MIC3003GFL Datasheet, PDF (23/74 Pages) Micrel Semiconductor – FOM Management IC with Internal Calibration

Micrel, Inc.
RX Power
Received power is sensed as a voltage appearing at
VRX. It is assumed that this voltage is generated by a
sense resistor carrying the receiver photodiode current
or by the RSSI circuit of the receiver. The value returned
by the A/D is therefore a voltage analogous to received
power. The binary values in RXOPh and RXOPl are
related to receive power by:
RX(mW ) = K × VREF × (256 × RXOPh + RXOPl / 16) • 65536
(4)
For a given implementation, the constant, K, will likely
have to be determined through experimentation or
closed-loop calibration, as it depends upon the gain and
efficiencies of the receiver. In SFF-8472
implementations, the external calibration constants can
describe up to a fourth-order polynomial in case K is
nonlinear.
B/ Internal Calibration
If the INTCAL bit in OEMCFG3 is set to 1 (internal
calibration selected), the MIC3003 will process each
piece of data coming out of the A/D converter before
storing the result in result register. Linear slope/offset
correction will be applied on a per-channel basis to the
measured values for voltage, bias current, TX power,
and RX power. Only offset is applied to temperature.
The user must store the appropriate slope/offset
coefficients in memory at the time of transceiver
calibration. In the case of RX power, a look-up table is
provided that implements eight-segment piecewise-
linear correction. This correction may be performed as a
compensation of the receiver non-linearity over
temperature or receive power level. If static slope/offset
correction for RX power is desired, the eight coefficient
sets can simply be made the same. The user has the
option to select between using preset hard-coded
delimiters values or programmable delimiters where
delimiters corresponding to the best linear approximation
intervals of a specific receiver can be entered. The latter
option will use an additional fourteen (14) bytes from the
OEM scratch pad A6h:208-221(DO_h-DD_h). OEMCFG6
bits [6:5] are used to select between these options. The
memory maps for the calibration coefficients are shown
in Tables 11 and 12. If the programmable delimiters
option is selected, the user must enter the seven
delimiters of the intervals that best fit the receiver
response. The diagram in Figure 3 shows the link
between the delimiters and the sets of slopes and
offsets.
Slopes Coefficients
The slopes allow for the correction of gain errors. Each
slope coefficient is an unsigned, sixteen-bit, fixed-point
binary number in the format:
[mmmmmmmm .llllllll]
(5)
MIC3003GFL
where m is a data bit in the most-significant byte and l is
a data bit in the least significant byte
Slopes are always positive. The decimal point is in
between the two bytes, i.e., between bits 7 and 8. This
provides a numerical range of 1/256 (0.00391) to
255.997 in steps of 1/256. The most significant byte is
always stored in memory at the lower numerical
address.
Offset coefficients
The offsets correct for constant errors in the measured
data. Each offset, apart from temperature, is a signed,
sixteen-bit, fixed-point binary number. The bit-weights of
the offsets are the same as that of the final results. The
sixteen-bit offsets provide a numerical range of –32768
to +32767 for voltage, bias current, transmit power, and
receive power.
The numerical range for the six-bit temperature offset is
–32 (–16 °C) to +31 (+15.5 °C) in increments of .5 °C.
The two most significant bits of the temperature offset
coefficient are ignored by the MIC3003.
Computing Internal Calibration Results
Calibration of voltage, bias current, and TX power are
performed using the following calculation:
RESULTn = ADC _ RESULTn × SLOPEn + OFFSETn
(6)
Calibration of RX power is performed using the following
calculation:
RESULT = ADC _ RESULT × SLOPE(m) + OFFSET(m)
(7)
where m represents one of the eight linearization
intervals corresponding to the RX power level.
The results of these calculations are rounded to sixteen
bits. If the seventeenth bit is a one, the result is rounded
up to the next higher value. If the seventeenth bit is zero,
the upper sixteen bits remain unchanged. The bit-
weights of the offsets are the same as that of the final
results. For SFF-8472 compatible applications, these bit-
weights are given in Table 10.
Parameter
Voltage
Bias Current
TX Power
RX Power
Magnitude of LSB
100µV
2µA
0.1µW
0.1µW
Table 10. LSB Values of Offset Coefficients
July 2010
23
M9999-072910-A
[e-mail] or [phone]