Radios and Antennas

The transponders, transceivers, amplifiers, switches and antennas that carry a robot’s data off the vehicle, with the mass, power, rate and coding each part supports and the result of any campaign published against it. The rates these parts achieve on a real path are on Link Budgets and Data Rates, the session and framing behavior on Protocols and Standards, and pass geometry and returned volume on Relay Architecture.

Radios and transponders selected and flown

Part	Manufacturer	Used by	Source
Electra-Lite UHF transponder	JPL	curiosity	[1][5]
Electra UHF transponder	JPL	MRO, MAVEN, TGO	[4][6]
CE-505 UHF transceiver	Not named	spirit, opportunity, phoenix	[4][36]
Small Deep Space Transponder	Not named	curiosity	[17]
X-band solid state power amplifier	Not named	spirit, opportunity	[3]
UHF coaxial transfer switch	Not named	curiosity	[1]
QinetiQ UHF transceiver	QinetiQ	Schiaparelli	[6]
SiFlex radio	Not named	ingenuity	[12]
Iris V2 software-defined radio	JPL	MarCO	[15]
DR-2000 transceiver	RF Monolithics	SPHERES	[22]
PicoStation M2HP	Ubiquiti	RASSOR 2.0	[24]
Mesh radio	Microhard	CADRE	[25]
UHF and S-band transmitter	Not named	LEV-1	[26]

Ratings and qualification results

• Electra-Lite UHF transponder, JPL, the rover end of the Mars relay link. Ratings: 3 kg per unit in two redundant units, drawing 21 W standby and 69 W transmitting in surface operations [1]. Against an Electra at the orbiter end it runs return rates in powers of two from 1 to 2048 kbps with (7,1/2) convolutional coding in the 390 to 405 MHz band, a radio capability demonstrated on adaptive-rate passes rather than a rate held across a pass [5]. Qualification: the flight radio’s box to mounting-platform temperature gradient held at 7 degC through the 16 day MSL rover system thermal test in the JPL 25 Foot Space Simulator, essentially unchanged between the cruise vacuum case and the 8 Torr GN2 surface case [9]. The chamber cannot reproduce Mars, so the deliverable of that campaign was a model correlated to within 5 degC rather than a validated design [9].
• Electra UHF transponder, JPL, the orbiter end on MRO, MAVEN and Trace Gas Orbiter. Ratings: forward 2 to 256 kbps and return 2 to 2048 kbps toward an Electra-Lite, with an open-loop recorder taking 8-bit in-phase and quadrature samples at about 150 kHz on MRO and 128 kHz on the MAVEN and TGO units [4]. Above about 32 kbps the user spectrum falls outside that sampled bandwidth, and a 20 minute capture at the maximum rate makes a file over 5 Gb [4]. The baseband processor module is a 24 MHz Sparc V7 with 256 Mbit of error-corrected SDRAM, of which 170.5 Mbit is available for mass data storage [37]. Qualification: interference from two MRO science instruments cost about 10 dB of Electra threshold performance in flight, reduced to about 3 to 4 dB by a post-launch firmware change and an interference-quiet mode during relay passes, against about 0.7 dB on MAVEN [8].
