We (people actually in the calibration industry) often hear things being called calibrators that obviously aren’t. A great example is the Emerson HART(tm) communicator. No offense intended, but these types of devices aren’t calibrators, they are electronic “screwdrivers” that enable a technician or engineer to adjust the calibration of a device.

Here are my defining points for a calibrator.

1. Can make a precision measurement or generate a precision signal (i.e., analog, not digital). Or BOTH.
2. Has a calibration that is traceable to national standards.
3. Has a known accuracy or degree of calibration uncertainty.

There is a lot that could be added to this list, of course; things such as conformance to international standards, EX ratings…

Another little list of mine is this list of characteristics that make up a good calibrator.

• Accuracy
• Flexibility or versatility
• Ease of use
• Reliability
• Ergonomics (small, lightweight, easy to read display, etc.)
• Value

To me, a great example of all of the above is the Martel BETA MC-1200 Multi-function Calibrator. It’s got a long list of features, top knotch performance and is a great deal for the money. Total cost of ownership is good because all they ever require is annual recalibration which anyone with a computer and adequate standards can do.

Martel calibrator user Bruce Schmeck is quoted in the March, 2010, issue of Control Magazine saying, “We use field calibrators…for all kinds of biochemical batches to monitor and verify temperature, pressure, flow and agitation.”

In the same article, one of our competitors surveyed users and unsurprisingly discovered that a lot of them aren’t calibrating the instruments in their plants.

The article goes on to point out that many users have been sold a bill of goods about their instrumentation from the makers of that instrumentation. To wit,”our digital instruments don’t drift, so they don’t need to be calibrated.”

BetaGauge 321A Dual Range Pressure Calibrator

Mr. Schmeck also explained in this article that it isn’t the digital part of new instruments that drifts; it’s the part where the rubber meets the road, so to speak. He cites the example of the metal diaphragm in a pressure transmitter. The metal will age and its response characteristics will change over time. The ONLY way to fix that is to use a traceable field calibrator to inject a known valid signal.

To sum it up, here’s another quote from Bruce Schmeck that I like, “Most calibrators need to be re-calibrated once per year.” I love it.

I’m not referring to an old time game show or parlor game as they used to be known. My question is, “Can you trust your vendor’s calibration certificate?”

There are a few indicators you can use.

1. What kind of reputation does the company have in the industry? You know, how long have they been around and do other customers respect their products and service? There are a number of industry surveys and reports that can help with this if you don’t feel comfortable making your own assessment.
2. Is the certificate traceable to NIST or other national standards body? This may seem obvious, but read the statement see if it makes sense to you. Are there unsupported assertions about things like TUR (Total Uncertainty Ratio) or Guide 25 approval?
3. Are the standards used properly identified and are they appropriate for the task at hand? For example, if a specification is % of reading, but the standard’s specification is % of scale, there’s a significant mismatch. You may also need to check to see if the standard has been properly maintained with acceptable calibration intevals.
4. Does the certificate include data that is clearly presented and meaningful? Are there sufficient data points to assess performance?
5. If possible, check the calibration at intervals different from the cardinal points on the certificate. Even a bump check can provide revealing information.

Finally, is the calibration accredited under an international standard (ISO 17025) or do you need such an accredited calibration. Even if it’s offered as an option, the vendor must have a pretty good idea that the normal calibration is valid.

When someone mentions calibrator uncertainty, they are talking about what we call accuracy in our specifications. To be strict about it, our accuracy specifications are inclusive of most or all errors. In scientific applications, you might see these all broken out.

Typical examples are errors from linearity, hysteresis, ambient effects and others. In most cases all of these errors are combined into a comprehensive specification. That makes it easier for a user to determine the applicability of the calibrator for a given purpose. This number would then be the total uncertainty of the calibrator for a given calibration.

However, if the calibrator is to be used to make a measurement, you might also have to consider the uncertainty of the primary element or sensor. Let’s take the example of the Martel BETA PTC-8001 being used as a thermocouple thermometer to measure a process temperature. Here, you need to also consider the uncertainty of the thermocouple and thermocouple extension wire to make a determination of the total uncertainty.

In calibrating a transmitter, the sensor is not used, so there is no need to consider sensor error in that case. For resistance sensors (RTDs), you would have to consider an error (lead wire resistance) if you are using only a 2 wire connection. For 3 and 4 wire connections, the lead wire error is too small to be of consideration since we have a compensating measurement.

For calibrating a thermocouple instrument, you must use the correct type of thermocouple extension wire for the cold junction compensation to work correctly. There is a very small error even with the thermocouple extension wire, but it is too small to be of any practical concern. The length of the thermocouple extension wire should be kept short. We typically recommend no more than 1 meter.

We’ll have more on uncertainty in transmitter calibrations in the next post.