Datasheet

LTC2990
10
2990fc
APPLICATIONS INFORMATION
unsigned number and the measured Kelvin (unsigned)
temperature represented as a 16-bit number, yields a
32-bit unsigned result. To scale this number back to a
13-bit temperature (9-bit integer part, and a 4-bit frac-
tional part), divide the number by 2
15
per Equation (5).
Similarly, Celsius coded temperature values can be scaled
using 16-bit fi xed-point arithmetic, using Equation (6).
In both cases, the scaled result will have a 9-bit integer
(d[12:4]) and the 4LSBs (d[3:0]) representing the 4-bit
fractional part. To convert the corrected result to decimal,
divide the fi nal result by 2
4
or 16, as you would the reg-
ister contents. If ideality factor scaling is implemented
in the target application, it is benefi cial to confi gure the
LTC2990 for Kelvin coded results to limit the number of
math operations required in the target processor.
T
K_COMP
=
Unsigned
()
η
CAL
η
ACT
2
15
T
K_MEAS
2
15
(5)
T
C_COMP
=
Unsigned
()
η
CAL
η
ACT
2
15
T
C_MEAS
+273.15 2
4
()
2
15
– 273.152
4
(6)
Sampling Currents
Single-ended voltage measurements are directly sampled
by the internal ADC. The average ADC input current is a
function of the input applied voltage as follows:
I
IN(AVG)
= (V
IN
– 1.49V) • 0.17[μA/V]
Inputs with source resistance less than 200Ω will yield
full-scale gain errors due to source impedance of <1/2LSB
for 14-bit conversions. The nominal conversion time is
1.5ms for single-ended conversions.
Current Measurements
The LTC2990 has the ability to perform 14-bit current
measurements with the addition of a current sense resis-
tor (see Figure 3).
In order to achieve accurate current sensing a few de-
tails must be considered. Differential voltage or current
measurements are directly sampled by the internal ADC.
The average ADC input current for each leg of the differ-
ential input signal during a conversion is (V
IN
– 1.49V)
• 0.34[μA/V]. The maximum source impedance to yield
14-bit results with, 1/2LSB full-scale error is ~50Ω. In
order to achieve high accuracy 4-point, or Kelvin con-
nected measurements of the sense resistor differential
voltage are necessary.
In the case of current measurements, the external sense
resistor is typically small, and determined by the full-scale
input voltage of the LTC2990. The full-scale differential
voltage is 0.300V. The external sense resistance is then a
function of the maximum measurable current, or R
EXT_MAX
= 0.300V/I
MAX
. For example, if you wanted to measure a
current range of ±5A, the external shunt resistance would
equal 0.300V/5A = 60mΩ.
There exists a way to improve the sense resistors precision
using the LTC2990. The LTC2990 measures both differential
voltage and remote temperature. It is therefore, possible
to compensate for the absolute resistance tolerance of the
sense resistor and the temperature coeffi cient of the sense
resistor in software. The resistance would be measured
by running a calibrated test current through the discrete
resistor. The LTC2990 would measure both the differential
voltage across this resistor and the resistor temperature.
From this measurement, R
O
and T
O
in the equation be-
low would be known. Using the two equations, the host
microprocessor could compensate for both the absolute
tolerance and the TCR.
R
T
= R
O
• [1 + α(T – T
O
)]
where:
α = +3930 ppm/°C for copper trace
α = ±2 to ~+200ppm/°C for discrete R (7)
I = (V1 – V2)/R
T
(8)
Figure 3. Simplifi ed Current Sense Schematic
V1 V2
LTC2990
0V – V
CC
R
SENSE
I
LOAD
2990 F03