Manual
15
Chapter 5: Computing the Temperature
A probe reading contains 15 values for product, water and temperature. The product and water values are 
each useful by themselves (a single product or water value may be converted with linear math to represent 
the oat position).
Unlike the product and water values, the temperature values are always accompanied by two reference 
values. A single temperature sensor value is of no use without its associated reference temperature 
values. The use of reference values eliminates the effects of circuit drift and produces exceptionally good 
repeatability. 
The 7100 Liquid Level Probe uses thermistors as the temperature sensing elements. All thermistors exhibit 
known, well dened non-linearity. In our application, the non-linearity of the temperature data is in the range of 
a few percent. (See Section 5.1: Improving Measurement Accuracy)
As shown in Section 5.1: Improving Measurement Accuracy, the rst step in computing the temperature of 
a given sensor is to use a linear equation to interpolate between the reference temperatures. The above 
mentioned non-linearity will then be present in the result. 
5.1: Improving Measurement Accuracy
The temperature measurement circuitry of the 7100 probe generates temperature pulse pairs with the time 
interval directly proportional to the resistance of a parallel connected thermistor (or reference temperature 
resistor) and a xed 37.4 K resistor. The 37.4 K parallel resistor serves to keep the time range of the 
temperature pulses within reasonable limits over the probe’s temperature range.
Variations in parameters of thermistors, time dening capacitors, and other parts of the probe introduce 
additional error to the temperature measurement. To reduce this error the 7100 probe transmits signals for 
two reference temperatures: +5 °C (low reference), and +50 °C (high reference). The following sequence of 
calculations compensates for errors using the reference signals, and takes into account the thermistor plus 
37.4 K resistor non-linearity.
Step 1 - Get reliable time counts for each temperature read, including the temperature references.
Example:
  1) Take 16 reads of a temperature.
  2) Discard 2 highest and 2 lowest readings.
  3) Take average value of the 12 remaining readings.
Step 2 - Calculate linear approximation of temperature (T
LIN
) using the following formula:
 T
LIN
 = [(R - L) * (H
REF
 - L
REF
) / (H - L)] + L
REF
Where:
  R = counts for thermistor,
  L = counts for low reference,
  H = counts for high reference,
 L
REF
 = low reference = 5 °C,
 H
REF
 = high reference = 50 °C.
T
LIN
 is a normalized value for temperature that compensates for errors caused by variations of the probe 
parts’ parameters. This is not actual temperature yet. It does not take into account non-linearity of the 
thermistor-parallel resistor combination.










