Data synchronization and time-keeping Thorbjrn Jrger Lehrstuhl fr - - PowerPoint PPT Presentation

data synchronization and time keeping
SMART_READER_LITE
LIVE PREVIEW

Data synchronization and time-keeping Thorbjrn Jrger Lehrstuhl fr - - PowerPoint PPT Presentation

Data synchronization and time-keeping Thorbjrn Jrger Lehrstuhl fr Systemtheorie Institut fr Mikrosystemtechnik - IMTEK Albert-Ludwigs-Universitt Freiburg Groundstation - 2 - 7. September 2016 Group Retreat 2016 Botanischer


slide-1
SLIDE 1

Data synchronization and time-keeping

Thorbjörn Jörger Lehrstuhl für Systemtheorie Institut für Mikrosystemtechnik - IMTEK Albert-Ludwigs-Universität Freiburg

slide-2
SLIDE 2
  • 2 -

Groundstation

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-3
SLIDE 3
  • 3 -

Flight Controller

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-4
SLIDE 4
  • 4 -

Clock synchronization

  • Reference pulse which occurs simultaneous in all systems
  • 1PPS signal of GPS is the natural choice for outdoor experiments

no information exchange between systems necessary Absolute and jump-free GPS time reference is also available Small jitter (15-50 ns) with cheap equipment, < 1 ns with better equipment Indoor Environment usually has to low signal strength

ARM FC GS

tN tN tN is a synchronous timing pulse tN

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-5
SLIDE 5
  • 5 -

Time keeping

  • Each measurement gets a timestamp in a unified time frame
  • Need for synchronous clocks

Flight Controller:

  • Self designed timing framework
  • 1PPS pulse for synchronization
  • Intervals between pulses measured with precise quartz oscillator
  • Absolute deviation below 10 µs
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-6
SLIDE 6
  • 6 -

Time keeping

  • Each measurement gets a timestamp in a unified time frame
  • Need for synchronous clocks

Groundstation:

  • Synchronized with UTC via GPS
  • Less than 1 µs deviation from true time
  • Essentially a Stratum 1 time server
  • Automatic synchronization on startup < 2 min
  • GS references ticks and PPS pulses to UTC
  • Keeps track of all quartz oscillators in the system
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-7
SLIDE 7
  • 7 -

Time keeping

GPS devices – Furuno GT-87 / GF-8701:

  • Lock on GPS, GLONASS, Galileo, QZSS, SBAS
  • Precise PPS generation < 15 ns
  • For PPS generation, only one satellite necessary
  • Sensitive enough for indoor operation (-161 dBm)
  • After startup and lock-on, separate devices generate PPS pulses with

less than 100 ns difference

  • Deviation gets smaller over time, as clock disciplination progresses
  • When lock is lost, drifts with 1µs/100s (hold-over)
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-8
SLIDE 8
  • 8 -

Result: Nicely timed measurements

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

380000 381000 382000 383000

  • 2,00E+009
  • 1,50E+009
  • 1,00E+009
  • 5,00E+008

0,00E+000 5,00E+008 1,00E+009

Number of ticks Package number

8 10 12 14 16 18 20 22

PPS counter

slide-9
SLIDE 9
  • 9 -

Synchronous control system

  • Measurements taken at a fixed frequency
  • All clocks in the system need to be synchronous

Synchronization method:

  • Measurements need to be triggered externally
  • Not all sensors support this
  • Difficult to implement and maintain
  • Manual calibration

Change one thing Repeat previous step

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-10
SLIDE 10
  • 10 -

Thor likes his hammer, so let’s look for an easy and “swift”, but effective solution

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-11
SLIDE 11
  • 11 -

Methods of “Forced Synchronization”

Resampling method: Just ignore the problem

  • Simply take the latest available value
  • Easiness:
  • Brutality:
  • Effectiveness: ? (Should work well, if the signal is slowly changing)
  • Introduces discretization noise and potential systematic errors

Resampling method: Ignore the problem a little less

  • Same as above, but increase the sampling rate
  • Easiness:
  • Brutality:
  • Effectiveness: ? (Linearly reduces the time since last measurement)
  • That is what we do/did but lets try a more subtle approach
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-12
SLIDE 12
  • 12 -

Methods of “Forced Synchronization”

Resampling method: Linear approximation

  • Take the last two values and approximate the current value
  • Easiness:
  • Effectiveness: ? (Should work well, if the signal is slowly changing)
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-13
SLIDE 13
  • 13 -

Methods of “Forced Synchronization”

Resampling method: Moving average aka. FIR filter

  • The N last samples are averaged with a windowing function
  • Higher sampling rates are beneficial
  • Necessary for fulfilling Nyquist criterion and to prevent aliasing
  • Reduces Gaussian noise
  • Introduces delay
  • Discontinuity of rectangular windowing function introduces side-

lobes to the signal spectrum

  • Rectangular window not optimal
  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-14
SLIDE 14
  • 14 -

Methods of “Forced Synchronization”

Resampling method: FIR filter with Hamming window

  • Much better suppression of side-lobes
  • Higher computational cost
  • Possible overkill

Discussion

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-15
SLIDE 15
  • 15 -

Delay overview

tAS, tAoA, tLAS, tIMU small, constant tSPI, tUART significant, jitter through interrupts tRS485 biggest contribution

ARM FC GS IMUs Airspeed AoA IMU GPS IMU

SPI I²C I²C SPI UART

GPS

UART

TAS

SPI

RS485 (Cable) RS485 (Cable)

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-16
SLIDE 16
  • 16 -

Sensor fusion

  • It is desirable to have all sensors fused at one point
  • Time keeping and synchronisation much easier
  • Easier to maintain
  • Fast and low jitter connection to controller
  • More time for NMPC calculations in each time step

FC Airspeed AoA IMU GPS

UART SPI I²C I²C

LAS

SPI UART Ethernet (Cable)

BB

SPI

GS

SSH

IMU

SPI

Past:

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger

slide-17
SLIDE 17
  • 17 -

Sensor fusion

  • It is desirable to have all sensors fused at one point
  • Time keeping and synchronisation much easier
  • Easier to maintain
  • Fast and low jitter connection to controller
  • More time for NMPC calculations in each time step

FC Airspeed AoA IMU GPS

UART SPI I²C I²C RS485 (Tether)

GS

SSH

LAS

SPI

IMUs

Present:

  • 7. September 2016

Group Retreat 2016 – Botanischer Garten – Thorbjörn Jörger