Chapter 3
RA-2 and MWR Instruments

3.1 RA-2 Instrument Description

RA-2 is a nadir-looking pulse-limited radar altimeter based on the heritage of ERS-1 RA functioning at the main nominal frequency of 13.575 GHz (Ku Band), which has been selected as a good compromise between the affordable antenna dimension that provides the necessary gain and the relatively low attenuation which experience the signals propagating through the troposphere.

A secondary channel at a nominal frequency of 3.2 GHz (S band) is also operated to estimate the errors on range measurements caused by the propagation of the radar signals through the ionosphere.

At the main operative frequency RA-2 shall autonomously detect, acquire, lock-on and track the earliest part of the radar echoes from ocean, ice and land surfaces without any interruption, irrespective of sudden changes in surface characteristics and elevation; after successful acquisition RA-2 shall autonomously start the tracking. Operations shall be accomplished by automatically changing the range resolution, the width, the position and the overall gain of the radar tracking window. The tracking shall always be performed with the highest resolution that allows the earliest part of the radar echoes to be maintained within the radar tracking window, in order to continuously measure on board their power levels and time positioning with respect to transmitted pulse power and time. Over ocean the resolution shall always be the highest available on-board. Furthermore RA-2 shall detect whenever the earliest part of the radar echo is no longer within the tracking window and autonomously recover from this condition.

The estimates of the time delay, radar cross section (σ0) and the standard deviation (σs) of the height distribution of the scatters on the Earth's surface are performed on ground by fitting to the samples of radar echoes, for both the Ku and S band, the model of the shape of the radar echoes from ocean (G. S. Brown). From these parameters, the satellite altitude, the wind speed magnitude, and the wave height of the oceans can be respectively retrieved. The estimations of these parameters can be averaged over a period of approximately 1 second to reduce random fluctuations.

The RA-2 transmits with constant repetition rates radar pulses of 20 μsec length; namely one Ku pulse is transmitted every 557 μsec, and one S pulse is transmitted every 4 Ku consecutive pulses which corresponds to a pulse repetition interval of 2228 μsec.

At Ku band pulses may be unmodulated (CW pulse) or linear frequency modulated with bandwidths selected among the three (320 - 80 - 20 MHz) available to adapt the range resolution to the observed scenario. In particular use of CW pulse is foreseen during the acquisition phase of the measurement mode which is required to initialize the tracking measurement phase. On the contrary, at S band a unique linear frequency modulated pulse 160 MHz bandwidth is used.

RA-2 is composed by the following sub-systems: Antenna, Ku-Band Front End Electronics (KFEE), S-Band Front End Electronics (SFEE), Ku-Band Transmitter (KTx), S-Band Transmitter (STx), Microwave Subsystem (MR), Frequency Generation and Conversions Unit (FGCU), Chirp generator (CG), Signal Processor Subassembly (SPSA), Low Voltage Power Supply (LVPS) and the Instrument Control Unit (ICU). A block diagram of the radar is shown in fig. 3-1. All the Subsystems with the exception of the Antenna are redounded to improve instrument reliability.

Each Front End connects the related transmitter and receiver input with the Antenna. Its main purpose is to isolate during transmission the high sensitive receiver and prevent it from being damaged by the high peak power level of the transmitted waveform which is however also injected in the receiver through a well controlled coupling path inside the FEE to calibrate the radar. On reception, when the transmitter is off, the FEE routes to the receiver the weak radar echoes impinging on the antenna through a very low loss path. In the MR each received radar echo is mixed with a delayed replica of the transmitted chirps (deramping operation) and down-converted to an Intermediate Frequency (IF) where signals can be more easily amplified and then split into their in-phase and quadrature components (I &Q) and filtered to 6.4 MHz. An Automatic Gain Control (AGC) adjusts the whole value of the receiver amplification to maintain the I and Q components at a constant and suitable level for the sampling.

FGCU provides all the frequencies which are necessary to the instrument. It contains the Ultra Stable Oscillator (USO) which is the frequency reference of the instrument.

CG generates the CW pulses and linear frequency modulated pulses through Surface Acoustic Wave (SAW) devices.

SPSA converts to digital samples the I & Q components of the signal and calculates the signal spectrum by an FFT on 128 points. After square modules extraction the results are accumulated over 55.7 msec to reduce signal fluctuations, leading to average Ku and S waveforms. In particular, 100 return echoes from the primary channel (Ku Band) and 25 from the secondary channel (S Band) will be accumulated during the specified time interval, being the Pulse Repetition Interval for the primary channel equal to 557 sec and for the S channel equal to 2228 sec (one S pulse in transmitted every four Ku pulses).

