On-Board Calibration System for the Range Delay of the BepiColombo KaT

1
MORE Team Meeting
On-Board Calibration System for the Range Delay
of the BepiColombo KaT
G.Boscagli (ESA-ESTEC) M.Mascarello (AAS-I)
27th February 2007
2
Introduction for BepiColombo KaT On-board
Calibration(for Ranging Delay)
3
Approach for On-Board Calibration

• The approach hereafter preliminary analysed is
  based on the idea of implementing this function
  directly inside the KaT unit
• Target ? Calibration as KaT Internal Unit
  Function
• The idea is to include inside the KaT unit both
  the SSPA and the Diplexer function.
• In general more compact solutions improve S/C
  design (mass, interfaces routing, etc) it is
  believed that also calibration performances
  should be improved following this approach.
• NOTE In this way the calibration doesnt take
  into account the wave-guides (from-to-antenna)
  and the antenna itself. They are outside the
  calibration loop. Which is the contribution of
  wave-guides and antenna in the overall end-to-end
  ranging budget error? At present it is understood
  that the main effect to be considered is related
  to the input/output mismatching variation due to
  temperature variation. This might cause
  multi-path effects and error in the end-to-end
  ranging measurements.

4
Approach for On-Board Calibration

• For this reason (Calibration as an internal KaT
  function) the KaT unit must be commanded (by the
  on-board computer) in two different modes
• Nominal Mode RF link via antenna, unit in
  coherent mode (down-link coherent with the uplink
  both for carrier and ranging signal) when RX in
  Tracking Mode.
• Calibration Mode RX and TX in Loop-back
  Configuration, unit running with the internal
  oscillator, RX coherent (in tracking) with the
  loop-back signal from the TX
• The approach hereafter proposed is based on the
  use of PN regenerative ranging, the reasons are
• The use of this ranging scheme simplifies the
  calibration scheme in particular for the
  ambiguity resolution (when compared with other
  approaches as the ESA STD or the NASA Tone
  Ranging).
• Note - The ambiguity must be solved since the
  KaT loop-back delay (TX-RX) is expected of the
  order of microseconds, while the WBRS band should
  be in the range 5-20 MHz.
• This comment might not be valid anymore in case
  the group delay variation (versus environmental
  conditions and including aging) is inside the
  WBRS ambiguity resolution. This is difficult to
  be predicted at this stage.
• The PN regenerative ranging is already
  implemented inside the BepiColombo X/X/Ka Deep
  Space Transponder, so it can be easily re-used
  for the KaT unit simplifying the multi-frequency
  operation as needed by BepiColombo for plasma
  cancellation.

5
Impact of Calibration on BepiColombo KaT
Front-End Architecture
NOTE this section has been written without
considering the current subsystem architecture
6
Impact of Calibration on KaT Front-End
Architecture
Preliminary RF Power Budget (TX Filter neglected)
RF Budget RF Budget
TX 2 W 33.01 dBm
TX coupler 30 dB 3.01 dBm
Attenuator- T1 30 dB -26.99 dBm
Passive Mixer 10 dB -36.99 dBm
Attenuator- T2 60 dB -96.99 dBm
RX coupler 50 dB -146.99 dBm
T2 is indicated as variable attenuator, it might
be commanded for selecting the proper RF power
(at RX input) for calibration minimum value
around -146 dBm (nominal value around -125 dBm,
TBC) .
Loop-Back Path
7
Impact of Calibration on KaT Front-End
Architecture

• According to the proposed approach, the new
  circuits/functions to be developed are
• The functions in the light-blue boxes 2 GHz LO,
  T1, T2, Mixer
• The functions in the light-green boxes TX
  coupler, RX coupler
• NOTE - The functions indicated in the light-red
  boxes represent the Diplexer and they are present
  in any case. In the current subsystem baseline
  (see dedicated slides by AAS-I on this issue),
  the Diplexer is indicated as external to the KaT
  the advantage of having it integrated inside the
  KaT unit is clear from a calibration point of
  view and for a more compact solution.
• According to the architecture as proposed in the
  previous slide, when in calibration mode the
  loop-back signal is routed back from the TX to
  the RX side and to the antenna as well.
• NOTE It might be useful (TBC) to introduce the
  control of the TX Power (2
• Watt SSPA) to minimise the TX power via antenna
  when in calibration mode
• question is the SSPA delay dependant on the
  selected Gain/Output-Power?

