Sensors and IMUs

Inertial and timing parts characterized against a named beam, with what the campaign covered and what it left open.

No inertial part flown on a planetary robot has a published qualification campaign behind it. What the literature does hold is a heavy-ion latchup screen on two commercial MEMS accelerometers, run for a JPL avionics build [1]. The latchup cross-section of one of them moves by about a factor of four between room temperature and 75 C, so a part screened warm and a part screened cold do not return the same answer.

MEMS parts screened for single-event latchup

Part	Manufacturer	Facility	Source
MS1002A MEMS accelerometer	Colibrys	Brookhaven National Laboratory	[1]
SF1600S MEMS accelerometer	Colibrys	Brookhaven National Laboratory	[1]
ASDMB 16 MHz MEMS oscillator	Abracon	Brookhaven National Laboratory	[1]
AD2S80A resolver-to-digital converter	Analog Devices	Texas A and M cyclotron	[2]

Campaign results

• MS1002A MEMS accelerometer, Colibrys. Heavy ion at Brookhaven National Laboratory, the beam line not named in the source. Qualification: single-event latchup cross-section 1.5e-6 cm2 at room temperature and 6.1e-6 cm2 at 75 C, both at LET 37.5 MeV-cm2/mg to a fluence of 1e7 ions/cm2 [1]. The non-destructive latchup LET threshold is 11.44 to 19.71 MeV-cm2/mg above 70 C. Supply current in latchup is 232 mA against 23 mA nominal, and the part recovers on a power cycle [1], so the design question is whether the bus can carry a tenfold current excursion for as long as detection and power cycling take [1].
• SF1600S MEMS accelerometer, Colibrys. Heavy ion at Brookhaven National Laboratory, biased at its maximum recommended voltage. Qualification: non-destructive latchup LET threshold 7.88 to 11.44 MeV-cm2/mg above 70 C, the lowest threshold of the two accelerometers [1]. Supply current in latchup is 340 mA against 20 mA nominal, again recovering on a power cycle.
• ASDMB 16 MHz MEMS oscillator, Abracon [1]. Heavy ion at Brookhaven National Laboratory. Qualification: non-destructive latchup LET threshold 11.44 to 19.71 MeV-cm2/mg [1].
• AD2S80A resolver-to-digital converter, Analog Devices. Heavy ion at the Texas A and M cyclotron. Qualification: no latchup to LET 85 MeV-cm2/mg [2].

What the screen does not cover

The Brookhaven and Texas A and M runs above are latchup screens and nothing else. No single-event upset, functional interrupt or total ionizing dose data was taken, and no event rate was computed, so a part recorded as free of latchup to LET 42.8 MeV-cm2/mg has not been qualified by that result [1]. Two or three parts of each type were tested. The highest LET available in the Texas A and M half of the campaign was 42.8 MeV-cm2/mg, below the 75 to 85 MeV-cm2/mg customary for heavy-ion qualification, so several of the clean results are bounded by the beam rather than by the part [1]. Ion range matters as much as LET for a die under a thick redistribution layer: silver at 2954 MeV reached 113 micrometers in silicon at LET 42.8 MeV-cm2/mg, bromine at 287 MeV reached 36 micrometers at LET 37.5, and one motor driver in the same campaign carried 12.76 micrometers of copper redistribution above its active silicon [1].

Inertial units and attitude sensors carried by flight and prototype programs

None of the units below has a radiation campaign published against it. What each carries is a performance figure taken on a bench or in a field test, and in three cases a heritage claim the project did not test itself [5].

Part	Manufacturer	Used by	Source
LN-200 fiber optic gyro IMU	Northrop Grumman	Lander Vision System prototype	[3][4]
STIM300 MEMS IMU	Sensonor	JPL dead reckoning study	[6]
STM300 IMU	Sensonor	moonranger	[5]
M-G370 IMU	Epson	Int-Ball2	[7]
nano-SSOC-D60 sun sensor	SolarMEMS	moonranger	[5]
ADXL-50 accelerometer	Analog Devices	Micro-rover concept study	[8]
CXTA02 two-axis inclinometer	Crossbow	Pioneer-AT test rover	[9]
Rate and acceleration IMU	Not named	fido	[10]

