Introduction to Errors
What is Error?
-
Error = Measured Value - True Value
-
Absolute Error: \(E_a = A_m - A_t\)
-
Relative Error: \(E_r = \dfrac{A_m - A_t}{A_t}\)
-
Percentage Error: \(E_p = \dfrac{A_m - A_t}{A_t} \times 100\%\)
-
Fractional Error: \(E_f = \dfrac{A_m - A_t}{A_m}\)
Types of Errors
Classification of Errors
-
Systematic Errors (Determinate)
-
Instrumental errors
-
Environmental errors
-
Observational errors
-
-
Random Errors (Indeterminate)
-
Unpredictable variations
-
Follow statistical laws
-
-
Gross Errors (Blunders)
-
Human mistakes
-
Misreading instruments
-
Systematic Errors - Details
Instrumental Errors:
-
Inherent shortcomings of instruments
-
Calibration errors
-
Loading effects
-
Aging of components
-
Non-linearity
Environmental Errors:
-
Temperature variations
-
Humidity effects
-
Vibrations
-
External electromagnetic fields
-
Pressure variations
Observational Errors:
-
Parallax error
-
Approximation in reading
Accuracy and Precision
Accuracy vs Precision
Accuracy:
-
Closeness to true value
-
Related to systematic errors
-
\(Accuracy = \dfrac{|\text{True Value - Measured Value}|}{\text{True Value}} \times 100\%\)
Precision:
-
Reproducibility of measurements
-
Related to random errors
-
Measure of scatter in repeated measurements
Key Point: High precision doesn’t guarantee high accuracy!
Resolution and Sensitivity
Resolution and Sensitivity
Resolution:
-
Smallest measurable change in input
-
Determines the fineness of measurement
-
\(Resolution = \dfrac{\text{Full Scale Range}}{\text{Number of Divisions}}\)
Sensitivity:
-
Ratio of change in output to change in input
-
\(Sensitivity = \dfrac{\Delta \text{Output}}{\Delta \text{Input}}\)
-
Higher sensitivity means better measurement capability
Threshold: Minimum input required to produce a detectable output
Statistical Analysis
Statistical Parameters
For n measurements: \(x_1, x_2, x_3, ..., x_n\)
Arithmetic Mean:
Deviation:
Average Deviation:
Standard Deviation and Variance
Standard Deviation:
Variance:
For Sample (n-1):
Probable Error
Probable Error
Definition: 50% probability that error lies within probable error limits
Probable Error:
Probable Error of Mean:
Standard Error of Mean:
Limiting Errors
Limiting Errors
Definition: Maximum error that can occur in measurement
For Addition/Subtraction:
For Multiplication/Division:
For Power: If \(R = A^n\), then \(\dfrac{\Delta R}{R} = \pm n \times \dfrac{\Delta A}{A}\)
Measurement Error Propagation
Error Propagation Formulas
For function \(R = f(A, B, C, ...)\)
General Formula:
Common Cases:
-
\(R = A + B\): \(\Delta R = \sqrt{(\Delta A)^2 + (\Delta B)^2}\)
-
\(R = A \times B\): \(\dfrac{\Delta R}{R} = \sqrt{\left(\dfrac{\Delta A}{A}\right)^2 + \left(\dfrac{\Delta B}{B}\right)^2}\)
-
\(R = \dfrac{A}{B}\): \(\dfrac{\Delta R}{R} = \sqrt{\left(\dfrac{\Delta A}{A}\right)^2 + \left(\dfrac{\Delta B}{B}\right)^2}\)
Instrument Errors
Instrument Error Classifications
Static Errors:
-
Calibration error
-
Offset error
-
Gain error (Scale factor error)
-
Non-linearity error
-
Hysteresis error
Dynamic Errors:
-
Lag error
-
Loading error
-
Insertion error
Loading Effects
Voltage Measurement:
-
Voltmeter draws current from circuit
-
\(R_{voltmeter} >> R_{circuit}\) (ideally infinite)
-
Loading error \(= \dfrac{R_{circuit}}{R_{circuit} + R_{voltmeter}} \times 100\%\)
Current Measurement:
-
Ammeter adds resistance to circuit
-
\(R_{ammeter} << R_{circuit}\) (ideally zero)
-
