Figure 1: Bottom view of the compact (24.5mmx30.5mm) hSensor Platform. Image courtesy of Maxim Integrated
Figure 1: Bottom view of the compact (24.5mmx30.5mm) hSensor Platform. Image courtesy of Maxim Integrated

When you’re thinking about making your own fitness tracker, smartwatch, or wearable biosensing device, Maxim Integrated has recently launched the hSensor Platform, offered as the MAXREFDES100# reference design, which is a great place to start. Although it is primarily designed as a development kit, out of the box it already offers some exciting functionality, allowing an easy start in exploration of the possibilities of biomedical sensing.

Sense Everything

The complete development platform comes with an ARM Cortex microcontroller (MAX32620) for doing the on-board processing. An accelerometer (LIS2DH) and accelerometer/gyro (LSM6D3) and barometric pressure sensor (BMP280) been added for measurement of motion and a Bluetooth low-energy communication section (EM9301) for wireless connectivity.

So far nothing revolutionary, but wait till you see the score of Maxim’s own health sensor components on board: The board features two temperature sensors (MAX30205) at both sides of the board for precision skin temperature sensing, a MAX30003 analog front end (AFE) for measuring a variety of biopotential signals, including electrocardiography (ECG, heart), electromyography (EMG, muscle), and electroencephalography (EEG, brain). In addition, the platform also supports a variety of optical measurements including pulse oximetry and heart-rate (HR) detection at three wavelengths, 880nm (infrared, IR), 660nm (red), and 537nm (green) using the MAX30101 optical sensor.

Figure 2: Schematic overview of the components present on the hSensor Platform. Image courtesy of Maxim Integrated
Figure 2: Schematic overview of the components present on the hSensor Platform. Image courtesy of Maxim Integrated

The board is very compact (25.4mm×30.5mm) so it’s small enough to be encased as a wrist-worn or chest-worn device. It’s powered by a CR2032 battery which (depending on the used components and communication bandwidth) will last for several hours (based on my experience).

What’s in the box?

The hSensor development kit ships with the following components to be up and running quickly:

  • The MAX32620HSP hSensor Platform (the board)
  • mBed compatible programming device (plugs into the board, combines the USB memory device programmer style (HDK) and Serial communication (CDC)
  • USB lead for directly connecting the hSensor to a host PC (communication)
  • USB lead for connecting the hSensor using the programming device
  • 3 flexible ECG leads which can be soldered to the tabs on the side of the board (electrodes not included)
  • Battery holder for CR2032 battery (battery not included)

If you want to connect the ECG leads, you also need some electrode patches like Covidien H124SG or these.

Figure 3: Overview of the components in the kit
Figure 3: Overview of the components in the kit

Powered by ARM/mbed

If you don’t know mbed yet, this is also a nice motivator to look into that. The common challenge that you have to get communication drivers running and an IDE downloaded and installed on your system to start developing code has been nicely circumvented in the mbed realm by working with a browser-based compiler and programming-by-file exchange.

Figure 4: The mbed online compiler importing the HPC-RPC application
Figure 4: The mbed online compiler importing the HPC-RPC application

The developer.mbed.org website offers an online compiler suitable for many ARM based platforms. The projects can of course be exported for off-line compilation if you have the correct toolchain installed — but the beauty is that as soon as you’re connected to the internet, you don’t need to. Serial or JTAG drivers are not necessary for uploading programs. An mbed device typically connects as memory device (like a USB stick or SD card) and a program file (*.bin) can be simply ‘saved’ to this device.

Figure 5: Download the compiled program
Figure 5: Download the compiled program

The MAX32620HSP is supported in the mbed online environment, and the available test sources work right out of the box. Even on Mac (with which I experienced problems in earlier mbed projects due to all the ‘hidden’ and catalogue files that need to be saved to memory devices) the programming works smoothly.

Figure 6: Import the necessary libraries for the MAX32620
Figure 6: Import the necessary libraries for the MAX32620

The supplied programming dongle can be used to transfer programs directly from your browser to the hSensor:

Figure 7: Programming dongle connected to the hSensor platform
Figure 7: Programming dongle connected to the hSensor platform

Test applications

The board comes pre-loaded with an application using Maxim’s HPC-RPC protocol for communication, as well as Bluetooth LE connectivity. These applications are by no means meant as finished products for OEM product development, but function as nice, open, well documented starting points. The board connects through a USB cable, registering as CDC serial device. A quick poke at the board using Arduino’s serial monitor (why not, the first one at hand, always check if you can do the “hello world” with an on-board LED) and test of the RPC protocol works as expected.