Ratings and qualification results

• LN-200 fiber optic gyro IMU, Northrop Grumman, formerly Litton. Three orthogonal fiber optic gyros and three silicon accelerometers, about the size of a coffee mug, delivering 400 samples per second over RS-485 [3]. Ratings: angle random walk 0.15 degrees per root hour on the commercial unit used in the Lander Vision System prototype, with the flight unit expected to be at least as good [4]. Qualification: in descent hazard detection testing its attitude output was the difference between 0.063 m and 0.1 m of reconstruction error and cost one detected hazard [3]. No radiation or environmental result is published for it here.
• STIM300 MEMS IMU, Sensonor. Tactical-grade MEMS inertial unit. Ratings: bench data with bias and misalignment corrections applied [6]. Qualification: propagated horizontal position error after ten minutes of dead reckoning was 1.28 km and 2.21 km on the two horizontal axes stationary, falling to 0.126 km and 0.03 km when the unit was rotated [6]. Vertical error stayed near 8 km in both cases, because rotation about the vertical axis does not average the vertical bias.
• STM300 IMU, Sensonor. MEMS inertial unit selected for MoonRanger. Ratings: carried on a heritage claim of 5 krad of total dose, with no test performed by the project [5]. Qualification: none published.
• M-G370 IMU, Epson. Commercial MEMS inertial unit adopted for the Int-Ball2 free flyer. Ratings: 10 g in a one-inch square package drawing 16 mA at 3.3 V, with an angle random walk quoted across the series at 0.03 to 0.06 degrees per root hour [7]. Qualification: none published. The vendor’s bias instability figure is printed in degrees per root hour, which is the wrong unit for that quantity, so the number is unusable as written [7].
• nano-SSOC-D60 sun sensor, SolarMEMS. Two-axis digital sun sensor. Ratings: heritage of 30 krad of total dose and 300 krad of 6 MeV protons, claimed rather than tested by the project [5]. Qualification: none published.
• ADXL-50 accelerometer, Analog Devices. Surface-micromachined capacitive force-rebalance accelerometer with on-chip conditioning and self test. Ratings: a 500 by 625 micrometer sensor with capacitor plates about 115 micrometers long and 4 micrometers wide [8]. Qualification: none published, and the study that names it records that the part is not inertial guidance grade.
• CXTA02 two-axis inclinometer, Crossbow. Pitch and roll sensor on the Pioneer-AT test rover. Ratings: plus or minus 75 degrees at 0.05 degree resolution, a vendor figure not independently verified in the source [9]. Qualification: none published.
• Rate and acceleration IMU, vendor not named, on the FIDO rover. Vibrating ceramic plate Coriolis rate sensors with surface-micromachined silicon accelerometers, sampled at 200 Hz [10]. Ratings: plus or minus 50 degrees per second at about 25 degrees per second per volt, and plus or minus 2 g at about 1 g per volt [10]. Qualification: a stationary Kalman filter converged on rate bias within 0.5 s, to about 0.01 degrees per second in one data set and about -0.02 in another, over nine ten-minute stationary sets taken on different days so that unit temperature varied.

Force, torque and ranging sensors with a measured result

Part	Manufacturer	Used by	Source
Mini-45 six-axis force-torque sensor	ATI	MER-class arm testbed	[11]
UTM-30LX-EW lidar	Hokuyo	LEMUR 3 perception mast	[12]
RVS3000-3D scanning lidar	Jena-Optronik	mev-2	[13]
Suspension torque transducer	Not named	viper	[14]