Loading error depends on ammeter resistance
Calibration and Standards
Calibration
Definition: Process of comparing instrument readings with known standards
Types:
-
Static calibration (DC conditions)
-
Dynamic calibration (AC/transient conditions)
Calibration Curve:
-
Graph of output vs input
-
Determines linearity
-
Used for error correction
Traceability: Chain of calibrations to national/international standards
Significant Figures
Significant Figures
Rules:
-
All non-zero digits are significant
-
Zeros between non-zero digits are significant
-
Leading zeros are not significant
-
Trailing zeros after decimal point are significant
-
Trailing zeros in whole numbers may/may not be significant
Operations:
-
Addition/Subtraction: Result has same decimal places as least precise
-
Multiplication/Division: Result has same significant figures as least precise
Normal Distribution and Confidence Intervals
Normal Distribution
Properties:
-
Bell-shaped curve
-
Mean = Mode = Median
-
68% data within \(\pm 1\sigma\)
-
95% data within \(\pm 2\sigma\)
-
99.7% data within \(\pm 3\sigma\)
Confidence Intervals:
-
68% confidence: \(\bar{x} \pm \sigma\)
-
95% confidence: \(\bar{x} \pm 2\sigma\)
-
99.7% confidence: \(\bar{x} \pm 3\sigma\)
Rejection of Data
Rejection of Outliers
Chauvenet’s Criterion:
-
Reject data if probability of occurrence < \(\dfrac{1}{2n}\)
-
Calculate deviation: \(d = |x_i - \bar{x}|\)
-
If \(\dfrac{d}{\sigma} > \text{threshold}\), reject the data
Q-Test (Dixon’s Test):
-
Compare Q with tabulated values
-
If Q > Q(table), reject the data
Measurement System Performance
Performance Characteristics
Static Characteristics:
-
Accuracy and Precision
-
Resolution and Sensitivity
-
Linearity and Hysteresis
-
Stability and Drift
-
Range and Span
Dynamic Characteristics:
-
Speed of response
-
Lag and Dead time
-
Fidelity
-
Dynamic error
Important Formulas
Key Formulas for GATE
-
Relative Error: \(E_r = \dfrac{A_m - A_t}{A_t}\)
-
Percentage Error: \(E_p = E_r \times 100\%\)
-
Standard Deviation: \(\sigma = \sqrt{\dfrac{\sum(x_i - \bar{x})^2}{n}}\)
-
Probable Error: \(P.E. = 0.6745 \times \sigma\)
-
Standard Error of Mean: \(S.E._m = \dfrac{\sigma}{\sqrt{n}}\)
-
Error Propagation (Product): \(\dfrac{\Delta R}{R} = \sqrt{\left(\dfrac{\Delta A}{A}\right)^2 + \left(\dfrac{\Delta B}{B}\right)^2}\)
-
Resolution: \(\dfrac{\text{Full Scale Range}}{\text{Number of Divisions}}\)
-
Sensitivity: \(\dfrac{\Delta \text{Output}}{\Delta \text{Input}}\)
GATE Tips
GATE Exam Tips
Important Points:
-
Systematic errors can be reduced by calibration
-
Random errors can be reduced by taking multiple readings
-
Accuracy depends on systematic errors
-
Precision depends on random errors
-
Probable error represents \(50\%\) confidence level
-
Standard deviation represents \(68\%\) confidence level
-
Always consider significant figures in final answer
-
Loading effects are crucial in voltage/current measurements
Practice: Error propagation problems are common in GATE
Numerical Example
Sample GATE Problem
Problem: A resistor has a nominal value of \(100~\Omega\) with \(\pm 2\%\) tolerance. A voltage of 10 V with \(\pm 1\%\) accuracy is applied. Find the percentage error in power calculation.
Solution:
-
Power \(P = \dfrac{V^2}{R}\)
-
\(\dfrac{\Delta P}{P} = \sqrt{\left(2\dfrac{\Delta V}{V}\right)^2 + \left(\dfrac{\Delta R}{R}\right)^2}\)
-
\(\dfrac{\Delta P}{P} = \sqrt{(2 \times 1\%)^2 + (2\%)^2}\)
-
\(\dfrac{\Delta P}{P} = \sqrt{4 + 4} = 2.83\%\)