How to make a radio receiver that meets the challenges of future autonomous sensor systems, i.e., ultralow-power consumption (50µW) and high sensitivity?

A feasible solution is to use an energy-friendly envelope detector and to reduce the receiver noise by applying a double-sampling technique. When using this approach sensitive receiver modules with power consumption as low as 51µW become reality.

Such an ultralow-power receiver can be used either as a wake-up receiver or as an ultralow-power component in an asymmetric radio link, paving the way for many new applications.

Frequency domain illustration of a double-sampling envelope detector. Only the important harmonics are shown for clarity.

Next-generation wireless autonomous sensor systems have only very limited budget (around 100µW) available to power the radio, digital signal processor and sensor. Ultralow-power radio receivers are therefore highly desirable. A popular solution to achieve low power consumption is to use an envelope detector.

This electronic circuit takes a high-frequency signal as input and provides an output which is the 'envelope' of the original signal. Although this circuit excels in low power consumption, it severely limits the receiver sensitivity since it attenuates low level input signal and adds excessive noise. So the real challenge is in reducing the receiver’s power consumption while still preserving a good sensitivity.

An interesting approach to do this is by reducing the receiver noise. This can be done by first amplifying the signal before the detector, hereby improving its signal-to-noise.

Next, a double-sampling technique is applied to the downconverting envelope detector to suppress the offset and 1/f noise. This not only improves the receiver’s sensitivity, but also flattens the output noise floor of the receiver.

The flat noise floor enables a flexible trade-off between data rate and sensitivity. By reducing the data rate and narrowing down the output bandwidth, the sensitivity of the receiver is improved accordingly. The proposed solution is therefore particularly advantageous for applications that do not require a high data rate.

To technically implement such a solution, we need an RF amplifier, an envelope detector, two baseband amplifiers, as well as a clock and control logic for double-sampling operation.

The input of the receiver is alternated between ground and antenna input, while the output is sampled twice to distinguish signal and noise. The incoming RF signal is modulated by the switch at the RF amplifier input to its double sidebands, which are f_CLK away from the original carrier.

After the envelope detection, the signal is down-converted to around f_CLK. The double-sampling frequency f_CLK is chosen beyond the 1/f corner frequency of the receiver, e.g. 10MHz, so that the downconverted signal is away from the 1/f noise. The baseband signal is amplified and sampled back to DC, while the 1/f noise and offset is up-converted to multiples of f_CLK and filtered out in the final output.

A matching network containing off-chip inductors determines the RF bandwidth and central frequency. With different of-chip inductors, the receiver is able to operate in different frequency bands, e.g. the 868/915MHz or 2.4GHz ISM bands.

Test board of imec and Holst Centre's wake up receiver.

Imec and Holst Centre have recently applied this concept to realize an ultralow-power receiver with good sensitivity. The receiver chip was implemented in a 90nm digital CMOS technology and occupies an area of 0.36mm².

The receiver was evaluated for 915MHz and 2.4GHz. Measurements on Si show a sensitivity of -75dBm (SNR>12dB) for the 915MHz receiver at 100kbps on-off keying (OOK) modulation. When scaling the data rate to 10kbps and filtering the out-of-band noise, the sensitivity is improved by 5dB.

For the 2.4GHz receiver, the sensitivity is -64dBm and -69dBm for 100kbps and 10kbps data rate respectively. This result demonstrates the effectiveness of double-sampling in improving sensitivity.

This concept opens the door to many new applications. An interesting use case is a wake-up module that monitors the communication channel and wakes up a main receiver when it is needed.

Today's battery-operated wireless communication systems – think of WiFi or Bluetooth – consume a lot of power even at times when there are no data to transmit or receive.

A first step to reduce the power budget of these radios is through duty cycling: activating the radio at regular intervals, and deactivating it in between. But this is still a sub-optimal schema; the radio will still be active when there are no data to receive or transmit. Adding a wake-up receiver such as this one allows keeping the main receiver inactive when no data traffic is present.

It can also be used in another way: as an ultralow-power receiver in an asymmetric radio link, where one side is a highly-sensitive transceiver, and the other side an ultralow-power node.

An example use is in smart buildings, where you could tag construction elements with cheap, low-power wireless sensors driven by energy harvesters. Other uses are in long-range radio-frequency identification (RFID), wireless sensor nodes for logistics, automotive, healthcare etc.

In this scenario, the ultralow-power receiver can be further duty-cycled, bringing its power budget down to a level that can be handled with energy harvesters. In the latter scenario, only around 50µW power is available for the receiver. Hence, our ultralow-power radio of only 51µW with small form factor is a major step forward to achieve this goal.

Guido Dolmans, Program Manager Event Driven Radio, imec and Holst Centre and Harmke de Groot, Program Director Ultra Low Power Radio and DSP, imec and Holst Centre