Low Temperature Tests
1. Problem Formulation
The specified temperature ranges of the electronic components used by Soclair Electronic in the standard measurement transducer modules vary considerably. Most passive elements are specified for operation in the temperature range -55°C to +125°C (U.S. Department of Defense MIL temperature specification). However, some of the active components (particularly op-amps) the temperature range is specified as -40/+85°C or 0/70°C. MIL components are often difficult to obtain, very expensive and not available in all housings. For these reasons, MIL components are not normally used in most of the circuits. This, however, does not mean that these measurement transducers are unsuitable for operation at very low temperatures. As the following tests show, such circuits can sometimes function perfectly well at temperatures as low as -130°C. Although one possible problem is the increased failure rate: if the electronics are frequently subjected to low temperatures or, even worse, are often subjected to large and sudden temperature changes then the failure rate can increase dramatically. We have undertaken various experiments on a few typical modules in order to determine the behaviour of Soclair Electronic measurement transducers at very low temperatures.
2. Experimental Set-up
A commercial climate test cabinet was used for temperatures down to -40°C. In order to reduce the module temperature even more we built a cryostat, cooled by liquid nitrogen. The temperature was measured by two Pt-100 sensors. One sensor (foil sensor) made good thermal contact with the module and the other measured the internal temperature of the cryostat. A SOCLAIR PCMS1 plug-in card, mounted in a laptop computer (16 channels, 16 bit), was used to gather and process the data. This system was configured for two Pt-100 sensors (module/cryostat) and for the module output voltages. Thus, all the data could be measured and recorded. Initially we tested three different modules for Pt-100 sensors:
- RTM 90-C-P, programmed for a range of 0/100°C = 0-10 V. This relatively complex module contains approx. 120 components, including various op-amps (SMD and DIL) as well as a CMOS DC-DC converter.
- RTM 70-D, fixed range module, -30/70°C = 0-10 V. This is a relatively simple, low-cost module with only two op-amps.
- RTM 70-D, fixed range module, 0/100°C = 0-10 V.
A relatively small temperature range (100°C) was chosen deliberately since, within such small ranges, the output signal is particularly sensitive to drifts.
The sensors for the three modules were realised by precision resistors located outside the cryostat. They simulated various temperatures within the current measurement range of the module. These resistors were connected using a four wire system.
For 9 days the modules were subjected to temperatures down to -130°C and comprehensively measured. The table below shows a small extract from the large amount of data collected.
T1 and T2 are the internal temperature of the cryostat and the module respectively. These two temperatures can differ by a few degrees (heat inertia following temperature changes, warming of the module itself). Columns "RTM90", "RTM70-1", and "RTM70-2" show the output voltage of the three modules (in volts). Columns "TC90", "TC70-1", and "TC70-2" show the calculated temperature coefficients in ppm/K corresponding to an output temperature of 16°C and FS (10 V). In order to calculate the error in the output value, the temperature difference (from 16°C) is multiplied with the temperature coefficient in ppm/K and the result divided by 100,000.
Temperature Drift According to Specification:
RTM90C - for a measurement range of 100°C, a total error of approx. 200 ppm/K is specified (70 ppm/K gain drift, 130 ppm/K offset drift).
RTM70D - for a measurement range of 100°C, a total error of approx. 450 ppm/K is specified (150 ppm/K gain drift, 300 ppm/K offset drift).
These specifications are guaranteed maximum values where the typical (measured) values are usually only a fraction of these worst case figures.
4. Summary of Results
Down to -30°C the temperature coefficient was around 30 ppm/K, at lower temperatures it increased constantly, at -80°C to 150 ppm/K and below -90°C the output voltage collapsed altogether. As soon as the temperature reached -90°C again, the output voltage jumped to its expected value. This process, whereby the temperature was brought down as low as -130°C, could be repeated many times.
Down to -60°C the temperature coefficient of one module was around 50 ppm/K and another was around 100 ppm/K. At -110°C, one module increased to approx. 150 ppm/K, at -130°C it was 130 ppm/K (and 210 ppm/K for the other module). These values are still well within the specification (450 ppm/K). This module continued to operate correctly even at -130°C.
During the measurements, the modules were subjected to many temperature changes (between -130°C and room temperature). The measurement data could always be reproduced and an abnormal long term drift could not be determined.
We expect all the other Soclair Electronic measurement transducers to behave in a similar manner.
Under certain conditions erroneous results could be obtained when warming the modules: if the module temperature is still below zero but the air temperature above zero then humidity in the air condensed. For a short period of time, this conducting moisture film lead to errors of varying magnitude.
Solution: "clothe" the module so that no water can condense on it. If this is not possible then we recommend that the module be coated at the factory with a protective paint layer (please state when ordering).