8
Impact of Calibration on KaT Frequency Plan

• The KaT frequency plan must be carefully studied
  in order to simplify the generation of the 2 GHz
  LO signal and to avoid internal RFI issues.
• NOTE - For instance considering the Cassini
  KaT frequency plan (next slide) we observe that
  the 1st IF chain is almost at the same frequency
  of the LO signal for calibration.
• The Cassini KaT turn-around ratio (next slide) is
  not included in the current CCSDS/ECSS
  recommendations for TTC applications. The values
  from the current ECSS-E-50-05B (Radio Frequency
  and Modulation, draft issue under public review)
  are

• The Ka/Ka turn-around ratio values are under
  discussion also in the frame of CCSDS, at present
  the draft recommendation (January 2007) is to use
  3599/3344 and 3599/3360

9
Cassini KaT Frequency Plan
10
CURRENT COMMUNICATIONS SUBSYSTEM BASELINE(from
BepiColombo SRR Data Package)This section has
been provided by AAS-I (Marco Mascarello)
Question Are there any difficulties (due to
on-board baseline architecture) for implementing
the above proposed approach for calibration?
11
Communications Subsystem Current baseline
12
Communications Subsystem (Option KaT Amplifier)
13
KaT on board calibration including Triplexer
Including a Triplexer inside the KaT, it would
be possible to calibrate all the paths till the
antenna interface.
14
Triplexer (from current BepiColombo SRR Data
Package)

• The Ka-Band Triplexer is a 4 port device in
  charge of splitting input and output signals. It
  will consist of a new development for BepiColombo
  based on existing technology.
• The splitting will be accomplished by an E-plane
  trifurcation. Each sub-band will be selected by
  an H plane filter. The foreseen useful bandwidth
  of the filters will be the following
• Rx Filter 50 MHz (TBC) within 34 200 to 34 700
  MHz
• Tx1 Filter 200 MHz (TBC) within 31 800 to 32
  300 MHz
• Tx2 Filter 50 MHz within (TBC) 31 800 to 32
  300 MHz
• As for the X-Band diplexer, the mechanical
  concept will be two symmetrical pieces.
  Interfaces will be standard WR28 waveguide
  flanges. The estimated dimensions for the
  assembly are 75 x 45 x 23.1 mm, while the
  estimated maximum mass should be less than 65 g.

15
Concern

• The above solution based on the Triplxer inside
  the KaT unit shows an important drawback
• The Ka-band DST signal is applied to the Antenna
  through the KaT unit.
• This represents a blocking point !
• Other solution must be addressed, for instance
• Keeping the Triplexer external to the KaT unit
  (calibration not anymore an internal KaT
  function).
• Analysing different mixing approach between DST
  and KaT signals at HGA input.

16
BepiColombo X/X/Ka DST PN Regenerative Ranging

• 1999 JPL
• Balanced Weighted-Voting Tausworthe (v2 and 4)

17
Introduction to Pseudo Noise (PN) Ranging
Sequence

• The term Pseudo-Noise (PN) ranging refers in a
  strict sense to the use of a ranging-sequence
  system in which the ranging sequence is a logical
  combination of the so-called range clock-sequence
  and several Pseudo-Noise (PN) sequences.
• The range clock sequence is the alternating 1
  and 1 sequence of period 2 chips.
• A Pseudo-Noise (PN) sequence is a binary ?1
  sequence of period L whose periodic
  autocorrelation function has peak value L and
  all (L1) off-peak values equal to 1.

Range Clock Frequency
18
Example for the introduction of the
Titsworth/Tausworthe generation scheme
Introduction to Pseudo Noise (PN) Ranging Sequence
Component Sequences or Probe Sequences
PN Sequence
As an example, considering the following
component sequences of period 2, 3 and 5,
respectively (the first period of each sequence
is underlined)
Seq. Gen. 1
Seq. Gen. 2
Seq. Gen. 3
Combined by majority logic give the following
period-30 sequence
PN Sequence
19
Example for the introduction of the
Titsworth/Tausworthe generation scheme
Introduction to Pseudo Noise (PN) Ranging Sequence

• Note that the period T of the PN sequence
  obtained with the Tausworthe scheme is given by
  
  with LCM Least Common Multiple

30 in the above example
Importance of having prime length component
sequences

• The correlation of this sequence (considered as
  /-1 sequence) with the component/probe sequences
  gives the following results

• Note that 2 3 5 10 operations of
  correlation are required instead of the 30
  operations needed in the classical approach. In
  fact, only 9 decisions are required because of
  the antipodal result of the sequence of period-2
  (the clock sequence). Only one of the two
  operation of correlation must be performed
  because the other correlation will be the
  negative of the other.

