Link This |
Email this |
Blog This |
Comments (0)
Measurement proverbs
October 31, 2008
The other day, I received some measurement proverbs that I'd like to share. The proverbs come from Karl Anderson, director of engineering at
Valid Measurements. Karl is also the inventor of the
Anderson Loop, an alternative to the Wheatstone Bridge. See the article we published about it in the January 1995 issue by
clicking here.
And now, measurement proverbs.
Measurement Systems Engineering is the art of Measurement Science.
The technical discipline of Measurement Systems Engineering (MSE) has a theoretical basis and body of knowledge comparable to the more widely recognized engineering disciplines. The key developer and ‘spiritual leader’ of this discipline is Prof. Peter K. Stein.
Highlighting the fundamental aspects of MSE can provide the basis for an engineer to apply wisdom in addition to knowledge in the process of making valid measurements. The best available resource I’ve encountered on this is the book ‘Applied Measurement Engineering’ by Chuck Wright.
What follows are some proverbs I’ve gleaned during my career as a measurement systems engineer. Most are expanded on in Chuck’s book.
A measurement always involves both a transfer of information and a transfer of energy.
The energy mentioned above is physical (material) stuff that is arranged or modified in order to encode/store/transfer information by means of the pre-established rules of a language.
On reflection, it becomes apparent that Information itself is not physical (material) stuff. That which is used to encode/store/transfer information is physical (material) stuff.
An ideal measurement is the indication one would obtain with perfect, noiseless instrumentation if it were not necessary to have the sensor present to disturb the energy field of the measured environment.
All measurements are made in order to inform a decision.
A valid measurement is an indication that will lead a knowledgeable observer to an appropriate decision. The measurement has all of the characteristics necessary to appropriately inform the observer. The observer is not misled by the data. (I would claim sole authorship of only this proverb. The others in this list I have expanded, adapted or edited.)
Measurements are made for only three purposes:
Design (so, what is really going on here),
Calibration (comparison to a standard) and
Control (providing feedback such that a process can be caused to approach its set point).
Each measurement purpose imposes significantly different requirements on the design of the measurement system.
All measurement approaches fall into one of two categories: Null balance and Offset.
Transducers fall generally into two categories: self-generating (like a thermocouple) and non-self-generating (like a potentiometer).
Measurement system noise has these same two characteristics: self-generating (arriving via capacitive, inductive or some other form of energy coupling) and non-self-generating (such as connecting wire resistance changes)
The measurement systems engineer must have an accurate understanding of the physics of the measurement process being employed.
(An unfortunate example: most engineers erroneously believe that a thermocouple’s output voltage is generated at its junctions.)
Every element in a measurement system will modify to some extent the measurement information it handles.
(Even a short wire has impedance and enables energy coupling to and from its environment.)
The response of a measurement system element is a function of its input conditions, its output conditions, its environment and its history.
Every element of a measurement system can be depended on to respond in every way it can to every characteristic of its past and present environment.
Calibrations must be accomplished in a manner that appropriately represents the relevant operating conditions of the measurement system in actual use. Otherwise calibrations will promote false confidence in the validity of measurements. (Example: Calibrating an undamaged wire segment of a damaged thermocouple.)
To be provably trustworthy, all measurement must be accompanied by validation information (preferably obtained automatically) which includes, but may not be limited to:
Appropriate calibrations
Assurance that the indications are primarily a function of what is desired to be observed
(Your response is really a function of the phenomenon you intend to measure.)
Assurance that the indications are insignificantly influenced by anything other than the phenomenon you intend to measure
(Noise of any kind from any source arriving by any path, including what you want to measure, is not invalidating the signal.)
Assurance that the measurement system has appropriate amplitude, frequency and phase response to deliver measurements with adequate sensitivity and fidelity
Assurance that a statistically significant end-to-end transfer function for the entire measurement system has been identified for every measurement channel.
With the above validation information in hand knowledgeable measurement systems engineers and users can bet their life on the data. They can quickly, politely and convincingly counter a charge that the data are “wrong”. (Another common descriptor is “garbage”.) Without it you have no real basis for confidence in the validity of data you analyze, deliver or receive.
Only with documentation proving that the measurement system is faithfully delivering what the sensor is feeling and is not materially effected by noise can the harried measurement systems engineer move the “garbage” monkey to the back where it belongs.
Posted by Martin Rowe on October 31, 2008 | Comments (0)
