Davidson Measurement Tech Tips

Field Calibration - The Pitfalls & Precautions

Pressure measurement is accomplished with standard pressure gauges and digital pressure indicators which give a visual display, transducers and transmitters with electrical output and switches that trip when reaching a preset value. No matter what device and how it is used, it is essential to carry out periodic calibration that guarantees performance. Even with the latest technology, the "Smart" or "Intelligent'"' transmitters still need a pressure applied to the diaphragm to confirm they are functioning to the required standard.

Unfortunately, we see too many errors when calibrations are carried out that add to the uncertainty of, and compound the errors in, measurement. This tech tip is the start of a series which we hope will help you identify areas for improvement.

Many factors should be looked at before any calibrations are done. Questions to ask are:

1. What is it that we are going to calibrate?

2. What is the environment where the calibration is going to take place?

3. What happens to the environment during the calibration process?

4. Who is going to carry out the calibration?

5. What training is needed to create high standard results?

6. What reporting mechanisms should be introduced?

7. What goals do we want to set for calibration?

8. What calibration equipment is needed to meet the desired goals?

Other questions will arise over the next few issues so we will start with defining the devices to be calibrated. Each of the following devices has the associated uncertainty:

A. Pressure gauges range from 0.15% to 5%
B. Transducers and transmitters from 0.02% to 1%
C. Pressure indicators from 0.025% to 0.5%
D. Pressure switches from 0.5% to 5%
E. Pressure controllers from 0.02% to 0.5%

The figures above are used as a general guide, they will vary from device to device, however, the numbers would cover the great bulk of products.

To be able to achieve a useful calibration for any of the above, the general rule of thumb is to use a calibration device that is at least 5 times better than the product under test, prefer- ably 10 times better. Thus to achieve a calibration for a 0.15% gauge you will need to use a calibrator that has an accuracy of at least 0.03% thus you have to select the unit very carefully.

However there is another factor to consider, that is the difference between percent full scale (%FS) and percent of reading (%rdg).

Terminal Straight Line (TSL)

This error defines the non-linearity curve for the device being used. In essence, the curve will have an error calculated which has the end points as the reference points for the curve. The graph below shows the relationship:

The above is obviously exaggerated, however, it is done to clearly show that the plot must go through the end points for the terminal straight line error to be quoted. The difference between the calibrated curve, the thick dark line. and the linear dotted line becomes the quoted error for non-linearity.

Best Straight Line (BSL)

Now. the BSL error is the line of best fit where the end points do not necessarily lie on the full scale and zero values for the sensor. Graphically, it can be represented by:

 With a BSL calibration it is important that you know the values for the zero and full scale CALIBRATED figures; you can then set your monitoring electronics accordingly.

A line of best fit, or BSL, will generally give you a better accuracy than the TSL figure.

Now to touch on the pitfalls when reranging any adjustable sensor, let's stick with the pressure transmitter.

Let's say the sensor has a full scale value of 1000 kPa and it has a turn down ratio of 10:1. which allows us to change the full scale to 100 kPa at the lowest end. If we did adjust it to 100 kPa our errors still refer to the upper range limit or 1000 kPa. So if our non-linearity was 0.1%, then the figures become:

0.1% of 1000 kPa = 1 kPa

1 kPa of 100 kPa = 1%

– a TEN FOLD increase in error.

Obviously, this may be reduced by a new calibration. but how often is that carried out? Now add the other errors of hysteresis, repeatability and above all temperature effects, then the errors magnify.

So the next time you have to rerange an adjustable sensor. do your calculations to realise the real error at the new range.