Harmony PI Filters are in stock. Order direct or through DigiKey.
BiomedicalWearablesEnergy harvesting

Wearables that match a chest strap: a TransSiP PI case study from Kanematsu

Kanematsu Futuretech Solutions, our distribution partner in Japan, published a user case on TransSiP PI in wearables: optical heart-rate monitors that reach clinical accuracy with one LED, and a sports watch that runs on body heat.

Kanematsu Futuretech Solutions, TransSiP’s distribution partner in Japan, has published a user case on what TransSiP PI does inside wearables. It is worth retelling here, because it shows something every wearable team runs into: the accuracy of the sensor is often decided by the quality of its power supply, not by the sensor itself.

The full case study is on Kanematsu’s website. This is our summary.

Efficiency is more than conversion efficiency

Wearables live on a small battery, and the newest ones on harvested energy. Teams work hard on DC-DC conversion efficiency, and rightly so. But the system’s efficiency also depends on how clean the supply is once it is converted. A noisy supply lowers signal quality, sensitivity and accuracy. The processor then spends more time on error correction and retries, which costs battery life and, in the end, user experience.

The two usual ways to convert power each fall short:

  • Switching regulators are efficient but noisy. Their switching noise also moves in time, a component we call Switching Noise Jitter (SNJ), which dominates when signals are weak.
  • Linear regulators are cleaner but less efficient, and the loss grows as on-chip voltages drop.
  • Ground bounce from I/O activity, and transient noise from power-saving duty cycles, add noise that conventional filters do not remove.

TransSiP found SNJ while investigating why GPS receivers became unstable. It does not show up in ordinary frequency-domain analysis, so we used Tektronix real-time spectral histograms to see it. That discovery led to the TransSiP PI products: the Symphony PI DC-DC chipset for clean buck conversion, and the Harmony PI Filter, which works with almost any DC source and removes noise in both the frequency and time domains.

Example 1: optical heart-rate monitors with one LED

Wrist-worn optical heart-rate monitors are generally less accurate than a basic chest strap, even with several LEDs, several photodetectors and advanced algorithms. Arterial pulses change the optical signal by about 0.086 dB or less, so supply noise on the LEDs and photodetectors is enough to bury them. The usual fix is more LEDs and more current, which costs power and still falls short in real use.

TransSiP PI goes after the other side of the problem: the supply. With a Harmony PI Filter cleaning the bias of the LED, the analog front end, the photodetector and the current ADC, a TransSiP PI-enhanced monitor matched a chest strap with a single LED and a single photodetector. The case study lists the thresholds for an optical monitor to count as clinically acceptable, and the PI-enhanced monitor met all of them:

Clinically acceptable when TransSiP PI-enhanced monitor
Within ±10 bpm of the chest strap 98% of the time, sedentary Met
Within ±10 bpm 95% of the time during the most intense activity (running) Met
Within ±10 bpm 92% of the time during the least periodic activity (gym) Met
Limits of agreement with the chest strap within ±15 bpm Met
Mean absolute error of 3 bpm or less Met
Standard error of the estimate of 10 bpm or less Met

That makes a PI-enhanced optical sensor a valid alternative to a chest strap, at lower power. The same approach applies to other optical biomedical monitoring. In Japan, MEDIROM’s MOTHER bracelet, which runs on body heat, uses four TransSiP PI products. See our biomedical case study for the measured results.

Example 2: a sports watch that runs on body heat

TransSiP PI is also a key enabler for fully autonomous, energy-harvesting products, where high conversion efficiency and a very clean supply both matter. In the case study, a TransSiP PI-enhanced sports watch measured distance traveled 10 times more accurately, and ran its power-hungry processes 25% faster, than without TransSiP PI.

Matrix Industries used both gains in its Power Watch 2, which runs from the wearer’s body heat, with optional solar harvesting. It offers GPS, heart-rate monitoring and calorie counting with no charging. As Douglas Tham, CTO of Matrix Industries, put it:

“In energy harvesting systems, efficient use of power is the key to real autonomy. TransSiP PI extends our edge: it gives us the power integrity we need to achieve high performance and ensure that our technologies are fully autonomous.”

Where else it applies

The case study lists the markets where TransSiP PI is in use: location-based services, wireless communications, homeland security, ultra-wideband, M2M (CAT-X, NB-IoT and 5G), data storage, computing and biomedical monitoring.

To try it on your own design, start with the Harmony PI Filter and its evaluation board. In Japan, contact Kanematsu Futuretech Solutions or TransSiP.

Sources

  1. Kanematsu Futuretech Solutions, case study: TransSiP PI, the secret of getting the most out of your system
  2. TransSiP case study: a body-heat-powered health tracker in Japan