• CE-505 UHF transceiver, manufacturer not named in the sources held. Ratings: forward 8 or 32 kbps and return 8 to 256 kbps, with no adaptive data rate, on both MER rovers and Phoenix [4]. Phoenix carried no X-band link at all, so this was its only path [36]. Qualification: a flight software change that starved the transceiver buffer cut entry telemetry latency from about 32 s to under 5 s at 8 kbps, and from 8 s to under 2 s at 32 kbps, so that telemetry would not be stranded in the buffer at loss of vehicle [7].
• Small Deep Space Transponder, Curiosity’s X-band direct-to-Earth chain. Ratings: the auxiliary oscillator frequency against temperature was fitted as a cubic with R squared 0.9977 against the oscillator sensor and 0.9984 against the baseplate sensor [17]. Qualification: the fit was made to cruise data in thermal equilibrium and is not a prediction of the entry excursion; what that cost the entry link is on Link Budgets and Data Rates.
• X-band solid state power amplifier, on Spirit and Opportunity. Ratings: 16.8 W of radio-frequency output, 42.25 dBm, inside an X-band subsystem drawing 71.8 W maximum and massing 5.367 kg [3]. Qualification: no campaign against this part is published in the sources held.
• UHF coaxial transfer switch, on Curiosity. Ratings: 0.130 kg, drawing 7 W for a 0.2 s pulse at 28 V and 0.25 A, operable from 22 to 36 V [1].
• QinetiQ UHF transceiver, on the Schiaparelli entry demonstrator. Ratings: 4.8 W transmitter power, return link at 401.585625 MHz from 8 to 1024 kbps in powers of two with optional rate 1/2 convolutional coding, forward link uncoded at 8 to 64 kbps on 437.1 MHz [6]. Qualification: the lander did not survive landing, so the flight record is the entry pass only.
• SiFlex radio, on Ingenuity. Ratings: about 1 dBi of antenna gain and 26.8 dBm of output as reported for the flown configuration [11], against a maximum transmit output of 28.8 dBm in the pre-flight design paper [10]. Qualification: the finding recorded against the part is obsolescence rather than performance. It is no longer available in the condition and quantity required, and the Sample Recovery Helicopter design moved to an improved radio built on the same design [12].
• Iris V2 software-defined radio, JPL, on the MarCO CubeSats. Ratings: 8 kbps at 1.05 AU with 4 W of X-band output through an external solid state power amplifier [15]. That rate is quoted without modulation, coding, ground aperture, elevation or margin, and pre-flight.
• DR-2000 transceiver, RF Monolithics, on SPHERES. Ratings: 916.5 MHz and 868.35 MHz, 18 kbps, 70 packets per second [22]. Qualification: the units reached the International Space Station in 2006 and had run more than 70 test sessions by 2015 [23].
• PicoStation M2HP, Ubiquiti, the ground station radio for the RASSOR 2.0 excavation prototype [24]. Qualification: a terrestrial commercial part on a ground prototype.
• Mesh radio, Microhard, on the CADRE rovers. Ratings: 1 Mbps minimum aggregate shared across three rovers and the base station, reduced further by motor interference while driving [25].
• UHF and S-band transmitter, on the LEV-1 lunar hopper. Ratings: 90 g in 60 by 40 by 25 mm, transmitting UHF and S-band and receiving UHF, claimed by its team as the smallest and lightest to transmit from the lunar surface [26]. Qualification: it operated on the surface, with S-band received at Usuda and Uchinoura from 380,000 km [26].

