Displacement transducers are everywhere in laboratory and industrial settings. From measuring minute changes in position of the Earth, rock samples in a mechanics laboratory, or positions of thrusters and control surfaces in aerospace. The DCDT (Direct Current Displacement Transformer) is a very high resolution and relatively low cost transducer making it a popular choice. While most characteristics of the DCDT have been well characterized by the manufacturer, we haven’t found much out there on any warm up timescales and associated drift. If you are new to LVDTs and DCDTs, be sure to checkout the TransTek Blog for some great transducer selection tips!
When Does Transducer Warm Up Matter?
Most of the time industrial applications turn the transducers on and they stay on for days, months, or even years. If they are power cycled the process they monitor likely has a startup time on the order of minutes to hours. Laboratory and field instruments are generally used in a much different way. Transducers are applied to samples in the laboratory and powered up before and experiment starts. The experiment may last hours before everything is shutdown. In the field we often turn off transducers when not in use to conserve power. Power conservation means smaller batteries, smaller solar panels, less shipping weight, and easier transportation to the deployment site.
Recently the team has been working on an outdoor measurement system that uses DCDTs to measure subsidence of soils in rapidly subsiding areas. The funny thing about geological measurements, is often a sampling interval of once per hour is considered “high rate” and is unprecedented. For systems with such long sampling intervals, at least from the electrical engineer’s perspective, it is very advantageous to power down the system into a low power state for most of the day. In fact, if we only sample for 1 minute of each hour over a day, that is 24 minutes out of 1440 minutes in the day, or about 1.6% of the day. That means we can easily save the majority of the power consumed by the system in an always on configuration simply by putting power hungry items to sleep or turning them off when not in use.
In doing this and experimenting on the bench we noticed some drift in the measurements that subsided after a while and did not seem to come back unless the system was powered off for some time. When powering DCDTs with near their maximum excitation voltage we have noted the cores can become quite warm. This lead us to suspect that it was probable that some combination of thermal expansion of the sensing core and a warm up of the electronics were contributing to the drift. In researching the subject of DCDT warm up we were surprised at how little information was available on the subject. Likely since 99% of the users of the transducers never turn them off! So we did what anyone would do – conducted our own tests and characterization in our environmental chamber.
We placed a DCDT and one of our new 24-bit RS-485 digitizer circuit boards inside the thermal chamber. They were powered off at the beginning of the test. We set the chamber to temperatures of -20℃, 0℃, and 40℃. After the chamber reached the temperature set-point we let the sensor soak for at least 30 minutes. The analog to digital converter was powered on so it would be thermally stable, but the DCDT was not. After 30 minutes the DCDT was powered on and readings taken at 0.5Hz.
The results were a bit surprising – yes, there is thermal drift during a “warm up” period, but the drift we were observing was absolutely miniscule. The drift was on the order of 2 or 3 millivolts which translates to about 0.0005 in (13 microns). So while there was drift, it was very small. In fact, it was well below the signals we were looking for. Really, this just tells us how amazing a 24-bit analog to digital converter is on a sensitive analog transducer.
We continued to test at other temperatures to see if the warm up time was dependent on the ambient conditions. Indoors this is not a huge concern as room temperatures may vary a few degrees, but outdoor swings can be many tens of degrees. Top that off with the fact that a lot of our equipment finds itself in very extreme conditions on ice sheets in Greenland or on volcanoes in Columbia.
Ambient temperatures do not appear to significantly impact the warm up times of the instrument. In general after about 15 minutes things appear to be mostly stable. You may be able to argue that at -20℃ the level off is slower. If that minor drift is critical for your application, like in precision measurement of surfaces in the lab, you should probably make sure you plug in your DCDTs a while before your experiment begins. Likely turning them on when coming in the morning of the experiment or getting a coffee after powering them on is sufficient. If your signal is large, then drift is likely no factor.
In fact, if you had a 16-bit analog to digital converter instead of 24-bit, the drift would be very difficult to even notice. It would be on the order of 20 or so bits of change! This is very near to the case of a higher resolution converter simply measuring noise with more resolution!
A few caveats before applying this information to your application. First, remember our displacement drift is for a 2 inch (50 mm) range transducer, if your transducer has a smaller or larger range it is likely that the percent change will translate, but we haven’t tested that yet! Also, the power supplied to the DCDT was from a linear regulator on-board the analog to digital converter which was also in the environmental chamber. Therefore we likely have some convolved warm up of the transducer and the power supply. For our system, that is the important number as that is the real operating condition of the system.
So, what did we decide in the end? For our application the resolution required is far above the drift observed. The drift is a exponential process though, as would be expected for a thermal process, so waiting just a few minutes can eliminate much of the drift. As long as our system warms up for the same amount of time for each measurement, we don’t expect any problems. We set our warm up time to be 5 minutes and made it a user adjustable parameter. Could it be 1 minute? Sure, but the power savings between 1 and 5 minutes was negligible in the total power budget.
The moral of the story is as our friend Elecia White says “everything is a temperature sensor, some things measure other stuff too.” If you don’t know what temperature effects are present in your system, you could very well be interpreting them as valid signals instead of noise. Contact us to find out how to calibration facilities can be used to help your team characterize thermal noise of your measurements systems and stop being lied to by your electronics.