1 of 61

FMCW Radar

EE541

Winter 2015

Michael Harriman

Bradley Hutchinson

Errol Leon

Adi Sanghani

2 of 61

Introduction

  • An FMCW Radar based on the MIT LL opencourseware “coffee can” design.
  • Frequency Range 2-2.5 GHz

3 of 61

Block Diagram [1]

4 of 61

Parts List

  • OSC1 - JTOS-2700 V+ VCO
  • PA1 - Gali-84+
  • Splitter - Wilkinson Microstrip
  • Antennas - Pyramidal Horn
  • Mixer - MBA-25L+
  • LNA - ERA-3+
  • Video Amp1 - LM741 and MC33078P
  • Mic: iMAC two channel microphone

5 of 61

Part List (continued)

  • Modulator: Function Generator or Arduino/ADC
  • Power: Power Supplies (5V, 9V, -9V)
  • Data Processing: MATLAB or python (provided by Raytheon)

6 of 61

Voltage Controlled Oscillator

7 of 61

Voltage Controlled Oscillator

JTOS-2700V Datasheet Values

Vtune (V)

Frequency (GHz)

Power Output (dBm)

0.5

1.94474

8.62

1

1.98936

8.11

2

2.0244

7.94

3

2.07081

8.39

4

2.11663

8.48

5

2.16434

8.16

6

2.21164

8.16

7

2.25979

8.53

8

2.32522

8.16

9

2.37459

7.52

10

2.42607

7.92

11

2.48008

7.72

12

2.53617

7.16

13

2.59233

7.52

14

2.64624

8.34

15

2.69575

8.07

16

2.74074

8.17

17

2.78242

8.7

18

2.82083

8.88

  • The VCOs output power was 2-5dBm less (depending on frequency) than specified on the datasheet.

8 of 61

Voltage Controlled Oscillator

  • Spectrum analysis of VCO with a 600MHz span centered at 2.1GHz.

9 of 61

PA: Gali-84+

  • Vcc=9V
  • R = 30 Ω
  • Gain: ~20 dB

10 of 61

PA Gali-84+

11 of 61

LNA: ERA-3+

  • Vcc=9V
  • R = 150 Ω
  • Gain: ~20 dB

12 of 61

ERA-3+ Measured S-Params

13 of 61

Mixer

14 of 61

Mixer

15 of 61

Initial RF Chain Test

  • VCO
  • PA
  • Test Spitter
  • Mixer
  • LNA
  • Cantennas
  • IF Output

16 of 61

Initial RF Chain Test

  • VCO Output:
    • 151MHz bandwidth
    • ~5dBm

17 of 61

Initial RF Chain Test

  • PA Output:
  • 14 dBm

18 of 61

Initial RF Chain Test

  • Test 4 way Splitter

terminated in 2 ports

  • 9dBm (~6dB loss)

19 of 61

Initial RF Chain Test

  • LNA Output
    • -25 dBm

20 of 61

Initial RF Chain Test

  • IF output
    • -30dBm
    • Lower Freq, but

too high

21 of 61

Wilkinson Splitter

  • A specific class of power divider circuit.
  • Can achieve isolation between the output ports while maintaining a matched condition on all ports.
  • Can also be used as a power combiner because it is made up of passive components making it reciprocal.

22 of 61

Wilkinson Splitter

Ideal Scattering Parameters

  • Left picture shows a typical output expected from a Wilkinson Splitter.
  • The S21, S31 are almost -3dB and the S11 is low near the design frequency.
  • Right picture demonstrates a very high isolation between output ports .

23 of 61

Wilkinson Splitter

Milled Splitter

ADS Layout Model

Benefits

  • In-phase with -3dB loss from ports 1-2 and ports 1-3
  • Lower loss than a resistive splitter
  • There is perfect isolation between ports 2 and 3.

24 of 61

Wilkinson Splitter

Splitter Schematic

Scatter Parameters

  • Wilkinson Splitter was designed and optimized in ADS for a duroid substrate.
  • The Scattering Parameters show close to ideal response at the designed frequency of 2.45 GHz.

25 of 61

Wilkinson Splitter

Experimental Results

S11

S12

S13

26 of 61

Wilkinson Splitter

Experimental Results

S21

S22

S23

27 of 61

Wilkinson Splitter

Experimental Results

S31

S32

S33

28 of 61

Arduino PWM Function Generator

  • First attempt at function generator
  • Utilizes PWM through a Sallen Key LPF to produce D-A conversion

29 of 61

Sallen-Key LPF

  • Resistors in schematic replaced with potentiometers for tuning.

30 of 61

PWM Ramp without LPF

Completed Ramp Output. Notice large amounts of jitter and distortion

31 of 61

Arduino DAC Function Generator

  • After determining jitter was too high for application, Arduino was recoded to control an external DAC.

32 of 61

Capture of Arduino Digital Ramp for DAC control

33 of 61

External DAC inadequate speed

Due to the datasheet specified mandatory hold time for data latching, combined with the high resolution requirement, the external DAC proved too slow to produce a 20ms ramp period. A alternative ramp generator using 555 timers will be attempted instead.

34 of 61

Alternate Ramp Generator

  • 555 Timer 50% Duty Cycle 40ms Period Square Wave
  • Two transistors to convert to Triangle Ramp (0-1.5V)
  • Op Amp to shift to 9-10.5V

35 of 61

Alternate Ramp Generator

LT Spice Sim

Offset 1.5V Ramp (9-10.5V for 2.4-2.5 GHz)

36 of 61

Alternate Ramp Generator Construction

Ramp Output

Level Shift

1.7V Ramp Shifted

21.8ms Ramp Period

37 of 61

Alternate Ramp Generator Construction

Driving VCO

VCO Output

2.416 - 2.524 GHz

38 of 61

Alternate Ramp Generator

  • Ramp generator worked sweeping the VCO, but integration the sync pulse to the mic channel didn’t work. Ran out of time to troubleshoot and used the function generator for the final demo.