MER’s rover UHF transceiver supported 8, 32, 128 and 256 kbps on both the forward and return link, with every link opened by the orbiter at 8 kbps [2]. The MINERVA-II rovers reached up to 32 kbps at over 20 km to the Hayabusa2 relay, half duplex on one frequency, time-division controlled by the relay so both rovers could work at once, with antennas on the top and bottom faces selected automatically by received signal strength [27].

Antennas

Part	Type	Used by	Source
KaPDA 0.5 m Ka-band antenna	Deployable parabolic mesh	RainCube	[13]
Electra quadrifilar helix	Quadrifilar helix	TGO	[6]
Surface platform low gain antenna	Quadrifilar helix	Schiaparelli	[6]
UHF monopole and UHF helix	Monopole and helix	phoenix	[7]
Helicopter Base Station Antenna	Monopole, 914 MHz	perseverance	[10]
Helicopter monopole	Monopole, 914 MHz	ingenuity	[11]
Backshell wrap-around antenna	Three-panel wrap-around	insight	[14]
High gain reflectarray	Flat reflectarray	MarCO	[15]
4.2 m relay antenna	Parabolic	Queqiao	[16]

• KaPDA 0.5 m Ka-band antenna, JPL, a parabolic mesh reflector deploying from a 10 by 10 by 15 cm CubeSat volume for RainCube [13]. Ratings: 32 to 36 GHz operation demands 0.4 mm RMS deployed surface accuracy, six times finer than the 2.4 mm RMS its S-band predecessor measured, and a 40 openings per inch mesh tensioned to 17.5 N/m, which takes about 250 N of deployment preload. Qualification: see the campaign below.
• Electra quadrifilar helix, on Trace Gas Orbiter with the flight-redundant Electra transceivers. Ratings: 6 dBic on boresight with a 3 dB beamwidth of plus or minus 40 degrees [6].
• Surface platform low gain antenna, an upward-looking quadrifilar helix on the Schiaparelli surface platform at the 401.585625 MHz return frequency [6]. Ratings: 6.2 dBic on boresight, against 1 to 2 dBic anti-velocity and 0 to -6 dBic at 45 degrees off axis for the backshell antenna used during entry [6].
• UHF monopole and UHF helix, both flown on Phoenix. Qualification: measured against each other in flight by elevation. The monopole performed better below 20 degrees and the helix better at high elevation, and because the monopole pattern was distorted by its ground plane and its polarization loss varied, its data volume predictions were unreliable and the circularly polarized helix became the operational workhorse [7].
• Helicopter Base Station Antenna, on Perseverance. Ratings: a vertically polarized monopole confined to a 5.6 by 5.6 cm ground plane, with S11 below -10 dB from 906 to 924 MHz, handling 1 W, required to survive -135 to 70 degrees C along with launch vibration and pyrotechnic shock [10]. Qualification: gain and S11 were measured on engineering and flight models mounted on rover and helicopter mockups [10].
• Helicopter monopole, on Ingenuity. Ratings: the whole telecom subsystem including radio, cables and antenna was held to 14 g, of which the antenna is 4 g or less, against a typical NASA flight radio in excess of 10 kg [10]. Qualification: the flown gain is about 1 dBi [11].
• Backshell wrap-around antenna, on the InSight entry vehicle. Ratings: three panels around a 45 degree half-cone backshell, giving about 135 degrees of usable gain about the backshell point [14]. Qualification: the deepest off-boresight null came from a mismatch between two of the three panels, about 45 degrees off the backshell point, and its depth is not quantified in the source [14].
• High gain reflectarray, on MarCO, a flat panel carrying the 8 kbps X-band downlink at maximum range [15].
• 4.2 m relay antenna, on Queqiao, parabolic and the largest communication antenna flown in deep space at the time, above 44 dBi at X-band [16]. Qualification: its narrow beam requires continuous guidance-driven pointing at the lander and rover, and with a directional surface dish the return runs at 285 kbit/s from the rover and 555 kbit/s from the lander, falling to 700 and 1400 bit/s on omnidirectional surface antennas [16].

The one full part-level campaign: KaPDA

KaPDA went from proposal to flight qualification in four years, and every failure the campaign found was a mechanism failure rather than a radio-frequency one [13]. Constant force springs in the mid-rib hinges lost their torque margin after the first axis of 14.1 G RMS qualification vibration: friction between the unrolled spring and the hinge root had been adding tension in the stowed state, and vibration let the spring relax to its lowest energy state so the same friction then resisted deployment. The fix was kick-off springs, and later a change to the hinge geometry. A deployment motor controller with a poor thermal path to the chassis stalled the deployment two thirds of the way through at 65 C ambient in thermal vacuum, against an 85 C part limit, so the qualified deployment range was cut from 10 to 50 C down to 10 to 40 C [13]. A press fit between a 300 series austenitic stainless sun gear and a 400 series martensitic stainless bearing went loose hot and let the gear walk off the bearing, with analysis showing it would crush the thin-section bearing cold; the joint was athermalized with an enlarged bore and a 0.4 mm epoxy bond line.

The flight model then measured 42.6 to 42.7 dBi on a radio-frequency range against a 42 dBi requirement, unchanged after deployment, after vibration and after thermal vacuum [13]. An earlier configuration that did not latch fully reached only 42.0 dBi, exactly meeting the requirement, which makes the latch state the difference between margin and none. The evidence base is one engineering model and one flight model, with no torque margin figures given and no on-orbit performance, since the document was written before launch [13].

Two other part-level results exist and neither is a radiation or life test. The Ingenuity radio’s receiver sensitivity thresholds were measured in the laboratory at -99 dBm for 250 kbps and -108 dBm for 20 kbps [10], and the same thresholds govern operations: -85 dBm or better supports the full rate, and below -105 dBm no link is supported [11]. Curiosity’s UHF radio was measured at the box level in a system thermal vacuum test [9]. The longest range these radios have demonstrated between two vehicles on Mars is 2.96 km, on a contact made on 26 November 2024, more than six months after Ingenuity stopped regular operations [31].

Ground apertures

The Deep Space Network held 12 active antennas across Goldstone, Madrid and Canberra as of early 2020, expected to reach 15 by 2024 [20]. Six 34 m beam waveguide antennas were funded out of operations savings, over a period in which staffing fell from 14 operators per antenna in 1980 to 4 by 1990, and the links one operator runs went from 1 before 2012 to 2 from 2012, with 3 in work [21]. The measured penalty for arraying two 34 m antennas against a single 70 m dish is on Link Budgets and Data Rates.