Pulse compression of LFM pulses is accomplished through the well established concept of deramping. The returning signal, actually composed of many chirps each reflected from a different facet of the surface observed, is then mixed with a delayed replica of the transmitted signal.

The Deramping Mixer generates signals which are the difference frequency between its two inputs. As the two inputs have the same rate of change of frequency, the output frequencies are constant tones. Being the input signals linear, mapping between the time offset of each individual chirp respect to the reference chirp into a frequency offset is then being generated.

As a result of the deramping process, targets with different range will give tones at different frequencies that can be resolved by a filter bank of adequate frequency resolution (comparable to the reciprocal of the radar pulse length) which is efficiently implemented through a simple FFT processing after analogue to digital conversion. Deramping allows to reduce the analogue signal bandwidth when receiving LFM pulses from scatters over a small range swath size like the one observed by an altimeter, strongly reducing the speed requirements of A/D converters. Furthermore, the deramping approach allows to strongly reduce instrument performance sensitivity to amplitude and phase distortions of RF subsystems up to the DRM stage. Just the amplitude distortions of filters and amplifiers down the DRM shall be taken into account and properly monitored through specific in-flight internal calibration procedure for their off-line correction on the altimeter echoes samples on ground.

The RA-2 instrument can be operated in different modes which belong to the following major classes:

Support modes

Operation modes

Support modes are used during instrument initialization procedures, and failure recovery procedures.

The Operation modes include the Measurement mode, the RF and Digital BITE (Built In Test Equipment) and the IF Calibration mode.

All the relevant information collected in any of the operation modes are transferred to ground in a standard layout denoted as Source Packet which contains 1.114 seconds of data (corresponding to 2000 Ku PRIs) organised in sets of 20 data blocks; each data block includes data collected over a period of 100 Ku PRIs.

RF and Digital BITE allow testing of the RF chains and of the digital sections respectively. Due to their specific verification oriented purpose, their use in orbit is not planned unless problems are encountered in the operations with the nominal chain.

The IF Calibration mode is instead specifically designed for periodic in-flight absolute calibration of amplitude distortions within the receiver noise bandwidth of the receiving sections after the deramping mixers (common to Ku and S chains). The instrument is operated to collect thermal noise samples and perform spectral analysis by FFT algorithm to retrieve the spectral shape of the noise within the bandwidth of interest. Noise spectra, averaged over sets of 100, are down linked on ground to calculate the IF filter correction mask.

• Measurement Mode

The Measurement mode is the principal operation mode of the altimeter, the main objective being the continuous tracking of the return echo time delay and the return signal power. During tracking additional functions are sequentially performed under control of a scheduler.

• Acquisition

Acquisition measurements will be performed at the beginning of the Measurement mode since no a priori knowledge of the satellite height over the surface is available. Furthermore an Acquisition phase is required whenever a loss of tracking is detected by the on board tracker of the RA-2 in order to properly reinitialize the Tracking.

Acquisition consists of an NPE (Noise Power Estimation), a Detection Phase and an AGC Setting Phase. Each of these phases shall in principle be skipped if requested from ground but default values shall be used in place. Even the whole Acquisition phase shall be skipped and the Tracking phase of the Measurement mode shall start with default preloaded values. This particular situation is known as "Preset Tracking"

The NPE phase is required to measure the instrument thermal noise level. It is accomplished by averaging over the Ku band noise samples collected in five time windows of 426.66 msec duration each and sampled at a frequency of 6.4 MHz. This results in 2731 noise samples for each of the 5 PRIs considered. During this phase the AGC is maintained constant at a default value AGCNPE contained in the Dispatch Area of the SPSA unit.

NPE can be repeated for two times in case of failure (i.e. comparison between the computed value and an allowed expected predefined range fails). After a second attempt a default value for the noise power level, extracted from the Dispatch Area, is used.

Completion of the NPE phase is followed by a FIRST DETECTION phase: 232 Ku band echoes, one every two Ku PRIs, resulting from the transmission of 20 _sec unmodulated radar pulses, are averaged together and compared with the previously computed noise power level to estimate the echo leading edge position within the pulse repetition interval. During this phase the AGC is maintained at a constant reference level AGC Det1 defined in the Dispatch Area.

