GBPPR Remote Respiration/Heart Beat Monitor Experiments


This experimental project is for a device which can be used to extract the respiration or heart rate signal from an individual by using only a beam of microwave energy in the X-band (10 GHz) region.  The main circuit consists of a commercial Doppler microwave unit followed by a low-noise 40 dB preamplifier and sharp 5 Hz low-pass filter.  All the signals we wish to detect will be in the sub-10 Hz region.

The contraction and expansion of the heart and lungs will impose motion on the target's chest wall.  We can detect these changes as a slight Doppler shift in the reflected microwave beam.  The final received signal can then be further processed to determine the frequency of these cardiopulmonary events.

The respiration (breathing) rate will be the easiest signal to detect, as the chest cavity movement is on the order of several millimeters, compared to only the faint surface vibrations of the heart beat.  Medical literature lists the period of a human's respiratory rate as being between 3 and 30 breaths per minute (16-24 average), with the heart rate varying from 40 to 200 beats per minute (43-94 average).  This corresponds to a possible frequency range of 0.05 - 0.5 Hz for respiration rates, and 0.6 - 3.3 Hz for the heart rate.

This particular experimental setup will be using a Microwave Solutions MDU1100 X-band Doppler integrated motion detector unit.  These little modules require only three connections, a clean source of +5 VDC, the Intermediate Frequency (IF) output with an impedance of around 400 ohms, and ground.  The MDU1100 module has built-in antennas for both transmit and receive.  They operate at around 10.5 GHz and have a RF output power (EIRP) of approximately +13 dBm (20 mW).  A higher powered Gunnplexer module from an old X-band police radar can also be used and will increase the overall detection range.

At 10.5 GHz, the air-skin barrier is fairly reflective to the RF energy.  The wavelength of the transmitted signal will dictate the system's overall spatial resolution.  At 10.5 GHz, the wavelength is 2.8 cm and the penetration depth in human tissue is only around 0.3 mm, though this seems to vary based on the medical literature I've found.  Operating at higher frequencies will provide you with the ability to more easily "zero in" on an individual target, but the RF signal will be severaly attenuated by any barriers between the radiating antenna and the target.  A lower operating frequency will better penetrate most barriers, but you'll loose the resolution and the antenna system will be significantly larger.  All this will need to be taken into consideration if you intend to use a device like this to locate victims trapped under debris or as a Sudden Infant Death Syndrome (SIDS) monitor for newborn babies.

Because the final output signal will contain very low frequencies, which will not be directly "listenable" to a human being, we'll need to digitally record the signal or at least transvert it to something easier to manage.  DATAQ Instruments has sevearl data acquistion boards available at fairly low-cost.  For this project, a DATAQ Instruments DI-194 board will be used.  The DI-194 (which is now obsolete) provides 10-bit measurement accuracy in four analog channels, has a +/- 10V analog measurement range, and up to 240 samples/second throughput.  It uses a simple RS-232 serial interface which also provides power to the unit.  You can use the free WinDAQ software to view and record any signals.

Pictures & Construction Notes

Overview of the low-noise preamplifier and 5 Hz low-pass filter circuit.

The amplifier is based around a Linear Technology LTC1051 "auto zero" precision dual op-amp.  It's configured in a standard non-inverting circuit with 40 dB of gain, rolling off at around 7 Hz.  A 10 kohm / 100 µF RC network forms a high-pass filter at around 0.16 Hz on the op-amp's input.

A Linear Technology LTC1062 5th-order, Butterworth low-pass filter follows the LTC1051.  The LTC1062 can be configured to low-pass filter anything up to 20 kHz.  A 6,800 pF capacitor on the LTC1062's pin-5 sets the filter's 3 dB response at around 5 Hz.

The second LTC1051 op-amp is configured as a two-pole 30 Hz low-pass filter.  This is to remove any high-frequency clock noise which may pass through from the LTC1062.

Both the LTC1051 and LTC1062 require a clean source of +/- 5 VDC.  The power lines to each IC have series 10 ohm resistors, ferrite beads, and a shunt 10 µF capacitor to ground.

Use non-polarized, high-quality, low-leakage capacitors and 1% metal film resistors for maximum performance.

Overview of the Microwave Solutions MDU1100 X-band Doppler integrated motion detector unit.

The four little squares are the transmit and receive antennas.

The connections along the bottom are, from left-to-right: IF output, ground, and +5 VDC.  The current draw is around 50 mA.

Detection range can be significantly increased by replacing this module with a Gunnplexer from a high-power X-band Doppler speed radar.

Setting the 3 dB corner frequency for the LTC1062 low-pass filter.

The LTC1062 contains an integrated divide-by-100 counter to determine the cut-off frequency.  This means the "Cosc" frequency at pin-5 of the LTC1062 should be 100 times the required 3 dB corner frequency.

In the above photo, the frequency at pin-5 of the LTC1062 is 594.5 Hz with a 6,800 pF polystyrene capacitor installed.  This gives a final 3 dB corner frequency of around 5.9 Hz.

Be sure to use a low-leakage and temperature stable capacitor in this timing application.

Overview of the DATAQ Instruments DI-194 data acquistion board.

Using the WinDAQ software.

This is an example of detected chest wall motion caused by the act of breathing.

At frequencies this low, it is best to detect the number of "events" in order to determine if you've received a true signal.

If you detect, say a peak of around 18 and 80 "events" per minute, it's safe to assume you've located a live human being.  Those 18 and 80 events in a one minute time frame correspond to the average respiration and heart rate for a healthly adult human.

The heart beat will appear as a series of sharp "spikes" riding on the slow rising respiration signal.  Separating those two signals using only common hardware and components is still something I'm working on.  This may be a better job for some DSP or software wiz...

There is a schematic for an optional (and untested) signal splitter at the end of this article.  This is a low-pass/high-pass filter centered at 0.5 Hz which can be used to break the two signals apart for further processing.

Audio / Video


Datasheets & Notes

Return to Homebrew Military & Espionage Electronics Page