20
More in general we can state that
Introduction to Pseudo Noise (PN) Ranging Sequence

• The ranging sequence is acquired by the receiver
  as the result of correlations between the
  received sequence and certain 1 periodic
  sequences (and their cyclic shifts) whose periods
  are divisors of the ranging?sequence period and
  that we will refer to as probing sequences.
• The probing sequences are related in some manner
  to the ranging sequence, e.g., the ranging
  sequence might be the sequence resulting from
  some sort of voting by the chips of all the
  probing sequences at the same chip time.
• The probing sequences must have the property that
  when all these in-phase decisions are correctly
  made, then these decisions determine the delay
  (modulo the ranging sequence period L) in chips
  of the received ranging sequence relative to the
  corresponding model of the ranging sequence. The
  (one-way) ambiguity (U) due to the period of the
  ranging sequence in meters is
• 
  

ranging clock frequency
chip rate
c speed of the light
21
The 1999 JPL PN Ranging scheme (Tausworthe
scheme)
Titsworth/Tausworthe generation scheme

• The combining logic is based on the following
  rule the ranging-sequence chip is a 1 if and
  only if either C1 has a 1 at that position or
  all five of the sequences C2, C3, C4, C5 and C6
  have a 1 at that position, or both.

In literature this sequence can be indicated
also as JPL 99 or Taus

• C1, C2, C6 are the so called Probing Sequences.

22
The 1999 JPL PN Ranging scheme (Tausworthe scheme)

• It is obvious from this combinational rule that
  the range clock will be strongly correlated with
  the ranging sequence, which facilitates locking
  on to the range clock at the receiver.
• Since the component sequences C2, C3, C4, C5 and
  C6 are all PN sequences with relatively prime
  periods 7, 11, 15, 19 and 23, respectively, the
  period of the 1999 JPL ranging sequence is L
  2x7x11x15x19x23 1,009,470 chips.
• The probing sequences in the 1999 JPL PN
  ranging-scheme are the range clock sequence
  together with the five component PN sequences.
• The total number of correlation operations
  required for the probing sequences, excluding the
  range clock, is thus 7 11 15 19 23 75.
  

23
The 1999 JPL PN Ranging scheme (Tausworthe scheme)
Correlation characteristics and spectrally
relevant properties of the ranging sequence and
probing sequences
Residual carrier Mod index 0.82 rad-pk
Clock Components at fRC
Chip Rate at fChip_Rate 2.5 Mcps
The spectrum shows a powerful clock component at
half the chip rate and below a noisy floor
originating from the combination process with the
other probing sequences. The fact the range clock
is strongly correlated with the ranging sequence
will facilitate locking on to the range clock at
the receiver. The chip is square-wave shaped.
24
Weighted-Voting Tausworthe PN Ranging-Sequence
Scheme

• The Weighted-Voting Tausworthe sequences are
  derived from the 1999 JPL PN Ranging sequence
  with an apparently small modification on the vote
  logic.
• The selection of different value for the clock
  vote (v2 or 4) provides
• flexibility in the choice of the strength of
  the range-clock component in the ranging sequence
• different level for the power allocated to the
  clock and the other ranging spectral components.
  

25
Balanced Weighted-Voting Tausworthe PN
Ranging-Sequence Scheme

• The Balanced Weighted-Voting Tausworthe sequences
  are derived from the Weighted-Voting Tausworthe
  sequences (scheme above) with an apparently small
  modification on the polarity of some probe
  sequences.
• As the 1999 JPL PN Ranging scheme (Tausworthe
  scheme) also the Weighted-Voting Tausworthe PN
  Ranging-Sequence Schemes (both for v2 and 4)
  present a DC component.
• A simple way to reduce the imbalance in the
  ranging sequence (and to produce what we call the
  Balanced Weighted-Voting Tausworthe
  ranging-sequence scheme) is choosing the PN
  probing sequences with the following first
  periods
• C1 1 ?1
• C2 1 1 1 ?1 ?1 1 ?1
• -C3 ?1 ?1 ?1 1 1 1 ?1 1 ?1 ?1 1
• -C4 ?1 ?1 ?1 ?1 1 1 1 ?1 1 1 ?1 ?1 1 ?1
  1
• C5 1 1 1 1 ?1 1 ?1 1 ?1 ?1 ?1 ?1 1 1
  ?1 1 1 ?1 ?1
• -C6 ?1 ?1 ?1 ?1 ?1 1 ?1 1 ?1 ?1 1 1 ?1 ?1
  1 1 ?1 1 ?1 1 1 1 1
• Note - The key to elimination of imbalance is
  the fact the negative of a real
• sequence has the same autocorrelation function
  as the original sequence.