The computed leading edge position is compared wrt an allowed predefined range (typically the expected orbit variation). If the value is out of range, the detection is repeated from the beginning with a new AGC Det2 value.

If the SECOND DETECTION phase test also fails, the Acquisition phase will restart from the beginning otherwise the AGC SETTING phase is executed.

During this phase the mean power level of the averaged echo is computed to properly set the AGC for the incoming Tracking phase. The mean power level is evaluated from the 100 signal samples around the estimated leading edge position. The computed AGC is then compared wrt an expected allowed range. In case of failure, a default value, defined in the Dispatch Area of the SPSA, shall be used in place.

The duration of the Acquisition phase is mainly driven by the result of the Detection phase. To give an idea the following 5 elementary cases are reported:

• Tracking

The instrument is continuously transmitting in both bands linear frequency modulated pulses. Only received Ku band waveforms are fully processed on-board; S band waveforms are, instead, simply collected and sent to ground in the science data telemetry.

For each Ku band echo 128 In-Phase/Quadrature samples are gathered using 8 bit A/D converters with 6.4 MHz sampling frequency.

Samples are then Hamming weighted and corrected for the fine Rx-delay information:

eq 3.1

 

where:

n = 0..NF-1 (NF = 128)

WH(n) is the Hamming weighting law

X(n) is the n-th I/Q sample

XW(n) is the n-th corrected I/Q sample

Δshift is the fine shift component of the Rx delay information

The Rx delay information computed by the on-board processor to give the position of the tracking window within the Ku PRI, is quantized according to the Tx/Rx clock period, derived from the Ultra Stable Oscillator frequency, i.e. 12.5 nsec; any fine adjustment within 12.5 nsec is accomplished in principle by right rotating the waveform spectrum in the FFT filter bank through the complex exponential term of equation (click here) -1 instead of moving the receiving window. Δshift thus represents the fine adjustment required once expressed in units of FFT filters.

The corrected 128 I/Q samples are then Fast Fourier Transformed:

eq 3.2

K = 0,..., NF - 1

The square modules of each transformed sample is extracted and an amplitude fine correction term is applied as a multiplicative factor:

eq 3.3

 

The fine amplitude correction term accounts for that part of the on-board attenuation, AGC, which cannot be applied through the RF attenuators since these devices are controlled with a resolution of 1 dB.

The computed waveform is finally accumulated over 100 Ku PRIs.

Two additional waveform samples shall also be computed by the on-board processor by performing a DFT algorithm on the corrected I/Q signal samples. A square modules extraction and an accumulation over 100 Ku PRIs is performed even in this case. The two DFT samples are computed in the middle of any two adjacent FFT filters; the indexes for the two selected DFT samples are specified in the Dispatch Area in the form of memory addresses for the selection of the sine/cosine table required in the evaluation of the DFT formula.

Except for the DFT samples computation, the above steps apply even for the S band echo samples processing; accumulation is in this case accomplished over 25 echoes and only the 64 central waveform samples of the average echo will then be passed into the source packet and transferred to ground.

The 128 samples of the Ku average waveform will instead be used to update the tracking window position and the AGC. To properly perform this operation a parallel task called Noise Power Measurement (NPM) is periodically (every 32 seconds) performed and consists in estimating the instrument mean noise power level by collecting noise samples for 10 μsec soon after the transmission of the Tx pulses within the Ku PRI. The computed noise power level, once it is multiplied by a proper scaling factor, is used to hard limit the Ku average waveform samples:

See

eq 3.4

if

eq 3.5

eq 3.6

if

eq 3.7

where Ts is the threshold computed from the noise power level estimate. The width W and the centre of gravity B of the binary vector T(i) are then evaluated:

eq 3.8

 

eq 3.9

 

The echo leading edge position wrt a reference FFT filter shifted of Δoffset filters on the left side of the centre of the FFT filter bank is then computed as:

eq 3.10

 

The leading edge positioning error is then converted in to time units through the chirp bandwidth currently in use:

eq 3.11

 

and also used as a key information by the Resolution Selection and Loss of Lock Detection logic to allow automatic switching among the three resolutions available at Ku band as well as to go back to the Acquisition phase whenever the tracking condition is considered lost.

See further details on the on-board tracker and the Resolution Selection Logic in "The On-board tracker and its autonomous adaptable resolution" technical note (PO-TN-ESA-RA-1316).

