Datasheet

Data Sheet ADPD1080/ADPD1081
Rev. B | Page 39 of 74
OPTIMIZING SNR PER WATT
The ADPD1080/ADPD1081 offer a variety of adjustable
parameters to achieve the best signal. One of the key goals of
system performance is to obtain the best system SNR for the
lowest total power. This goal is often referred to as optimizing
SNR per Watt. Even in systems where only the SNR matters
and power is a secondary concern, there may be a lower
power or a high power means of achieving the same SNR.
Optimizing for Peak SNR
The first step in optimizing for peak SNR is to find a TIA gain
and LED level that gives the best performance where the
number of LED pulses remains constant. If peak SNR is the
goal, use the noise section of Table 4 as a guide. It is important
to note that the SNR improves as a square root of the number of
pulses averaged together, whereas the increase in the LED
power consumed is directly proportional to the number of LED
pulses. In other words, for every doubling of the LED pulse
count, there is a doubling of the LED power consumed and a 3 dB
SNR improvement. As a result, avoid any change in the gain
configuration that provides less than 3 dB of improvement for
a 2× power penalty; any TIA gain configuration that provides
more than 3 dB of improvement for a 2× power penalty is a
suitable choice. If peak SNR is the goal and there is no issue
saturating the photodiode with LED current at any gain, the
50 kΩ TIA gain setting is an optimal choice. After the SNR per
pulse per channel is optimized, the user can then increase the
number of pulses to achieve the desired system SNR.
Optimizing SNR per Watt in a Signal Limited System
In practice, optimizing for peak SNR is not always practical.
One scenario in which the PPG signal has a poor SNR is the
signal limited regime. In this scenario, the LED current reaches
an upper limit before the desired dc return level is achieved.
Tuning in this case starts where the peak SNR tuning stops. The
starting point is nominally a 50 kΩ gain, as long as the lowest LED
current setting of 8 mA does not saturate the photodiode and the
50 kΩ gain provides enough protection against intense background
light. In these cases, use a 25 kΩ gain as the starting point.
The goal of the tuning process is to bring the dc return signal to a
specific ADC range, such as 50% or 60%. The ADC range choice is
a function of the margin of headroom needed to prevent saturation
as the dc level fluctuates over time. The SNR of the PPG waveform
is always some percentage of the dc level. If the target level cannot
be achieved at the base gain, increase the gain and repeat the
procedure. The tuning system may need to place an upper limit
on the gain to prevent saturation from ambient signals.
Tuning the Pulse Count
After the LED peak current and TIA gain are optimized,
increasing the number of pulses per sample increases the SNR
by the square root of the number of pulses. There are two ways to
increase the pulse count. The pulse count registers (Register 0x31,
Bits[15:8], and Register 0x36, Bits[15:8]) change the number of
pulses per internal sample. Register 0x15, Bits[6:4] and Bits[10:8],
controls the number of internal samples that are averaged together
before the data is sent to the output. Therefore, the number of
pulses per sample is the pulse count register multiplied by the
number of subsequent samples being averaged. In general, the
internal sampling rate increases as the number of internal
sample averages increase to maintain the desired output data
rate. The SNR/Watt is most optimal with pulse count values of
16 or less. Above pulse count values of 16, the square root
relationship does not hold in the pulse count register. However,
this relationship continues to hold when averaged between
samples using Register 0x15.
Note that increasing LED peak current increases SNR almost
directly proportional to LED power, whereas increasing the
number of pulses by a factor of n results in only a nominal √(n)
increase in SNR.
When using the sample sum or average function (Register 0x15),
the output data rate decreases by the number of summed samples.
To maintain a static output data rate, increase the sample
frequency (Register 0x12) by the same factor as that selected
in Register 0x15. For example, for a 100 Hz output data rate
and a sample sum or average of four samples, set the sample
frequency to 400 Hz.
Applying a Reverse Bias to the Photodiode
The photodiode capacitance contributes to higher noise in the
signal path. Applying a reverse bias to the photodiode reduces
the capacitance of the photodiode, resulting in better noise
performance. To apply a reverse bias to the photodiode, set
Register 0x54, Bit 7 to 1. The actual reverse bias is then
determined by the settings in Register 0x54, Bits[11:10] for
Time Slot B and in Register 0x54, Bits[9:8] for Time Slot A.
Set these bits equal to 0x2 applies ~250 mV of reverse bias
across the photodiode. There is also an option of setting the
cathode of the PD equal to the positive supply voltage, which
can result in up to 0.9 V of reverse bias; however, any noise on
the supply is introduced directly into the signal so this may
actually result in higher noise levels. The recommended setting
is to set Register 0x54, Bits[11:10] and/or Register 0x54,
Bits[9:8] equal to 0x2 for an ~250 mV reverse bias.
Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.Downloaded from Arrow.com.