Aperture becomes the binding constraint when a vehicle’s own antenna pointing goes wrong. After IM-1 Odysseus tipped over at Malapert A, the 34 m stations could not reliably close the S-band downlink and the 70 m antennas were committed elsewhere, so the 64 m Parkes telescope was required [18]. Frame error rate on the 34 m stations ran from 0.43 to 1.00 over days L+3 to L+6, against 0.0001 on a 34 m station five days before landing [18]. Parkes had been rebuilt for Mars support: the X-band receiver installed in August 2003 ran a system temperature of about 25 K over a 50 MHz bandpass centred on 8.4 GHz, and the combined surface and receiver upgrade was stated as about 6 dB of sensitivity gain, without measurement conditions [19].

Radio telescopes also serve as carrier-detection assets during entry. A 12 antenna sub-array of 45 m dishes at the Giant Metrewave Radio Telescope, equivalent to a single 156 m dish, was predicted to receive the Schiaparelli carrier only a few dB above the array noise floor, enough to detect the carrier and not to recover data [6]. The InSight entry analysis reached the same conclusion for Green Bank and Effelsberg: the direct-to-Earth UHF path was never capable of carrying the 8 kbps telemetry and both telescopes sat below the detection threshold in the first-percentile Monte Carlo case [14].

Surface and short-range radios

Range figures for commercial short-range radios come from terrestrial field tests, and the geometry that defeats them is the vehicle’s own structure. An Axel rover 802.11n link matched its specified throughput to 30 m, degraded rapidly beyond that and failed outright by 90 m, with packet loss approaching 100 percent between 90 and 100 m on outdoor line of sight with 9 dBi antennas [29]. Covering the same antenna with an aluminum plate standing in for the rover body cost essentially all of the signal [29]. A MoonRanger Wi-Fi board with two switched omnidirectional antennas reached 160 m in the team’s own field test, double the 80 m the lander guaranteed, at a 30 percent drop from peak short-range performance and in an unstated terrestrial environment with no vacuum, dust or thermal conditions [28]. Mesh radios from Silvus and Persistent Systems reached 400 m of line of sight in smooth-walled NIOSH mine tunnels and still needed droppable relay nodes for side passages [30].

Data interface and radio-frequency parts screened at a named beam

The link parts above have no radiation screening published against them. The devices that carry data across an isolation barrier or a differential pair do, and they belong to no other page in this tree.

Part	Manufacturer	Facility	Source
IL715T GMR digital isolator	NVE	Naval Research Laboratory	[34]
IL815T GMR digital isolator	NVE	JPL cobalt-60	[34]
DS90LV010A bus LVDS transceiver	Texas Instruments	Heavy ion, facility unnamed	[33]
DS90LV032A LVDS quad receiver	Texas Instruments	Heavy ion, facility unnamed	[33]
MAX2112 tuner	Maxim	Texas A and M cyclotron	[32]
SJA1000 CAN controller	Cobham	No radiation result published	[35]

• IL715T GMR digital isolator, NVE. Giant magnetoresistive digital isolator. Ratings: ten devices at each supply voltage [34]. Qualification: functional failure at 120 krad(Si) at a very high 240 rad(Si)/s, two of ten failing at 5 V and seven of ten at 3.3 V, with nearly all failed by 150 krad [34].
• IL815T GMR digital isolator, NVE. Ratings: the failure level depends on how two pins are tied. Qualification: parametric and functional failure at 100 krad(Si) at 100 mrad(Si)/s and 5 V with sync pulled high and output enable grounded, supply currents and timing out of specification [34]. Both pins pulled to 5 V is the worst case and the recommendation is to fly with both grounded, so the part’s dose capability is a property of the board around it.
• DS90LV010A bus LVDS transceiver, Texas Instruments. Ratings: screened for a JPL avionics build [33]. Qualification: non-destructive latchup below 42.8 MeV-cm2/mg.
• DS90LV032A LVDS quad receiver, Texas Instruments. Ratings: same campaign as the transceiver [33]. Qualification: non-destructive latchup below 42.8 MeV-cm2/mg.
• MAX2112 tuner, Maxim. Direct-conversion satellite tuner. Ratings: commercial [32]. Qualification: upset threshold below 6.3 MeV-cm2/mg with a saturated cross section about 2e-5 cm2, and no latchup to 83.2 MeV-cm2/mg at 25 C over 1e7 ions/cm2 [32].
• SJA1000 CAN controller, Cobham, with the UT64CAN333x transceiver. Ratings: total input and output delay 40 ns in the controller and 745 ns plus 1.5 microseconds in the transceiver, both datasheet values [35]. Qualification: no flight part was tested in the source that names them, which quotes the figures only to compare against an Arduino MCP2515 at 200 to 2100 ns and an MCP2551 at 1240 ns plus 5 microseconds.