|
Consider the following analogy. If you needed to measure Child A's height, which of the following methods do you think would be more accurate?
- Use a tape measure to measure the height of Child A from the floor to the top of Child A's head, or
- Use a tape measure to first measure the height of Child B, then, with Child B and Child A standing back to back, use a ruler to measure the difference in height between the two children?
Thermocouples, of course, work on the second principle. If the measurement of cold junction temperature is near-instantaneous and highly accurate - and the difference in temperature is measured precisely - AND the instrument environment is stable - it is possible to achieve reliable results. A problem, though, is that every degree of error in the cold junction temperature measurement directly impacts the thermocouple measurement. This error potential is complicated by the fact that a typical cold junction environment is subjected to a wide variety of heat influences - making it a highly dynamic reference - even when the instrument is located in a controlled lab.
To revisit the analogy of measuring Child A's height, it would be as if Child B, the reference, was constantly squirming and moving. In such a situation, the measurements can come under serious questioning.
Real world temperature problems
So how does this work in the pharmaceutical validation realm? Consider the following example of temperature measurements made inside a 40°C +/-1°C temperature-controlled incubator. In the graph below, the top (blue) line represents the temperature as measured by a thermocouple system. The other graph line represents data as measured by a thermistor-based data logger. Now, which is the correct information? Is the chamber really cycling significantly? Because if it is, it is out of specification.
Or are the conditions as steady as the data logger (with a thermistor sensor) indicates? The right answer could mean the difference between a costly repeat of the validation or moving on to the next application.

To find out the answer, it helps to see the complete picture. To do that, we'll include the information reported on the thermocouple probe's cold junction reference temperature. As can be seen in the graph below, the temperature peaks and valleys on the thermocouple line correlate strongly with the measurements made at the cold junction reference point. This correlation is indicative of one of the major problems of cold junction temperature measurements: thermal lag. Even though the thermocouple system has been designed to compensate for temperature fluctuations at the cold junction, the problem of thermal lag introduces errors that directly affect the accuracy of the thermocouple measurements. In this example, the result is a significant adverse effect on the data reported in the incubator.

All of which raises the question: If I'm using thermocouples and there is a temperature problem, is it a real problem? Or was it caused by something that happened at the cold junction? For example, if one observes a "temperature blip" in the data, was it caused by a failure within the chamber, or was it caused by a thermal draft onto the thermocouple instrument?
Accurate cold junction temperature measurement is difficult
Problems arise with cold junction temperature compensation in thermocouple instruments because it is difficult to accurately (and instantaneously) measure the actual cold-junction temperature. There will always be a thermal lag. Consequently, the goal of these instruments is to simply attempt to measure as close as physically possible - but errors will always exist. The problem is exacerbated when the cold junction temperature is in flux, as in the example above where the cold junction was affected by the normal cycling of the air conditioning system.
To minimize cold-junction error, validation engineers typically perform thermocouple calibration a few minutes after the measuring instrument is powered up, allowing the cold junction to stabilize after warm-up. However, in the field, measurement accuracy at the cold junction can be significantly affected by varying air flow patterns and the proximity of nearby heat sources such as people, chambers, or heating or cooling elements. In such dynamic conditions, it becomes difficult to obtain repeatable results and, consequently, thermal validations can often fail post-calibration procedures.
A more accurate solution
The solution? In most cases, a highly stable environment, careful procedures and a good deal of patience can help to minimize errors. Or better yet - if your temperature measurement ranges support it - use a temperature measuring instrument that does not rely on cold junction references. Alternatives include thermistor-based data loggers such as Veriteq's validatable data loggers which provide high accuracy absolute temperature measurements over the range of -40°C to 85°C.
|