Heart & Breath processing device

I had to design a device used for real-time processing of signals of a biolocation system that receives breath and heartbeat parameters using interferometry method and a scheme containing a hybrid tee (magic tee). One of the usages of this biolocation system is to discover people trapped under debris after e.g. an earthquake.

Input Description

A complex signal that a human chest would generate: it contains at least two waveforms (one for the lungs oscillations, one for the heart oscillations) and some amount of the medium noise.

1. Wavelength: 3cm
2. Microwave generator power: ≤ 10µW
3. Range: ≤ 5m
4. Breath rate: 15..50 min^-1
5. Heart rate: 40..150 min^-1

Limitations

The generator power must be kept as low as possible, because we don't want to even further harm a person who's life feels like a mess already.

I couldn't use any type of a standalone computer system due to economic reasons (yes, the diploma contains all that stuff as well), so I decided to go with something smaller and much cheaper, and thought of a microcontroller: you write some code once, you flash the MC, you give it some power, and it works – very convenient. I had to learn the ways of the force microcontrollers, the CodeVisionAVR program and a bit of C language.

Also, there's never any funding, so all the things I'd have to test – I'd have to simulate them somehow.

Design Process

First, I had to design the horn antenna for the tee to assume for the input signals and the generator power limitation.

Second, I had to come up with an appropriate amplifier so that the useful signal is not lost in the noises. I "used" AD8628ARTZ by Analog Devices. No one of course could provide me with a real amplifier, so I just simulated its params in Proteus.

Next, I had to tell the heartbeat from the breath, and to do that I used the spectral analysis and designed two digital filters: a low-pass filter for breath and a band-pass filter for heartbeat. I calculated the coefficients, used Matlab for the prototypes, and it all worked quite well.

Device Structure

1. An amplifier, to raise the amplitude of the input signal
2. A low-pass filter (LPF), to get rid of the useless noise
3. A microcontroller "scheme", to process the signal
4. An ADC to convert a constant signal to bits
5. A breath "circuit" 1. A low-pass filter 1. A limiter 1. A counter
6. A heart "circuit" 1. A band-pass filter 1. A limiter 1. A counter
7. Some LCD, to indicate the processed signals

flowchart LR
  input[Input]
  amp[Amplifier]
  lpf[LPF]
  lcd[LCD]
  vd[Divider]
  out[Output]

  input --> amp --> lpf --> vd --> out
  lpf --> adc
  br --> lcd
  hr --> lcd

  subgraph mc[ATMEGA8&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp&nbsp]
    adc[ADC]

  adc --> blpf
  adc --> hbpf

  subgraph br[Breath channel]
    style br margin-bottom: 100
    blpf[LPF] --> blim[Limiter] --> bc[Counter]
  end

  subgraph hr[Heart channel]
    hbpf[BPF] --> hlim[Limiter] --> hc[Counter]
  end
  end

Loading

Choosing an MC

For the MC I used ATMega8 by ATMEL. It's not like I was choosing from a lot of options, it just had all the things I needed: ADC, counters and a comparator.

Even though I didn't optimize the code, or used the internal stack or EEPROM (I was 23 back then, cut me some slack!), I still think it all turned out very well, as the flashed code has taken 96.5% of the MC storage memory.

Simulation

I used Proteus to:

1. Simulate the input signal, using two different sine-signals summed up in a summator
2. Add some possible medium noise, using a high-frequency signal in another summator
3. Recreate the amplifier scheme, using active components (R, C)
4. Monitor the amplified noisy signal that becomes the input signal of the MC
5. Flash the MC with the .hex file I got from the CodeVisionAVR build
6. Show the MC output on an LCD

flowchart LR
  input[Input]
  noise[Noise]
  amp[Amplifier]
  osc[Oscillograph]
  mc[Flashed<br>ATMEGA8]
  lcd[LCD]

  input --> noise --> amp --> osc
  amp --> mc
  mc --> lcd

Loading

Amplifier Simulation Scheme

LCD Output

Motherboard Design

I also had to design the motherboard to house the device.

What I Feel

Even though I didn't have the result in physical world (which would be very nice, of course), by the time that I finished this project and defended the diploma, I felt very satisfied about how I was able to dive into lots of new fields of knowledge, to make myself "do it", and how everything felt into its right place. I still feel very proud of myself for all that work, and I don't care that the code is very sloppy.