TMP03/TMP04
–5–
Table I. Counter Size and Clock Frequency Effects on Quantization Error
Maximum Maximum Maximum Quantization Quantization
Count Available Temp Required Frequency Error (25C) Error (77F)
4096 125°C 94 kHz 0.284°C 0.512°F
8192 125°C 188 kHz 0.142°C 0.256°F
16384 125°C 376 kHz 0.071°C 0.128°F
Optimizing Counter Characteristics
Counter resolution, clock rate, and the resultant temperature
decode error that occurs using a counter scheme may be deter-
mined from the following calculations:
1. T1 is nominally 10 ms, and compared to T2 is relatively
insensitive to temperature changes. A useful worst-case
assumption is that T1 will never exceed 12 ms over the
specified temperature range.
T1 max = 12 ms
Substituting this value for T1 in the formula, temperature
(°C) = 235 – ([T1/T2] × 400), yields a maximum value of
T2 of 44 ms at 125°C. Rearranging the formula allows the
maximum value of T2 to be calculated at any maximum
operating temperature:
T2 (Temp) = (T1max × 400)/(235 – Temp) in seconds
2. We now need to calculate the maximum clock frequency we
can apply to the gated counter so it will not overflow during
T2 time measurement. The maximum frequency is calculated
using:
Frequency (max) = Counter Size/ (T2 at maximum
temperature)
Substituting in the equation using a 12-bit counter gives,
Fmax = 4096/44 ms ⯝ 94 kHz.
3. Now we can calculate the temperature resolution, or quanti-
zation error, provided by the counter at the chosen clock
frequency and temperature of interest. Again, using a 12-bit
counter being clocked at 90 kHz (to allow for ~5% tempera-
ture over-range), the temperature resolution at 25°C is
calculated from:
Quantization Error (
°
C) = 400
×
([Count1/Count2] –
[Count1 – 1]/[Count2 + 1])
Quantization Error (
°
F) = 720
×
([Count1/Count2] –
[Count1 – 1]/[Count2 + 1])
where, Count1 = T1max × Frequency, and Count2 =
T2 (Temp) × Frequency. At 25°C this gives a resolution of
better than 0.3°C. Note that the temperature resolution
calculated from these equations improves as temperature
increases. Higher temperature resolution will be obtained by
employing larger counters as shown in Table I. The internal
quantization error of the TMP03 sets a theoretical minimum
resolution of approximately 0.1°C at 25°C.
Self-Heating Effects
The temperature measurement accuracy of the TMP03 may be
degraded in some applications due to self-heating. Errors intro-
duced are from the quiescent dissipation, and power dissipated
by the digital output. The magnitude of these temperature er-
rors is dependent on the thermal conductivity of the TMP03
package, the mounting technique, and effects of airflow. Static
dissipation in the TMP03 is typically 4.5 mW operating at 5 V
with no load. In the TO-92 package mounted in free air, this
accounts for a temperature increase due to self-heating of
∆T = P
DISS
× θ
JA
= 4.5 mW × 162°C/W = 0.73°C (1.3°F)
For a free-standing surface-mount TSSOP package, the tem-
perature increase due to self-heating would be
∆T = P
DISS
× θ
JA
= 4.5 mW × 240°C/W = 1.08°C (1.9°F)
In addition, power is dissipated by the digital output which is
capable of sinking 800 µA continuous (TMP04). Under full
load, the output may dissipate
P
DISS
= 0.6 V
()
0.8 mA
()
T 2
T1 +T 2
For example, with T2 = 20 ms and T1 = 10 ms, the power
dissipation due to the digital output is approximately 0.32 mW
with a 0.8 mA load. In a free-standing TSSOP package, this
accounts for a temperature increase due to output self-heating
of
∆T = P
DISS
× Θ
JA
= 0.32 mW × 240°C/W = 0.08°C (0.14°F)
This temperature increase adds directly to that from the quies-
cent dissipation and affects the accuracy of the TMP03 relative
to the true ambient temperature. Alternatively, when the same
package has been bonded to a large plate or other thermal mass
(effectively a large heatsink) to measure its temperature, the
total self-heating error would be reduced to approximately
∆T = P
DISS
× Θ
JC
= (4.5 mW + 0.32 mW) × 43°C/W = 0.21°C (0.37°F)
Calibration
The TMP03 and TMP04 are laser-trimmed for accuracy and
linearity during manufacture and, in most cases, no further
adjustments are required. However, some improvement in per-
formance can be gained by additional system calibration. To
perform a single-point calibration at room temperature, measure
the TMP03 output, record the actual measurement tempera-
ture, and modify the offset constant (normally 235; see the
Output Encoding section) as follows:
Offset Constant = 235 + (T
OBSERVED
– T
TMP03OUTPUT
)
A more complicated 2-point calibration is also possible. This
involves measuring the TMP03 output at two temperatures,
Temp1 and Temp2, and modifying the slope constant (normally
400) as follows:
Slope Constant =
Temp 2 −Temp1
T1@Temp1
T 2@Temp1
−
T1@Temp 2
T 2@Temp 2
where T1 and T2 are the output high and output low times,
respectively.
REV. B