/System/ReadVer

Returns the current software version

/Led/On

Turns on the onboard LED .. etc. Using this protocol all registers of available devices can be checked, programmed, as well as the scheduling of ‘missions’ — consisting of register settings, sampling frequencies and output formats.

The sources for this HSP-RPC protocol are available for the mbed environment. In the mbed environment there is also a clear specification of the interface command as well as the API for the used components.

A PC (windows) GUI application uses this protocol to configure the board, program ‘missions,’ and visualize data. A graphic link between parameters and schematic layout facilitates this. This PC application runs on a system with Windows 7 (no success at WindowsXP or Win10 in a virtual box — but hey, that’s probably just me) and depends on the MicroSoft .NET framework 4.0. The development team at Maxim is planning to release the sources for this GUI application, as well as the sources for the Android app as well in the coming months.

For the PC connection, wireless connectivity has not been explored or designed further because of the difficulties to get one single universal working combination of Bluetooth dongle/hardware, Bluetooth stack, etc. This also means that for viewing the signals using the GUI, the platform has to have an electrical connection to your system (no galvanic separation) so when checking ECG/EEG signals the system HAS to be disconnected from the power mains! Not only for safety reasons, but also the delicate ECG voltages are influenced by the noise and grounding through the laptop’s supply. The graph shown below is generated with hSensor connected to laptop, power supply disconnected. When using Bluetooth for monitoring on a mobile device, the galvanic separation is guaranteed.

Figure 8: (my) ECG measurement using the PC GUI (running on Windows7)
Figure 8: (my) ECG measurement using the PC GUI (running on Windows7)

The Android app works out of the box on a system running Android 4.4 or higher. By default the signals of the two temperature sensors and barometric pressure sensor are provided. For measuring ECG or accelerometer, a ‘mission’ profile needs to be made (using the Windows GUI application) which can be started from the Android app. A ‘mission’ can also be started from the hSensor using the tiny button.

Figure 9: Android app showing my heartrate (ECG), temperature, accelerometer, and barometer
Figure 9: Android app showing my heartrate (ECG), temperature, accelerometer, and barometer

Testing

To get the ECG readings described in the previous section, I fixed the electrodes to my body in a 3-lead configuration. This can be done in several ways, resulting in different readings, see Figure 10. A good source to get started on ECG electrode placement is for example this: ECG Primer.

Figure 10: 3 lead ECG placement options - image from the HRW Project
Figure 10: 3 lead ECG placement options. Image from the HRW Project

The PC GUI offers options for data visualization and recording. In the previous section the plots from ECG measurement are shown. Here is also a reading of the optical heart rate sensor. It works very quickly using a finger tip placed on the sensor. Measuring the blood flow at a different spot (your wrist for example) works less fluently (you have to sit VERY still).

Figure 11: Optical heart rate sensor data
Figure 11: Optical heart rate sensor data

Conclusion

The Maxim hSensor Platform provides however an excellent starting point for exploring the possibilities of Maxim’s range of health sensors. A large effort has been put into getting users up and running quickly, resulting in well documented ‘getting started’ apps, an open architecture, and open development tools.

In order to get the product ‘Maker Ready’ Maxim organized a series of events such as a hackaton (Aggies Invent) in December 2016 for about 60 students of the Texas A&M Health Science Center, Texas A&M Engineering, and Texas A&M Children’s Hospital, resulting in a large number of working proof-of-principle setups. The winning team of this event utilized the barometer on the hSensor to measure the pressure in mattresses to create a scale for people who cannot stand.

Figure 12: Students hard at work during Aggies Invent. Image courtesy of Maxim Integrated
Figure 12: Students hard at work during Aggies Invent. Image courtesy of Maxim Integrated

Compared with the Bitalino board (which comes in roughly the same price range) the Maxim hSensor offers more integrated functionality and much more configurability. The software and tools are open and well documented. Presently, the Bitalino system offers more connectivity options such as examples for MatLab, LabView, etc. but the developers of the hSensor plan to release more examples and additions to their SDK in the coming months. The sensors on Maxim’s hSensor offer more configuration options and precision (also more complexity) making the platform more suited for quantitative measurement than the more qualitative oriented Bitalino. For projects relying on quantitative measurements in more detail and precision, for projects closer to clinical use or clinical trials I would certainly recommend checking out Maxim’s hSensor. It is really astonishing to take a peek at your own biosignals into such detail with a very small effort.

For me personally it is a great candidate to continue the hunt for ‘AWE’ – the feeling of goosebumps (see this project on Make:). Since measuring goosebumps directly is a very complex process, aggregation of other sensory information is probably the only way to go. Heart rate, temperature, motion are all in there somewhere – with the hSensor we’re getting a head start..