Similarly for the AGC correction, the waveform amplitude is estimated first:

eq 3.12

 

The error wrt a reference power value Pref, defined in the Dispatch Area of the SPSA unit, is then evaluated:

eq 3.13

 

Both errors εp and εAGC are fed in input to a-b filters whose purpose is to update estimates of the tracking window position and of the AGC. The general structure of the α- β filter is the following:

eq 3.14

 

eq 3.15

 

eq 3.16

 

The values xp (predicted rate) and xc (predicted value) are updated every 100 Ku PRIs since the error terms are available only after a Ku average waveform has been collected. Since the tracking window position and the AGC values are required every Ku PRI a linear interpolation technique starting from xp and xc will be used to generate values at the required rate. These values will be split in coarse and fine correction terms as already briefly indicated.

During Tracking the instrument is also performing internal calibration measurements without interruption of tracking by coupling the output signal in the receiver through the calibration path of the Front End. One Point Target Response (PTR) measurement is performed every source packet (1.114 seconds) interleaving between the Ku and the S band. In particular, one S PTR measurement is performed every 4 source packets, leaving the other 3 source packets available for the PTR measurement at Ku band. In this case, since more than one chirp bandwidth is managed by the on board processor, only the chirp bandwidth currently in use at the time of the calibration task execution is then effectively calibrated. The In-phase and Quadrature samples of the PTR measurement will be used on ground to retrieve the flight calibration data needed for instrument errors correction.

On request by macrocommand individual echoes can be included in the scientific data stream and sent to ground: single Ku channel return echoes after A/D conversion are sent to ground without performing FFT, square modules extraction and accumulation. The only constraint is that no more than 1.114 seconds (2000 PRIs) of individual echoes shall be acquired and transmitted.

A special operating function within the tracking phase is also available on request by macrocommand. It is the "Preset Loop Output" and consists in opening the two on-board tracking loops (one or both) for a predefined duration.

• RF and Digital BITE

BITE mode is initiated by macrocommand and it is suitable for both hardware and software check. It is divided into RF and Digital BITE for verification of the Tx/Rx chain and digital processing, respectively. RF BITE is executed from the Measurement mode by performing and open loop calibration using the same technique as for open loop calibration during tracking. RF BITE is executed cyclically until a mode change request is received. Three test phases are identified within the RF BITE:

the first phase lasts 2400 Ku PRIs (24 Data Blocks: 1 full source packet- except for first data block, always spare - and 4 data blocks of the successive one);

the second one lasts 600 Ku PRIs (6 Data Blocks);

the third one lasts 1000 Ku PRIs (10 Data Blocks).

In the first phase averaged waveform data are collected 6 times for each of the 3 Ku chirp + S chirp bandwidths using every time a different AGC value.

In the second phase an A/D conversion process is executed over a time window of 30 μseconds leading to the collection of 192 I/Q samples. A square modules extraction and an averaging over 100 Ku PRIs will then take place and samples transferred in the source packet in place of the 128 Ku and 64 S waveform samples of a data block. This type of measurement is executed 6 times using the following resolutions: CW - 320 MHz - 80 MHz - 20 MHz - 160 MHz (S band) and 1 dummy.

In the third phase, 128 I/Q samples are acquired, then a square modules extraction and an averaging process over 100 Ku PRIs takes place.

The total duration is of 8 data blocks. The source packet is completed with one spare data block. In the first five the 320 MHz chirp bandwidth is used, in the second five the 160 MHz chirp bandwidth is used in place. The 128 averaged waveform samples are finally transferred in the data blocks of the source packet in place of the 128 Ku waveform samples during tracking.

Digital BITE is instead based on an open loop tracking technique (Preset Loop Output) as the one already foreseen during the Tracking phase, using preloaded I/Q signal test samples. Digital BITE is executed cyclically with a minimum duration of one source packet, i.e. 20 data blocks are filled in with Digital BITE data.

• IF Calibration Mode

Switching from Measurement mode to the IF Calibration mode is required to monitor changes in the IF filter mask caused by significant variation of the mean operating temperature of the instrument.

In IF Calibration mode the instrument collects thermal noise samples and performs Fast Fourier Transform over sets of 128 noise samples.

Average noise spectra resulting from the averaging of 100 consecutive FFT outputs after square modules extraction are transferred in the data blocks of the source packet in place of the 128 Ku waveform samples. They will be used on ground to estimate the IF filter shape. The IF Calibration mode is commanded from ground via a macrocommand.