39 of 61

Pyramidal Horn Antenna

  • 2 Horn Antennas
  • 14 dB Gain
  • WR-430 Waveguide Flange (10.9cm x 5.461cm)
  • Hardline Coax Feed
  • Copper Clad PC Board

40 of 61

Pyramidal Horn Antenna

  • MATLAB

G = 25; % linear

f = 2.4E9; % freq

ae = 0.51; % aperture eff

a = 0.1092; % a WG dim

b = 0.05461; % b WG dim

% Design Equations

l = 3E8/f;

syms As;

Asol = solve(As^4 - a*As^3 + (3*b*G*l^2)/(8*pi*ae)*As-...

(3*G^2*l^4)/(32*pi^2*ae^2) == 0);

% A = 0.45*l*sqrt(G)

A = double(Asol(1))

B = G*l^2/(ae*4*pi*A)

;

R1 = A^2/(3*l);

RH = R1*(A-a)/A;

lH = sqrt(R1^2+(A/2)^2)

R2 = B^2/(2*l);

RE = R2*(B-b)/B;

lE = sqrt(R2^2+(B/2)^2);

l = sqrt((RE)^2+((A-a)/2)^2+((B-b)/2)^2);

la = sqrt((RE)^2+((B-b)/2)^2)

lb = sqrt((RE)^2+((A-a)/2)^2)

s = B^2/(8*l*R2); % should equal 0.25

t = A^2/(8*l*R1); % should equal 0.3750

41 of 61

Pyramidal Horn Antenna Calc Dimensions

A Side- Trapezoid Dimensions:

base 1: 10.9cm

base 2: 28.61cm

Height: 15.65cm

B Side - Trap Dimensions:

base 1: 5.461cm

base 2: 21.3cm

Height: 16.14cm

42 of 61

Pyramidal Horn Antenna Calc Dimensions

  • Waveguide Depth:
    • λg / 2 = 7.62cm
  • Feed Location:
    • λg / 4 = 3.8cm
  • Feed Height:
    • λ0 / 4 = 3.125cm

43 of 61

Pyramidal Horn Antenna 1 Construction

44 of 61

Cantenna S11 Parameters

  • High VSWR (>2)

45 of 61

Pyramidal Horn Antenna 1 Tuning

Initial

Trimmed Feed Probe

  • VSWR (<2)

46 of 61

Pyramidal Horn Antenna 2 Construction

  • Thinner PC Board Used

47 of 61

Pyramidal Horn Antenna 2 Tuning

Initial

Trimmed Feed Probe

  • VSWR (<1.8)

48 of 61

Full System Test in Lab

    • Wilkinson Splitter
    • Horn Antennas
    • IF Amp/Filtering
    • two channel mic input
    • Function Gen Modulator

49 of 61

Full System Test in Lab

  • More Power than before (10dB)

50 of 61

Full System Test in Lab

  • Function Generator Output
  • Sync Pulse used for software.
  • 40ms Period, 100% Duty Cycle, 0.5V Low, 10V High, High Z
    • 2-2.44 GHz (only half bandwidth used for test due to pulse length

51 of 61

Full System Test in Lab

Filtered IF Frequency Domain

52 of 61

Full System Test in Lab

Audacity Recording

Ch1 - Sync

Ch2 - Data

53 of 61

Software-Python Code

  • Pythonized code provided by Raytheon was used
  • Frequencies and minimum pulse length were changed
  • Y axis had to be inverted since it was incorrectly mirrored in the code
  • Trigger signal was inverted to match our ramp
  • Zooming was performed on output images to see desired area

Sample screen for code

54 of 61

Testing in Dexter Lawn

  • Testing was performed in Dexter Lawn of Cal Poly in order to get high range
  • The same setup from before was used; same function generator was used
  • Many different permutations of movement were performed to compare data

55 of 61

Test Subject Walking from Radar

  • This image was generated from a person walking away from the radar
  • The person is visible to the radar up to 9.1m
  • Average velocity was 1.13m/s, which is calculated from the distance from the radar with time

56 of 61

Test Subject Crisscrossing

  • This image was generated from a person crisscrossing going from the radar out and then back
  • The person can be seen by the radar up to 11.1m going away and 11.5m coming back
  • Average velocity is 0.83m/s going away and 1.24m/s coming back

57 of 61

Golf Cart and Bicycle Test

  • The image was generated from a person driving to the radar in a golf cart and a bicyclist passing through
  • The brighter line is the golf cart and the dimmer line is the bicyclist
  • The golf cart averaged 2.96m/s and the bicyclist averaged 0.91m/s (bicycle is at an angle from radar)

58 of 61

Project Demonstration

  • The components were placed on the cart as shown
  • The cart was pointed outside to take readings
  • Function generator used a ramp as before
  • iMic microphone system was connected to laptop as before to collect data

59 of 61

Demonstration Results

  • The image from the demo was obtained from a person coming towards the radar from afar and another person going away from the radar
  • The people intersect at 10.96m from the radar at 3.97s
  • The average velocity of person 1 (going away) is 2.99m/s and the average velocity of person 2 (coming in) is 3.52m/s

60 of 61

References

  1. http://web.stanford.edu/~kimth/www-mit/radar/images/block.jpg
  2. http://www.mathatube.com/geometry-area-of-trapezoids.html

61 of 61

Social Media