a test, supply voltagepresents small drops which make clear that monitoring this key variable is crucial to perform the mentionedcorrections. igura 3 . Temperature measurements performed by LabOSat and satellite. Upper bound of the gray area representstemperature after a Standard Test and the lower bound represents the temperatures before the test starts. Red dots:temperature measured by LabOSat.
Figura 4 . Supply voltage measurements performed by LabOSat.
According to the simulations of Samwel et al ., TID for LEO with a 1.5 mm thick of Al shielding, which issimilar to the environment where LabOSat is working, will be 1 krad/yr at maximum. Assuming a constantdose during a year and considering the sensitivity of each dosimeter is expected that one dosimeter shifts V GS around 48 mV and the other one around 40 mV, for a period of 175 days. These values represent 79 and 66ADC counts, respectively.As showed in Equation 1, gate-source voltage measurement is performed indirectly. LabOSat senses thevoltage at the gate relative to ground and then subtracts the value of a reference voltage which polarize thesource pin. That reference is not independent of V DD . In this way, the variations in the gate source voltagecaused by V th shifting for the first days of the mission are masked by the variations of the supply voltage.Furthermore, no clear trend was observed in V GS . It is important to remark that although the pMOS transistoris polarized in the Zero Temperature Coefficient (ZTC) point, VGS measurements were found to beuncorrelated with the temperature (results not showed here). n this mission all ReRAM devices were configured to perform current-voltage (I-V) experiments so as toobserve if resistive switching parameters were modified by environmental conditions. In first place, nodegradation was observed over the current and voltage source. Control devices i.e. fixed resistance loads,allowed us to check that characteristics of sources remain the same as measured at our lab. Secondly, somedevices were found to be in high impedance after several I-V loops. After studying the pinout of damaged devices and the packaging properties like epoxy, wire-bonding andpads material, we concluded that high variations of temperature (T) might cause this failure. In spite of beinginside the satellite LabOSat recorded temperature variations ( Δ T) of 15ºC which could mechanically stressthe bonded structure caused by dilation-contraction cycles. Nonetheless, other devices kept switching asshown in Figure 6. Different runs are plotted for titanium dioxide and manganite ReRAM. The TiO devicespresented hysteresis the first 30 days in orbit. After day 31, the issues related with encapsulation of thesilicon die produced a permanent damage, which led to high impedance DUTs. The manganite devicescontinue to exhibit hysteresis and are currently switching between two states. However, switching parameterschanged from our lab tests as can be seen in the I-V curves. We attribute this behavior to the intrinsicvariations that suffer low endurance ReRAM devices. This effect may not be related to incident radiationbecause dosimetry does not measure significant TID.
Figure 6 . Current-voltage response of ReRAM devices for diferent days in LEO.
Left: titanium dioxide ReRAM.Right: manganite ReRAM.
The analysis results of the first 175 days indicates that there is no observable V th shifting in the TFT devices.However, taking into account the dosimeters measurements, it is expected that the incident TID does notgenerate changes in the threshold voltage yet. This is just a partial analysis since it is a 3 years mission.Nevertheless, we expected that the thin gate oxide will not accumulate electrical charge so V th shifting willnot occur.In Figure 7, the I D vs. V DS curve for one device is shown for different reports. Also, threshold voltage wascalculated from the characteristic output curve using Equations 2 and 3. The time evolution of this parameterdoes not exhibit a clear pattern, if there was one. Figure 7 shows the histogram of the extracted V th for thesame device with an inset of its value as a function of report number.
6. Conclusions
LabOSat has been working in LEO for more than 170 days. Regarding performance, all subsystems areworking as expected. On Earth calibration, supply voltage and temperature measurements in orbit allows usto carry out corrections to improve accuracy in DUTs experiments.
TID measurements are still inconclusive,ut consistent with the expected results, signaling very lows radiation levels. ReRAM devices still exhibithysteresis. Variations of I-V curves are due to the low endurance instead of absorbed dose. It is important toremark that this is a 3-years mission, so the results presented here are partial results.
Figure 7 . Histogram of the extracted threshold voltage for one device. Inset: time (report number) evolution of thisparameter.
6. Acknowledgment
The authors would like to thank to R. Ferreira, E. Paz, W. R. Acevedo and D. Rubí who participated in the fabrication ofReRAM devices; P. Stoliar, L. Hueso and E. Lopez who were involved in fabrication of xFET devices; M. G. Inza whofacilitated the dosimeters. The authors acknowledge financial support from ANPCyT PICT 2013-0788 ”MeMOSat” andUNSAM-ECyT FP-001.
7. References