There are typically a lot of components to the accuracy specification for the sensors we use in our HVAC systems, as illustrated below, which are the specifications extracted from the data sheet for a sensor I was working with a while back.
As you can see (depending on your screen’s resolution, you may have to click on the hyperlink to the actual document to really “see”), in addition to linearity, hysteresis, and
repeatability, things like ambient temperature, relative humidity, vibration, supply voltage and mounting position can come into play, impacting the accuracy of the current reading and the long term stability of the device. If the sensor is connected to a transmitter, many if not all of the issues affect the transmitter in addition to the actual sensing element, compounding the potential for error.
And, it doesn’t stop there; the wiring, scaling resistors, analog to digital (A to D) converters, change of value settings, digital to analog (D to A) converters and bi-focal glasses (is that a three or an eight?) that are in the path from the parameter we are measuring to our eyes and brains can all introduce an error. The Control Design Guide discusses some of these factors in detail if you are curious and want to know more.
When I talk about this stuff in class and out in the field, one of the more common, unrecognized potential sources of error is mounting position effects. For some sensors, like the one illustrated in Figure 1, where the sensing mechanism does not have to move to generate a signal, these effects are negligible. But for others, like the one illustrated below, where the sensing mechanism (a diaphragm) must move to generate a signal, the effects can be quite pronounced and significant.
The sensor is a low range differential pressure sensor that I was deploying yesterday along with a data logger to monitor filter performance and loading. It was set up to generate a signal that varied between 4 and 20 milliamps as the differential pressure across it varied from 0 to 1 inch water column. (That’s 0 to 0.0361 pounds per square inch for those of you who are more familiar with the psi unit of measure; bottom line is that it’s a very, very low
pressure being measured.)
While I am not intimately familiar with the inner workings of the sensor (and didn’t want to risk breaking it by taking it apart) the deflection of the diaphragm created by imposing a differential pressure across it is likely the mechanism used to generate an output. Since the pressure being measured is very small, the diaphragm would likely need to be very flexible and easily deflected to be able to move enough to generate a signal. Even at that, given the low level pressure being measured, its likely that the full scale deflection is fractions of an inch rather than inches.
If you consider the very low level parameter that is being measured combined with how gravity might interact with the diaphragm that is used to measure it when the diaphragm is in the horizontal vs. vertical position, you can probably begin to understand how mounting position might come into play with regard to the accuracy of this sensor. Irrespective of mounting position, the diagram will deflect as the signal moves through its full range. In addition, gravity will likely cause the diaphragm to deflect more when the transmitter is in the horizontal position rather than in the vertical position, irrespective of the applied signal. The transmitter electronics have no way of knowing that the sag or deflection produced by gravity is not a deflection produced by a differential pressure. Thus, the effect of gravity on the diaphragm will influence the output of the transmitter and the
effect will vary with the mounting position.
Theory aside, seeing is believing. Since I had all of my toys hooked up to make sure everything was working properly before Jesse Miller (the Kaiser Permanente Maintenance Assistant I was working with) and I crawled into the filter plenum to mount the sensors, I thought it might be interesting to show what happens to the sensor output as I rotated the sensor from horizontal to vertical and beyond vertical.
The following pictures illustrate what happened. The meter is reading milliamps dc and, as you can see, there is nothing connected to the transmitter’s sensing ports, which are the little nibs sticking out the of the bottom (you can see one of them right above my index finger).
Here is a video clip of me rotating the sensor for those who want to see it happen in real time. The quality is not great because I used the video mode on my 5 year old digital camera to generate it, but still, you can see the numbers on the meter change as I rotate the sensor.
Note the following:
- The specified mounting orientation in the manufacturer’s literature (the cut sheet illustrated above) is with the diaphragm in the vertical. (The manufacturer’s literature is the paper you see blowing around on the construction site when you visit it during the construction cycle, which may explain why some folks are not aware of mounting position effects).
- When I rotate the sensor to the vertical position, the output is virtually 4 ma, which is what it should be with no input pressure applied to the input ports.
- In the horizontal position, the output of the devices is in the range of 3.5 ma. Depending on whose system it was hooked up to, this output might generate an “Input Out of Range” error, or it may be converted to show a negative pressure of -0.03 inches w.c. (about 3% of full scale) when there was actually no differential pressure applied to the device.
- As I rotate the sensor beyond the vertical position by about 30°, the output increases to about 5.2 milliamps. This would correspond to a pressure of about 0.08 inches w.c. (8% of full scale) when there was actually no differential pressure present.
- The effect produced by mounting the sensor in a position other than vertical is not a linear function of the offset from vertical. In other words, rotating the sensor backward from vertical by 90° produced an error of 3% of full scale while rotating it forward from vertical by approximately 30° produced an error of 8%.
The bottom line is that how you mount a sensor (among other things) can have just as much impact on its accuracy as the quality of the components used in sensor itself. So, pay attention to the sensor mounting positions when you are out in the field developing a punch list or checking calibration. It could have a significant impact on the result you are trying to achieve.
Senior Engineer – Facility Dynamics Engineering