Ratings and qualification results

• Mini-45 six-axis force-torque sensor, ATI. Wrist force-torque sensor on a Mars Exploration Rover class arm. Ratings: specified at plus or minus 1 percent of full range, 1.45 N and 0.05 N m [11]. Qualification: measured against a single-axis load cell with loads applied on three axes to a 40 N threshold at several workspace locations, giving an RMS error of 0.41 to 1.19 N across runs and least-squares slopes of 0.962 to 0.988 [11]. Sensors were zeroed immediately before each run, so the result bounds short-run accuracy rather than drift.
• UTM-30LX-EW lidar, Hokuyo. Scanning lidar on the LEMUR 3 perception mast [12]. Ratings: 270 degree field of view, swept by the mast for a full 360 degree scan. Qualification: 0.1 to 30 m range with a 30 mm point cloud resolution in the climbing workspace, and operation stated both in total darkness inside a lava cave and in direct sunlight [12].
• RVS3000-3D scanning lidar, Jena-Optronik. Rendezvous lidar on the Mission Extension Vehicle 2, its emitter modified from ISS-approach heritage with an added amplification stage and reduced beam divergence [13]. Ratings: no retroreflectors on the client vehicle, so position comes from the centroid of returns. Qualification: on-orbit tracking of an uncooperative client in three degrees of freedom beyond 2 km, and a six degree of freedom solution with a client model overlaid from 60 m down to 2 m [13].
• Suspension torque transducer, vendor not named, in the VIPER wheel module. Ratings: plus or minus 297 N m at 0.07 N m of resolution, telemetered at 10 Hz [14]. Qualification: described by its own program as a measurement not previously available on a planetary rover, and the enabler for deriving wheel normal load in situ [14]. No environmental or life result is published.

Data converters, references and timing screened at a named beam

Part	Manufacturer	Facility	Source
DAC121S101 12-bit DAC	Texas Instruments	JPL cobalt-60	[15]
RHF1201 12-bit 50 Msps ADC	ST Micro	JPL cobalt-60	[15]
ISL71590SEH temperature transducer	Renesas	JPL cobalt-60	[16]
AD7760 24-bit ADC	Analog Devices	Texas A and M cyclotron	[2]
LTC1419AIG 14-bit ADC	Linear Technology	Texas A and M cyclotron	[17]
LTC2440IGN 24-bit ADC	Linear Technology	Heavy ion, facility unnamed	[1]
ECS-3961-040 4 MHz oscillator	ECS	Heavy ion, facility unnamed	[1]

Campaign results

• DAC121S101 12-bit DAC, Texas Instruments [15]. Ratings: biased, at 100 mrad(Si)/s [15]. Qualification: parametric failure at 100 krad(Si), with severe integral and differential nonlinearity and zero-code failures between 150 and 200 krad, and no measurable recovery after a 168 hour biased anneal.
• RHF1201 12-bit 50 Msps ADC, ST Micro [15]. Radiation-hardened converter. Ratings: degradation depends on how the part is operated. Qualification: integral and differential nonlinearity degrade parametrically from 100 krad(Si), and the degradation worsens with higher sample frequency, higher polarization resistance and higher supply voltage, so differential mode at minimum polarization resistance is the recommended configuration [15].
• ISL71590SEH temperature transducer, Renesas. Ratings: unbiased is the worst case [16]. Qualification: temperature error out of specification at 100 krad(Si), worst error -5.7 microamps at 300 krad, with no functional failure.
• AD7760 24-bit ADC, Analog Devices [2]. Ratings: tested at 25 C and 85 C at normal incidence [2]. Qualification: both temperatures latched up, threshold below 8.3 MeV-cm2/mg, with a saturated latchup cross section of 5e-4 cm2 at high transfer.
• LTC1419AIG 14-bit ADC, Linear Technology [17]. Ratings: the MSL flight lot was tested separately from catalog parts. Qualification: no latchup above 86.2 MeV-cm2/mg at 20 C and above 78.0 at 40 C, but the flight lot latched at 50 to 55 MeV-cm2/mg at 60 C and the event was destructive once current limiting was removed [17]. Supply current in latchup reached 450 mA with the protection defeated and the clamp raised to 1.9 A, and the part did not recover on a power cycle. Temperature moves this part from clean to destructive across a forty degree span.
• LTC2440IGN 24-bit ADC, Linear Technology [1]. Ratings: commercial delta-sigma converter [1]. Qualification: destructive latchup observed, with no threshold published.
• ECS-3961-040 4 MHz oscillator, ECS [1]. Ratings: crystal oscillator on the same JPL avionics screen as the MEMS parts above [1]. Qualification: destructive latchup observed, with no threshold published.

Two of the parts in this last table were destroyed and three were bounded by the highest transfer the beam could reach, so the table separates parts that failed from parts that were not pushed hard enough to fail [1][17].