ECE 445 – Senior Design Project

Document Sample
ECE 445 – Senior Design Project Powered By Docstoc
					   ECE 445 – Senior
     Design Project
Sensor Board for Intruder Detection
                              Yong Jae Lee
                       Andrie Tanusetiawan
                                   Xi Zeng
Project: Create a sensor board compatible with
  Prof. Jones’s microprocessor board. The goal is
  intruder detection.

Requirements:
- PCB must be within 3” × 5” × 1”
- < 20 mW power consumption
- Powered from a single voltage source @ 3.3 V
The circuit contain two sensors:
         A microphone
              And
   A 2-axes accelerometer
Microphone Circuit
Four components:
- microphone

- high-pass filter

- gain amplifier

- low-pass filter
Microphone
EM6027-42: a condensor mic

Needs to be properly biased:
-Leads are not interchangeable
-Correct resistor value improves output
   signal strength
Microphone
Biasing Test:
-Connect speaker to function generator

-Fix volume

-Fix distance between speaker and mic

-Vary potentiometer, and measure output
Microphone
Microphone
Conclusion:
-Use R = 12 kΩ

-Tolerance of 5% is adequate
High-pass Filter
Two purposes: AC
  coupling, kills low-
  frequency thermal
  noise (~30-40 Hz) from
  microphone

Active filer: 2nd order
  Butterworth Sallen-Key:
Av =
High-pass Filter (cont.)
To implement Butterworth:
Set C1 = C2 ≡ C, R1 = 2R2, R ≡
  fc =

Our Design:
C1 = C2 = 0.01 μF ± 3% tolerance
R1 = 100 kΩ (± 3%) + 12 kΩ (± 3%) = 112 kΩ
R2 = 56 kΩ (± 3%)
→ R ≡ 79.2 kΩ, C ≡ 0.01 μF → fc = 201 Hz
High-pass Filter (cont.)
Our circuit is powered from just one voltage
 source (+3.3 V) and microphone outputs
 contain negative components. Thus, using
 Sallen-Key as given would kill all negative
 components. It would be even worse if the
 op-amp utilized is not rail-to-rail (as our
 case).
High-pass Filter (cont.)
Solution: introduce a DC-offset
High-pass Filter (cont.)
          Some Data of the High-pass filter stage:

   C1 = 10.2nF; C2 = 10.2nF; R1 = 110.1kΩ; R2 = 55.3kΩ
High-pass Filter (cont.)
   Tolerance: At 3% tolerance, fc shifts by less than 15Hz
                   (according to theory)

             Actual Measurement Confirmation
Gain Stage
Amplifying our signal:

Av = -100




The capacitor is for AC coupling.
The resistors bridge is for DC offset.
Gain Stage
Why separate gain stage from high-pass filter?


                                        From: Texas Instruments,
                                        Op Amps for Everyone




Answer: Stability
Low-pass Filtering
   Anti-aliasing
   Microprocessor may have built in anti-aliasing
    filters for the A/D, but included just in case
   Professor Jones suggests treating sampling
    frequency = 8kHz  Nyquist Frequency is at
    4kHz
2nd order Butterworth Low-pass filter




   Butterworth Requirements: R1 = R2 ≡ R, C2 = 2C1, C ≡ √C1*C2,
    fc = 1/(2pi*RC)
   Original Design: C1 = 0.01 µF, C2 = 0.02 µF, R1 = R2 = 10kΩ 
    C ≡ 0.014142 µF, R ≡ 10kΩ, fc = 1125Hz
   Problem: OP490 has very low-bandwidth (price to pay for low
    power)  distortion above 2000Hz
   Furthermore, distortions result in signals with peak-to-peak much
    larger than theoretical calculations
2nd order Butterworth Low-pass filter
Simple, passive RC low-pass filter




   R = 8.2kΩ ± 5% tolerance, C = 0.01uF ± 5% tolerance

   fc = 1/(2pi*RC) = 1940 Hz

   Attenuation is a lot worse, but no distortion
RC Low-Pass Filter
Conclusion - Microphone




   OP490: 4 op-amps
   Swings to 0.6V below Vdd and 0.5V above ground
   Circuit uses only 3 op-amps  power wasted on 1 op-amp
Anechoic Chamber
   Background/thermal noise: 27.5mV peak-to-peak
Microphone Fields
 From Fundamental of Acoustics, 4th ed., by
   Lawrence Edward Kinsler:




When rm = r1 (m = 1), this indicates the first local
 maximum that shows the border of near field and
 far field. Let f = 100 Hz.
Microphone Fields (cont.)


                  r1 < 0, which means that
                     there is no physical
                     meaning since the ratio
                     a/λ is small enough.
                     This also means that
                     there is no near field.
Anechoic Chamber Testing
Anechoic Chamber Testing
Omni-directionality
   Microphone is omni-directional
Directivity
   1/r response for far-field
Microphone Fields (cont.)
Conclusion:
 Only far field is accountable for our
 microphone. The reason is that the aperture
 of the microphone is so small that the near
 field radius is too small to be accounted for.
Accelerometer
   ADXL 213
   Can measure both dynamic (vibrations) and static
    (gravity) accelerations
   Measurement of static acceleration allows it to be tilt
    sensor
   Outputs: analog signals and digital signals whose duty
    cycles are proportional to acceleration
   Advantage of digital signals: duty cycle outputs can be
    directly measured by a microcontroller without the need
    of A/D conversion
Accelerometer Testing

