There are a lot of factors involved in calibrating your device. Some things to consider:

- the expected range of your measurements
- temperature
- comparability and accuracy

The range of expected measurements must be known before the device is calibrated.

- A device calibrated for measuring hydroponic solutions won’t be able to measure the salinity of an aquarium.
- Conductivity isn’t a linear slope in practice. If a device is calibrated for 0.7 mS/cm to 2.0 mS/cm, you can expect a very close measurement inbetween that small range. If the range was wider, say from 0.7 mS/cm to 5.0 mS/cm, your measurements inbetween will very likely be off.

Temperature is very important in determining conductivity. Compensation is used to adjust the measured value, which may be taken at any temperature, to a particular temperature. For example, if you have a 1413 uS/cm calibration solution, it will only measure that at 25 C. If that solution was 20 C, it will measure 1278 uS/cm. You can either increase the temperature to 25 C, or apply compensation to your reading to get 1413 uS/cm. In an uncontrolled setting, you have to compensate to control for temperature.

- There are many different methods to compensate for temperature. The prevailing standard is to adjust measurements to 25 C. A measurement adjusted to 25 C is often times noted in someway (1.413 mS/cm 25°, 1.413 mS/cm @25 etc), however there is no standard and there may be no indication.
- The adjustment is dependant on several factors. The predominant factor being the ions in solution.
- There are linear and non-linear compensation algorithms.
- Coefficients are a factor of the calculation and vary by what is being measured.
- Any method is just an approximation.

If a device is calibrated without using temperature compensation, you cannot take compensated measurements accurately. In other words, you must always use or never use compensation to obtain valid results.

uFire devices use a linear calculation. For measuring freshwater, a coefficient of 1.9 is used, saltwater is 2.1. A different coefficient can be easily used with the library, and uncompensated measurements can be taken, giving a lot of latitude in implementing a custom routine.

It is important to recognize that there are many factors that go into deriving a conductivity measurement: temperature compensation is an approximation, there are several steps in the measurement that involve floating point math, coefficients, and choices in method etc., that all contribute to the final result.

Calibration solutions aren’t the same as the solution to be monitored. If the calibration solution is comprised of sodium chloride and what you are ultimately measuring isn’t isn’t sodium chloride, the resulting measure will be close, but not exact.

When comparing results, it is important to fully understand how one measurement was derived. Differing methods and conditions will produce slightly different results.