Datasheet

 2006-2012 Microchip Technology Inc. DS22025C-page 21
MCP3909
The multiplier output gives the product of the two
high-pass filtered channels, corresponding to instanta-
neous real power. Multiplying two sine wave signals by
the same ω frequency gives a DC component and a 2ω
component. The instantaneous power signal contains
the real power of its DC component, while also contain-
ing 2ω components coming from the line frequency
multiplication. These 2ω components come for the line
frequency (and its harmonics) and must be removed in
order to extract the real-power information. This is
accomplished using the low-pass filter and DTF
converter.
4.7 Active Power Low-Pass Filter and
DTF Converter
For the active power signal calculation, the MCP3909
uses a digital low-pass filter. This low-pass filter is a
first-order IIR filter, which is used to extract the active
real-power information (DC component) from the
instantaneous power signal. The magnitude response
of this filter is detailed in Figure 4-5. Due to the fact that
the instantaneous power signal has harmonic content
(coming from the 2 components of the inputs), and
since the filter is not ideal, there will be some ripple at
the output of the low-pass filter at the harmonics of the
line frequency.
The cut-off frequency of the filter (8.9 Hz) has been
chosen to have sufficient rejection for commonly-used
line frequencies (50 Hz and 60 Hz). With a standard
input clock (MCLK = 3.58 MHz) and a 50 Hz line fre-
quency, the rejection of the 2ω component (100 Hz) will
be more than 20 dB. This equates to a 2ω component
containing 10 times less power than the main DC com-
ponent (i.e., the average active real power).
FIGURE 4-5: LPF1 Magnitude Response
(MCLK = 3.58 MHz).
The output of the low-pass filter is accumulated in the
digital-to-frequency converter. This accumulation is
compared to a different digital threshold for F
OUT0/1
and HF
OUT
, representing a quantity of real energy mea-
sured by the part. Every time the digital threshold on
F
OUT0/1
or HF
OUT
is crossed, the part will output a
pulse (See Section 4.8 “Active Power FOUT0/1 and
HFOUT Output Frequencies”).
The equivalent quantity of real energy required to out-
put a pulse is much larger for the F
OUT0/1
outputs than
the HF
OUT
. This is such that the integration period for
the F
OUT0/1
outputs is much larger. This larger integra-
tion period acts as another low-pass filter so that the
output ripple due to the 2ω components is minimal.
However, these components are not totally removed,
since realized low-pass filters are never ideal. This will
create a small jitter in the output frequency. Averaging
the output pulses with a counter or a MCU in the appli-
cation will then remove the small sinusoidal content of
the output frequency and filter out the remaining 2
ripple.
HF
OUT
is intended to be used for calibration purposes
due to its instantaneous power content. The shorter
integration period of HF
OUT
demands that the 2ω com-
ponent be given more attention. Since a sinusoidal sig-
nal average is zero, averaging the HF
OUT
signal in
steady-state conditions will give the proper real energy
value.
-40
-35
-30
-25
-20
-15
-10
-5
0
0.1 1 10 100 1000
Frequency (Hz)
Normal Mode Rejection (dB)