Experimental Methods Tips, Questions & Answers (Q & A)
Go to: Other Q&A * BACK to KOSTIC Homepage * ©1990-1997 M. KosticQ: Are the Lab Handouts posted on the Web sufficient for the lab assignments?
A:
Probably not. The lab handouts describe the assignments, apparatus and the processes to the extent necessary for the experiment execution and reduction of data. However, you have to review the related theoretical background from your science courses (i.e. prerequisites) in order to better understand the phenomena behind the lab experiments. Additional library or Internet research may be necessary and valuable for better understanding and comparison of your results with similar ones available elsewhere. The more references you use the better you'll understand the purpose and meaning of a lab assignment. The objective behind every lab is NOT to have a "show-and-tell" presentation, BUT to have a meaningful demonstration of physical phenomena with measurements of relevant quantities and application of as much as possible of the corresponding engineering analysis and comparison with the similar results from the references. The emphasis is more on quality than quantity.Q:
Will a multimeter read zero Volts if it is not connected to anything?
A: Not really. A good voltmeter may be picking up electrostatic charge in the air or any other surrounding noise. It should read zero Volts if you short its plus and minus terminal. Otherwise you should zero the meter or take into account its zero drift/shift.Q: If a thermocouple is connected to a precise voltmeter (both being in a room), will it read "zero" voltage corresponding to the zero temperature difference of its junctions?
A: The multimeter voltage reading should be zero when you measure room temperate without the reference junction (the thermocouple connected directly to the meter) Then the reference temperature is the same as measured temperature, both equal to the room temperature, therefore the temperature difference and generated voltage should be zeros. However, your reading will never be exactly zero due to many reasons: - if the tip of the thermocouple is wet it may be sub-cooled due to evaporation;
- the room temperature is non-uniform;
- he meter is not perfect (make a short-circuit to check its zero),
- the measured circuitry may be picking up some "noise," since it acts as an antenna; etc.
Q: If a thermocouple (being in a room) is connected to a precise voltmeter, will it read the table voltage corresponding to the room temperature?
A: Yes and No! Yes, if you have the reference junction in the ice-bath, since the table values are given for the zero reference-junction temperature. However, if only one-junction thermocouple (which is often the case) is connected to a voltmeter, the voltmeter circuitry will make up the other (reference) junction. Since the both junctions are at the same (room) temperature the reading should be zero volts, except for measurement errors. The reference junction is actually separated by the meter's circuitry, but as long as the meter terminals and circuitry are at the uniform temperature (any level) and accounted for with the corresponding emf value for that temperature, it should not matter, according to the rule of "intermediate metal interference."
Q:
How long does it take for a process to come to the steady state? How do we know it?
A: Good question! There is no an easy answer. It depends on the process and our criteria - what tolerances we may accept! Remember our "Dynamic response of a thermocouple sensor" lab (see the Figure). Every transient process under the constant extraneous conditions will eventually come to an equilibrium or (quasi) steady state. The transition is usually exponential with an asymptotic approach to the steady state (theoretically, it takes forever!). The rate of change is maximum at the beginning and decreases in time. You have to write down your measured reading (and to plot it if necessary). When the rate of change and/or the change itself become very small, you may assume that the steady state is achieved for your practical purposes. If the extraneous conditions are not constant, but oscillate, your final state may never be steady but oscillate to some degree. If the oscillations are relatively small with respect to the total change you may assume that you have a quasi steady state. In any event your measured quantity in a quasi steady state should not change for more than accepted uncertainty (error, tolerance) even if allowed to last forever. In other words, you should estimate a possible deficiency (error) of your steady state and take it into account in your uncertainty analysis. Only experience and knowledge of the process phenomena may help you know when you achieved the "tolerable" steady state. Usually, 90% of the theoretically possible change (100%) is good for most engineering purposes, for other more demanding applications 95% or 99% of the maximum possible change are needed. Letting your experiment process run for some time after you final measurement is the only way to confirm your judgement about the steady state within the tolerable errors (uncertainty). If it seems that the steady state do not come at all, check that the extraneous conditions and the process are under desired control. If nothing helps, you should try to give reasons and justifications for the outcome of your measurements. Remember that nothing is perfect, but everything is for a good reason!
Q: What is difference between an instrument resolution and accuracy and how repeating measurements and statistics relates to them?
A: Nothing is perfect and it relates to our definitions of concepts (our modeling and simplifications of reality) and our communications (whether we are talking about the same things). Resolution is the output scale precision, to what smallest part we are able to "resolve" an instrument reading. If it is a digital scale, it is the last digit and instrument rounds the reading to the closest displayed digit for us. If it is an analog scale, than it is the smallest fraction we could visually resolve the indicator position against the scale divisions, regardless whether they are marked or we imagine them (then somewhat subjective).
Accuracy of an instrument is separate from its display resolution and depends on quality of instrument components and calibration. For example, a scale may have display resolution in grams and systematically be off for say 10 grams - that is accuracy.
In addition to resolution and accuracy, when we repeat the measurements we usually get variability (scatter) in repetitive readings due to inevitable changes in time of instrument properties, measurement procedure and/or measured variable. Statistics helps here by describing and averaging ("smoothing") those variabilities. However, if the variabilities are smaller than the instrument resolution we may never detect and be aware of it. For example, when we measure a battery voltage with large instrument resolution (say 0.1V) the measurement reading appears very steady, but if we use a very precise voltmeter (say 0.0001V resolution) then we may be measuring (detecting) very dynamic fluctuation, which may be due to many things, like instabilities in instrument properties, measured signal, and/or interference and noise from the surrounding (regardless that we are displaying 0.0001V the accuracy may be off for 0.01V for example) - and that is always the case. So, the statistics (of many repetitive measurements) does not help with systematic instrument errors, the latter may be corrected with proper (re)calibration. Hope this will not confuse and be of help.
Q: Will
A: Not