Measurement Systems Engineering: Sampler (contd)

The Transient Case: A Sampler
Part 2 of 2

IN PART ONE, I established that we must always ask certain questions of our data before we accept them as evidence of what we are trying to measure. Now let's take this concern to another level . . .

Before we can be sure that the data are valid, we must also address the dynamics of the measurement system's own behavior. Such issues apply regardless of the physical phenomenon being measured!

[Two typical traces (superimposed): amplitude over time]

For example, consider these two typical records (shown superimposed). Various aspect of these traces can be misleading. In order to avoid fooling ourselves, we must question each of these aspects:
1.	High frequency in the measurand? .. Or, embellishment by the measurement system (such as transducer ringing, or ripple created in data handling)?

2.	Actual pulse shape? .. Or, distorted by the measurement system?
3.	Rise-time of the wave? .. Or, rise-time of the measurement system?
4.	Actual pulse peak? .. Or, peak altered by the measurement system's amplitude response (nonlinearity) or frequency response for magnitude (bandwidth) or phase (nonlinearity)?
5.	Actual cavitation, bounce-back, rarefaction, stress-reversal, etc.? .. Or, undershoot that is a characteristic of the measurement system?
6.	Permanent set in the measurand? .. Or, zero-shift in the measurement system?

Conclusion: .. Once again, we must address certain questions in order to prove that our data are "valid"---that is, to prove that we really are measuring what we think we are measuring.

(If you wish to learn how to prove methodically that data are valid, you may be interested in. "The Dynamics of Measurement Systems"---my second of two videotaped short courses consisting of lectures with integrated demonstrations. After viewing the first course, any engineer with exposure to differential equations can grasp these lectures and use the results.)

Return Home