MAX1462 Datasheet, PDF (14/16 Pages) Maxim Integrated Products – Low-Voltage, Low-Power, 16-Bit Smart ADC

Low-Voltage, Low-Power,
16-Bit Smart ADC
Table 9. MAX1462 Conversion Timing
PARAMETER
Sensor Warmup Time
ADC Time
DSP Time
EOC Pulse Width
Conversion Time
SYMBOL
t_WARM
t_ADC
t_DSP
t_EOC
t_CONV
MIN
35
130,585
3220
8
133,805
MAX
—
130,585
3364
8
133,949
UNITS
ms
XIN clk cycles
XIN clk cycles
XIN clk cycles
XIN clk cycles
where
Equation (3)
x = ^{DL − DOFF}
D_S − D_OFF
Equations 2a, 2b, and 2c form a system of three linear
equations, with three unknowns, Of₀, Of₁, and Of₂.
Solve for Of₀, Of₁, and Of₂.
The small sensor excitation versions of equation 1 can
be ratioed to obtain:
Equation (4a)
(Y_CS − Y_AS ) + (G₁ T_AY_CS − T_CY_AS ) +
G₂


T_A2Y_CS
−
T_C2Y_AS


=
0
Equation (4b)
(Y_CS − Y_BS ) + (G₁ T_BY_CS − T_CY_BS ) +
G₂


T_B2Y_CS
−
T_C2Y_BS


=
0
Equations 4a and 4b form a system of two linear equa-
tions and two unknowns, G₁ and G₂. Solve for G₁ and
G₂. Equation 1 can now be readily solved for the last
unknown, Gain.
Arithmetic manipulation can magnify measurement
errors and noise. Quantization of the calibration coeffi-
cients is another reason to consider adjusting the Gain
and D_OFF coefficients. To do this, load the MAX1462
registers with the calculated coefficients Gain, G₁, G₂,
Of₀, Of₁, Of₂, and D_OFF. Assuming the oven is still at
temperature C and the S sensor excitation is still
applied, measure the output D_CS. Change to the L sen-
sor excitation, and measure D_CL. Compute the new
Gain coefficient using equation 6. Remeasure D_CL, and
compute the new D_OFF coefficient, given by equation 7.
Equation (6)
GAIN_new
=
Gain ^DL
D_CL
− D_S
− D_CS
Equation (7)
D_OFFnew = D_OFF + D_L − D_CL
where:
Equation (5a)
Y_AS
=
Signal_AS
D_S − D_OFF
+ Of₀ + Of₁T_A
+ Of₂ T_A ²
Equation (5b)
Y_BS
=
Signal_BS
D_S − D_OFF
+ Of₀ + Of₁T_B
+ Of₂ T_B2
Equation (5c)
Y_CS
=
Signal_CS
D_S − D_OFF
+ Of₀ + Of₁T_C
+ Of₂ T_C2
The final calibration coefficients may now be written
into the MAX1462 EEPROM. The unit is now ready for
final test.
This algorithm minimizes the error directly at the six test
conditions, AS, AL, BS, BL, CS, and CL. Space the
temperatures A, B, and C widely to minimize the SNR
of the measurement. If there is a large error remaining
in the finished product, move the calibration tempera-
tures closer to the peak error temperatures. Similarly,
full-scale sensor excitation may not be the best calibra-
tion condition if the sensor has nonlinearities. Move S
and L away from full scale.
14 ______________________________________________________________________________________