   Hooke’s Law: F = kx
   There is a net force up
    against gravity, the
    accelerometer
    experiences effective
    acceleration = (F + mg) /
    m = kx’ / m
Accelerometer Testing
(continued)
   Theoretical acceleration from datasheet of ADXL213:
    A(g) = (T1/T2 – 0.5) / (0.25)
   Sensitivity is calculated using the above equation with
    A(g) = 1, with T1 and T2 obtained from oscilloscope
    when the board is hanging on the spring
   It is the nominal value (units %/g) that accounts for the
    static acceleration of 1 g felt by the accelerometer
Linearity of Spring


   In linear region (F=kx): period of
    oscillations should be constant
   Stopwatch measure  1.29 s
   Due to physical constraints: only
    measured up to 18cm (distance
    pulled)
   Spring deformed at about 80 cm
Components of Accelerometer

   Cdc = 0.1µF – decouples
    accelerometer from the noise
    on power supply
   Rset = 1MΩ - used to set the
    period of the duty cycles
   Cx = Cy = 0.1µF – low-pass
    filtering for anti-aliasing and
    noise reduction
Components of Accelerometer
(continued)
   Microcontroller has 8kHz sampling frequency: Nyquist rate =
    4kHz; with Rset = 1M Ω , T = 1/125s, f = 125Hz

   Filtering for noise reduction and anti-aliasing can be achieved
    with capacitors Cx and Cy

   Equation for -3db bandwidth
    F-3db = 1 / (2pi * 32kΩ * C(x,y)) = 5µF / C(x,y) (from ADXL213
    Datasheet)

   For Cx = Cy = 0.1µF, F-3db = 50Hz (Bandwidth)
Error measured
Tolerance Values

   PCB1: 99.8nF, 989k Ω
   PCB2: 100nF, 980.5k Ω
   PCB3: 100.1nF, 980k Ω
   PCB4: 100.1nF, 981.5k Ω
   PCB5: 99.9nF, 989k Ω

   Tolerance for Cx, Cy: < 1%
   Tolerance for Rset: < 2%
Over all Circuit
   Vdd = 3.3V, I = 0.772mA
   P = 2.55mW
Transient/Turn-On Time:
 Microphone: ~195ms
Transient/Turn-On Time:
 Accelerometer: ~220ms, both axes


      X-axis                        Y-axis
Stress Testing
   Real Life scenario: sensor board placed near
    a furnace vent
   Simulate using blow dryer
   Each stress session: heat to ~50° C for 5
    minutes
   Measure before stress session, measure
    immediately after, measure after 10 minutes
   Repeat
Stress Testing
 Stress Test, round 1:
 -Test in Senior Design Lab
 -microphone, pre-stress:
 1) 62.5mV peak-to-peak background noise
 2) connect a speaker to function generator at 250 Hz, set to some volume → 641
 mV peak-to-peak readout
 -immediately post-stress:
 1) 67.2 mV peak-to-peak background noise → no significant change
 2) With same sound source at same volume: 843.8 mV peak-to-peak readout →
 significant change
 -after 10 minutes:
 1) Same readout with sound source
Stress Testing
 Stress Test, round 2:
 -same test environment
 -use same board that was stressed
 -microphone, pre-stress:
 1) 62.5mV peak-to-peak background noise
 2) connect a speaker to function generator at 250 Hz, set to some volume →
     fluctuation between 1.063V to 1.375V
 -immediately post-stress:
 1) 65.6 mV peak-to-peak background noise
 2) With same sound source at same volume: 1.016V to 1.391V
 -start round 3 immediately, here are immediate post-stress results
 1) 62.5mV ambient
 2) 1.031V to 1.422V with same sound source
Stress Testing
 Accelerometer Performance, long after stress test 3
Stress Testing
 Circuit Turn-On Time isn’t changed by heat stress:


        Stressed Board          Time

        Microphone:             198ms

        Accelerometer:

                         axis x: 224ms

                         axis y: 223ms
Stress Testing
 Conclusion
 1) Both accelerometer and microphone are altered permanently by heat
     stress
 2) Microphone:
    - microphone generated thermal noise not changed
    - more sensitive to 250Hz; other frequencies?
 3) Accelerometer:
    - accelerometer output (acceleration) changed by around 0.11 m/s2
 4) Power Consumption post stress:
    Vdd = 3.3V, I = 0.788mA → 2.6mW
    Power consumption not changed significantly
Cost Per Circuit Board
Part                        Unit Cost ($)           Quantity       Subtotal ($)
microphone: EM6027-42                       2.271              1                    2.271
OP490 (PDIP)                                 3.74              1                     3.74
12kΩ                                          0.2              2                      0.4
100kΩ                                         0.2              3                      0.6
56kΩ                                          0.2              1                      0.2
2kΩ                                           0.2              1                      0.2
200kΩ                                         0.2              1                      0.2
8.2kΩ                                         0.2              1                      0.2
0.01uF (non-polarized)                        0.2              3                      0.6
1uF (electrolytic)                            0.2              1                      0.2
accelerometer: ADXL 213AE                   14.55              1                    14.55
1MΩ                                           0.2              1                      0.2
0.1uF                                         0.2              3                      0.6
PCB Board                                     2.7              1                      2.7
Unit Total                                                                        $26.661
Credits
   Prof. Michael Oelze
   Prof. Jennifer Bernhard
   Prof. Scott Carney
   Dwayne Hagerman
   Prof. Douglas Jones
   Sriram Narayanan

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:2/5/2013
language:Unknown
pages:49