An Engineer's Guide to Choosing Different Heart Rate Detection Techniques
September 09, 2019
Today's wireless and wearable technology not only enables complex data to be non-invasively collected but also allows it to be analyzed and displayed instantaneously.
Heart rate data can provide valuable information on an individual's physical condition. Its usage ranges from personal fitness monitoring, through detecting irregularities such as arrhythmias, to hospital-grade patient monitoring of vital signs. Today's wireless and wearable technology not only enables complex data to be non-invasively collected but also allows it to be analyzed and displayed instantaneously, as well as stored for analysis later.
Such technology has opened up new fitness and wellness consumer markets for wearable devices, such as fitness bands, and chest strap sports monitors. In the medical sector, non-invasive heart rate monitoring is used to identify problems such as reduced blood flow to the heart, and to assess risks such as heart attack and thrombosis. Even in a hospital ward, nurses can now perform regular ‘observations,' such as arterial oxygen saturation, respiration rates, and hydration levels, using a simple finger-grip heart rate monitor and portable analyzer.
Demand is rising for non-invasive and/or wearable devices, while the complexity of data to collect is also increasing, adding significant technological challenges in terms of data acquisition, signal conditioning, and processing. For medical applications especially, measurements have to be reliable, accurate and secure.
There are two primary methods for measuring heart rate. The first utilizes optical techniques to detect changes in light absorption or reflection as blood passes through vessels close to the skin. Optical techniques can also be used to estimate oxygen saturation in the blood (pulse oximetry). The primary technical challenges are low power consumption, ambient light rejection and ambient noise cancellation.
The second method, biopotential measurement, uses voltage sensing electrodes to detect the electrical activity generated by heart muscle tissue, which transmits to the skin. The data are used to generate an Electrocardiogram (ECG), widely used by medical experts to determine cardiac health. Bioimpedance measurements can also determine respiration rate and relative amplitude. The key technical challenges associated with this method are low power operation, motion compensation and countering other interference, such as noise.
Optical Pulse Oximeter
Fortunately for developers, many dedicated, application-specific optical data acquisition systems for heart rate monitors are available. The MAX86140 from Maxim Integrated, for example, is optimized for detecting optical heart rate, oxygen saturation, and muscle oxygen saturation in monitors in contact with the wrist, finger, ear and other locations.
While optical heart rate monitoring typically requires a single light source, a pulse oximeter needs two. For fabulous accuracy and to increase the range of measurements possible, multiple light sources are often employed. The MAX86140 and MAX86141 offer single and dual optical channels respectively.
On the transmitter side, three programmable high current LED drivers can be configured to drive up to 6 LEDs. With two devices working in master-slave mode, the LED drivers can drive up to 12 LEDs. A critical feature of these devices is their robust, proprietary Ambient Light Cancellation (ALC) circuit, particularly useful for ensuring accuracy under bright conditions. Also, the system can cope with rapid changes in light levels.
Other key features include low-noise signal-conditioning Analog Front End (AFE) with 19-bit sigma-delta ADC, voltage reference and temperature sensor. The ADC output data rate can be programed from 8 to 8192 samples. The devices require minimal external hardware. A 128-Word FIFO provides on-chip storage for digital output data and enables connection to a microcontroller.
Operating on a 1.8V main supply voltage with separate 3.1V to 5V LED driver power supply, both devices provide many power saving facilities. They have flexible timing and shut down configurations as well as control of individual blocks. This enables optimized measurements at minimum power levels. A dynamic power down mode is available for lower sample rates, below 128 sps. A proximity mode function can reduce energy consumption when the sensor is not in contact with the skin.
The optical controller can be configured for a variety of measurements. One, two or three LED drivers can be pulsed sequentially, making measurements at multiple wavelengths, as required for pulse oximetry. When the LED drivers are pulsed simultaneously, heart rate measurements can be taken on a wrist-mounted unit. The LED drive level can be adjusted to compensate for increased noise levels, such as when high interfering ambient signals are present.
Biopotential ECG Measurement
An ECG measures heart rate and can provide details on individual signals, giving professionals additional detail for cardiac investigations. It also allows for more reliable heart rate monitoring in fitness applications, particularly when a chest strap is used. Biopotential measurements typically require significantly less power compared to optical sensors for the same level of accuracy. However, the processing of ECG signals can consume battery power rapidly. In addition, ECG readings are highly sensitive to motion and other interference sources. Thus, in fitness applications, motion compensation is particularly important, and movement can also be a significant source of noise.
Once again, dedicated application specific devices are available for this type of application. Consider the MAX30003 from Maxim Integrated, described as a complete Biopotential Analog Front End (AFE) solution for wearable applications. See Figure 2. This single channel device features a clinical grade ECG AFE with high-resolution ADC delivering 15.5-bits effective resolution with 5μV peak-to-peak noise. In addition, it has ESD protection, EMI filtering, internal lead biasing, DC leads-off detection and soft power-up sequencing. High input impedance ensures minimum signal attenuation at the input during dry start.
Motion compensation and removing interference from motion artifacts is achieved by ensuring the Common Mode Rejection Ratio (CMRR) of the AFE is as high as possible. The MAX30003 has a CMRR characteristic typically up to 115dB. Selectable lead bias resistors help improve CMRR as well as contributing to high input impedance. Various low-pass and high-pass filter options are available to limit bandwidth, important for attenuating noise from static electricity and high-frequency signals. For fitness applications, the single power high-pass corner frequency should be set at 5Hz, while for clinical use, it can typically be down to 0.5Hz or 0.05Hz. For sports use, the common mode low-pass corner can be set at 34Hz, the optimum level for limiting clothing noise during dry starts.
The ultra-low power consumption of 85 µW at 1.1V supply voltage gives a longer battery life. The leads-on detection feature operates during standby/deep sleep mode (70nA). A 32-word FIFO can contain up to 32 results of ECG data conversions, saving on host microcontroller activity, as it stays asleep longer, thereby reducing power consumption. Similarly, the MCU can be programmed not to process potentially invalid data. A built-in algorithm for R-R peak intervals gives further power savings, such that the MCU only consumes about 1µA, compared to 50 to 100 times more than when hosted on the MCU.
Maxim offers a design platform for developing a wearable health or fitness product based on this device. The MAXREFDES100 Health Sensor Platform includes the hardware building blocks on a single PCB and an ARM mbed-based programming board as a Hardware Development Kit (HDK). In addition to the MAX30003, the MAX30101 adds pulse oximetry functions through optical elements including LEDs and photodetectors, plus low noise electronics with ambient light rejection. Also available is the MAX30205 clinical grade temperature sensor, while the recommended power block is the MAX14750, providing multiple outputs for the MCU, AFE, and the digital interface.
For more on wearables, check out: