Previously, heart rate was monitored using electrocardiograms (ECG), where electrodes connected to the body measure electrical activity in the cardiac tissue. Whilst professional systems use up to 10 electrodes, single lead ECG measurement is more common in the sports world.
Here, a chest strap with two electrodes is worn on the chest. Whilst ECG waveforms can be seen, most systems only measure heart rate. Since chest belts are not comfortable to wear, the industry is looking for alternatives, like integrating the electrodes directly into a shirt.
The AD8232 is an analogue ECG amplifier developed for low power and body worn applications such as this. An input amplifier with a gain of 100V/V is accompanied by a high pass filter to block the offset voltage generated by the half cell potential of the electrodes on the skin. An output buffer rejects the higher frequency component generated by muscle activity. This low power front end, consuming 170uA, can be used in combination with the ADUCM350 SoC to create a high performance heart rate monitor or single lead ECG system.
A new trend for measuring heart rate is the photoplethysmogram (PPG), which uses optical principles to retrieve cardiac information. PPG has mainly been used in systems to measure oxygen saturation in the blood (SPO2), but heart rate monitors can be detected when PPG technology is integrated in body worn systems, such as a wrist worn device; something which isn't possible with bio potential systems.
In optical systems, light is projected into the surface of the skin and the absorption of light by red blood cells is measured using a photosensor. During a heart beat, blood flow and volume changes, resulting in scattering of the amount of light received. When used on blood rich sites, such as a finger or the earlobe, a red or infrared light source is preferred. However, a wrist worn device needs to use green light in order to pick up information from the veins and capillaries just beneath the skin's surface. The ADPD142 supports both approaches. It features a complete optical front end, with integrated photo sensor and two current sources with LEDs, designed for reflective measurement.
The main challenges for measuring PPG in a wrist worn device are dealing with ambient light and the artifacts generated by motion. Ambient light can have a big influence on the measurement results. Light from fluorescent and energy saving lamps, for example, produce AC components can be seen as measurements. To cancel out these interferers, the analogue front end uses two independent structures to reject them. After this signal conditioning block, a 14bit SAR A/D converter is integrated. The results can be sent to an MCU or processor over the I2C interface for post processing.
In parallel to the optical receiver signal chain is a synchronised transmit path with an independent current source than can drive two LEDs with a current programmable to 250mA. Since the LED currents are pulsed in the range of microseconds, the average power dissipation is reduced, extending battery life.
Conditions such as ambient light, the wearer's skin or or sweat between sensor and skin might impact the sensitivity at the receiving side. However, because the LED excitation can be configured 'on the fly', an auto adaptive system can be created.
There are two versions of the ADPD142: the ADPD142RG has a green and red LED to support optical heart rate monitoring; while the ADPD142RI supports a red and infrared LED for SPO2 applications.
It can be a problem to cancel motion artifacts in sports watches. Motion between the photo receiver and LED and the skin will influence the optical signal the frequency of this sensitivity shift might be seen as a measurement of heart rate, so it is important that motion is compensated for. The tighter the device is attached to the body, the lower the impact; however, it is nearly impossible to cancel this out mechanically.
There are various ways to measure motion. One could be optically, using multiple LED wavelengths. Common signals would detect motion, whilst differential signals would retrieve the heart rate. However, a motion sensor can also be used.
For wearable and battery operated application, the ADXL362 a three axis MEMs motion sensor, is suited to battery powered wearables. It has a 12bit programmable A/D converter and, depending on the sampling rate, scales its power consumption dynamically. At 100Hz, the power consumption for all three axes is 1.8uA, increasing to 3uA at a data output rate of 400Hz.
Since the ADXL362 can also be used for user interfacing – for example, a tap/double tap application – higher sampling rates might be required.
While these sensor solutions are attractive for wearable products, what is missing something to connect the sensors together, run software algorithms and either store, visualise or transmit the results.
The ADUCM350 is a recent introduced sensor fusion SoC. It has a high performance analogue front end (AFE) with an integrated 16MHz Cortex-M3 processor core.
Because it is configurable, the device can be used with nearly any sensor. It has a programmable wave form generator to power analogue sensors with either an AC or DC signal and convert sensor outputs into the digital domain. The real 16bit A/D converter has no missing codes, samples at up to 160ksample/s and has INL and DNL error rates of better than ±1LSB.
The AFE can be operated as a standalone and does not require the Cortex M3 processor. A programmable sequencer controls the complete measurement engine and results are stored directly into memory.
When complex impedance measurements need to be taken in applications such as glucose measurement, body mass index or tissue discrimination, a built in DSP accelerator provides a 2048 point single frequency discrete Fourier transform. The Cortex processor supports various communication ports including, I2S, USB, MIPI and a (static) LCD driver. Furthermore it has onboard Flash-memory, SRAM and EEPROM and supports five power modes to maximise battery life time.
Power always is a critical factor in portable and body worn devices. Despite new battery technologies bring more capacity, battery size remains an issue. One option is to consider energy harvesting, but the first question is always which harvester to use.
Light and heat seem to be the most appropriate solutions for wearable devices as the sensors used in them don't have high output power. To bridge the gap between harvester and battery, Analog Devices has developed the ADP5090. This switch mode power supply can boost input voltages as low as 100mV to 3V. During a cold start, where the batteries are completely discharged, a minimum input voltage of 380mV is required. However, in normal operation, any input signal down to 100mV is used and converted to a higher potential. Protection circuits are included for safe operation and the device works with almost any battery technology.
Jan-Hein Broeders is healthcare business development manager, Europe, with Analog Devices.
