IEEE Solid-States Circuits Magazine - Fall 2019 - 18

WuRX sensitivity needs to be as good as the
main receiver; otherwise, the effective range of
the system is limited.
where CI is the connection interval
and SCA is the sleep-clock accuracy.
The accuracy of the sleep clock
used determines the amount of guard
time needed to ensure that the data
packet is received during the target wake time. A less-accurate sleep
clock requires a longer guard time
and causes higher average power consumption because the radio is active
for a longer period. As an example, if
the most relaxed BLE sleep-clock accuracy setting of 500 ppm is used and the
connection interval is 1 s, the guard
time must be 500 μs to ensure that
the receiver is awake at the appropriate time and does not miss the data
packet. Given that the data packets
can be short, it is possible to spend
more energy listening and waiting
for the data to arrive than actually
receiving and processing those data.
To avoid such inefficiency, a sleep
clock is desired that has sufficient

accuracy to use the tightest BLE sleepclock accuracy setting of 20 ppm.
However, there is often a power/
accuracy tradeoff in implementing
the sleep clock. It is on continuously,
so it adds directly to system power
consumption. The sleep clock needs
to be both very low power and very
accurate for the lowest system power.
This is illustrated in Figure 2, where
typical BLE node-current consumption
is calculated for a short data packet
sent with a 1-s connection interval
using a sleep clock that has either
500- or 20-ppm accuracy. A tighter
accuracy allows a shorter guard time,
which reduces the average current
consumption, particularly for use
cases such as this example, where
the data packets are very short. Several methods to generate the sleep
clock are possible, including low-frequency crystal oscillators, microelectromechanical-systems oscillators,

1-s Connection Interval, 500-ppm SCA,
9-µA Average Current Consumption
Postprocessing

temperature-compensated crystal
oscillators, and integrated oscillators
(typically resistance-capacitance-based).
Each option offers different tradeoffs
among power, form factor, cost, and
frequency accuracy [3].
Figure 3 provides an example
wireless node's current consumption
using synchronized data transmission as a function of the sleep-clock
accuracy and connection interval. If
the node wakes up frequently, the
average power consumption will be
limited by the radio power consumption. If a sleep clock with poor accuracy (1,000 ppm in this example) is
used, the average power consumption can't be decreased even if the
connection interval is increased. If a
very accurate sleep clock is used, the
power consumption at long connection intervals is limited by the sleep
current. From this plot, it can be seen
that synchronized data transmission is most relevant to applications
where the delay is not critical, such
as environmental monitoring. In those
cases, a longer latency can be tolerated to achieve the longest battery
life possible.

1-s Connection Interval, 20-ppm SCA,
6-µA Average Current Consumption

Sleep

Postprocessing

Sleep

Preprocessing

Rx/Tx

Preprocessing

Radio Set Up,
PLL Lock
Rx/Tx

Radio Set Up,
PLL Lock
Guard Time

Guard
Time

FIGURE 2: An example of the current consumption for a BLE wireless node with two different sleep-clock accuracy settings. A tighter accuracy
enables a shorter guard time and lower current consumption. SCA: sleep-clock accuracy.

FA L L 2 0 19

IEEE SOLID-STATE CIRCUITS MAGAZINE

IEEE Solid-States Circuits Magazine - Fall 2019

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Fall 2019