26
BepiColombo X/X/Ka DST Code Phase Acquisition
The current model of BepiColombo X/X/Ka DST is
programmable and can handle the different
schemes JPL99, BT2 and BT4. The Regenerative
Ranging Channel is composed by
From Carrier Quadrature branch

• the Chip Tracking Loop (CTL) for ranging code
  clock component phase and frequency recovery

• the In-phase Integrator output is provided to
  Code Correlators Six Correlators running in
  parallel for probe sequences (C1,. C6) position
  recovery

• the Down-link Code Generator
• (In this case only the JPL99 case is represented)

27
BepiColombo X/X/Ka DST Chip Tracking Loop (CTL)
The mid-phase integrator output is multiplied by
/-1 in order to provide the right correction to
the loop. In a certain way the multiplication by
/-1 replaces the transition detector typical of
a DDTL, considering that the PN sequence
resembles a square-wave.
Filtered Loop Error
Quadrature Carrier Branch Output
CTL NCO Base Frequency
Scaled Carrier Loop Error
28
BepiColombo KaT Calibration based on PN
Regenerative Ranging
29
Impact of Calibration on BepiColombo KaT Baseband
Processing

• We need a separate PN code generator on the TX
  side clocked by the on-board oscillator
• In the current X/X/Ka DST design the TX PN code
  is generated coherently with the received up-link
  PN code (see previous slide).
• The TX PN NCO and the RX PN NCO must be clocked
  with the same oscillator, avoiding any timing
  error between the two signals.
• At the start of the calibration procedure
  (defined by a strobe signal common to RX and TX
  processing functions) the two PN code generators
  (RX and TX) must be identically initialised.
• The loop back ranging signal acquired by the RX
  provides the delay from TX to RX (Loop-Back
  Delay).
• The PN code phase acquisition (using the Probe
  Sequences) is used for ambiguity resolution
• The phase difference between TX and RX ranging
  clock provides the accurate delay measurement

30
Impact of Calibration on BepiColombo KaT Baseband
Processing

• The phase difference between the RX and TX PN
  Ranging Clock can be measured using the
  filtered phase error loop term of the CTL

Kd
KNCO
E
CTL second order loop
31
Impact of Calibration on BepiColombo KaT Baseband
Processing

• Open Loop CTL Transfer Function

• In the X/X/Ka DST the CTL is digitally
  implemented inside the RX Digital Section (Ts is
  the loop sampling time), using the Z transfer
  function we have

CTL second order loop digital representation
aTs
X
Kd

Ts
CTL Detector
X
ßTs2
E
RX NCO N-BIT

B (nominal chip rate)
FCLK1/Ts
32
Impact of Calibration on BepiColombo KaT Baseband
Processing

• After the transient phase, the error term E
  provides the measurement of the delay between the
  TX and RX ranging clock signal. In radiant we can
  write

• While in time we have

• It is evident that for typical loop sampling time
  of the order of 40 MHz and N32 bit NCO the
  phase/time resolution is well below the required
  BepiColombo ranging delay accuracy. The
  calibration resolution in time due to the digital
  loop implementation is

• The Probe Sequence acquisition phase and the CTL
  error term (E) must be transmitted via telemetry
  down-link (using the X/X/Ka DST link). This
  information (after proper post-processing) can be
  used (on-ground) to evaluate the accurate
  Loop-Back delay.

33
Impact of Calibration on BepiColombo KaT Baseband
Processing

• Notice that the KaT Ranging Delay (RX gt TX) and
  the Loop Back Ranging Delay (TX gt RX) might be
  different, this is due to
• The TX/RX different paths (between nominal and
  calibration mode) in the Front-End (Attenuators,
  Mixer, Couplers)
• Different routing of the signal in the baseband
  digital processing (ASIC gates).
• Probably the first contribution could be kept
  small (and negligible also under variations of
  environmental conditions) in terms of overall
  error budget. This to avoid further complications
  in terms of calibration.
• Also the second contribution (inside the DSP)
  might be kept negligible however if not
  negligible, the delta (between the KaT Ranging
  Delay and the Loop Back Ranging Delay) can be
  measured at unit level in the LAB. Notice that
  this contribution is almost independent from the
  temperature since it is related to the clock
  drift (Note - the X/X/Ka DST is embarking an
  OCXO).

34
CONCLUSIONS
35
Conclusions

• In order to improve the calibration performances
  and to minimise the on-board complexity it is
  suggested to integrate inside the KaT unit the
  SSPA, the Diplexer (), the RF Calibration
  Front-End (Attenuators, Couplers, mixer, LO).
• () The possibility to integrate the
  Diplexer/Triplexer is not clear (under
  discussion).l
• The use of PN Ranging (as per X/X/Ka DST)
  simplifies the calibration function in particular
  for ambiguity resolution.
• Minor changes are foreseen for the base-band
  digital signal processing(). The approach is to
  transmit the CTL error term via TLM link for
  post-processing at the G/S.
• () However these have an impact on the current
  X/X/Ka DST FPGA/ASIC