Mech 430: Final Formula Sheet

Sensor Static and Dynamic Characteristics:

Overall Accuracy of a System 𝐴𝑠𝑦𝑠𝑡𝑒𝑚 = √𝑠𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑠

𝑛 𝑛

Δ = ∑(𝑦𝑖 − 𝑦) = ∑(𝑦𝑖 − 𝑏 − 𝑚𝑥𝑖 )2

2 2

𝑖=1 𝑖=1
Regression 𝑛 ∑𝑛 𝑛 𝑛
𝑖=1 𝑥𝑖 𝑦𝑖 – (∑𝑖=1 𝑥𝑖 )(∑𝑖=1 𝑦𝑖 )
Least Squares Method 𝑚= 2
𝑛(∑𝑛 2 𝑛
𝑖=1 𝑥𝑖 )−(∑𝑖=1 𝑥𝑖 )
(∑𝑛𝑖=1 𝑥𝑖2 )(∑𝑛𝑖=1 𝑦𝑖 ) − (∑𝑛𝑖=1 𝑥𝑖 𝑦𝑖 )(∑𝑛𝑖=1 𝑥𝑖 )
𝑏= 2
𝑛(∑𝑛𝑖=1 𝑥𝑖2 ) − (∑𝑛𝑖=1 𝑥𝑖 )

𝑦
Gain 𝐾=
𝑥

Full Scale Input 𝐹𝑆 = 𝑥𝑚𝑎𝑥 − 𝑥𝑚𝑖𝑛

Full Scale Output 𝐹𝑆𝑂 = 𝑦𝑚𝑎𝑥 − 𝑦𝑚𝑖𝑛

1 1 (𝑥 − 𝑥 ′ )2
Normal Distribution 𝑝(𝑥) = exp [− ]
𝜎√2𝜋 2 𝜎2

𝑃(−𝑧1 < 𝛽 < 𝑧1 ) = 2𝑃(𝑧1 )

Interval Probabilities 𝑥1 − 𝑥′
𝑧1 =
𝜎
1
Sample mean: 𝑥̅ = 𝑁 ∑𝑁
𝑖=1 𝑥𝑖
1
Sample variance: 𝑆𝑥 2 = 𝑁−1 ∑𝑁
𝑖=1(𝑥𝑖 − 𝑥̅ )
2

Finite Statistics Sample standard deviation: 𝑆𝑥 = √𝑆𝑥 2

𝑥𝑖 = 𝑥̅ ± 𝑡𝑣,𝑃 𝑆𝑥 (𝑃%)
𝑆𝑥
Standard deviation of means: 𝑆𝑥̅ =
√𝑁
True mean value 𝑥′ = 𝑥̅ ± 𝑡𝑣,𝑃 𝑆𝑥̅ (𝑃%) 𝑣 = 𝑁 − 1
Chauvenet’s Criterion (Data 1 |𝑥𝑖 −𝑥̅ |
(1 − 2𝑃(𝑧0 )) < 2𝑁 𝑧0 =
Outliers) 𝑆𝑥
Three-Sigma Test 𝑥̅ ± 𝑡𝜈,99.7 𝑆𝑥
𝑆𝑥
𝐶𝐼 = ±𝑡𝑣,𝑃 𝑆𝑥̅ = ±𝑡𝑣,𝑃 (𝑃%)
Measurements required √𝑁
𝐶𝐼 𝑆𝑥
𝑑= = 𝑡𝑣,𝑃
2 √𝑁
 Design-Stage: 𝑢𝑑 = √𝑢02 + 𝑢𝑐2

RSS Method: 𝑢𝑥 = ±√∑𝐾 2

𝑘=1 𝑢𝑘

1
Zero-order: 𝑢0 = ± 2 𝑟𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 (95%)

𝑑𝑦
Error Propagation: 𝑢𝑦 = (𝑑 ) 𝑢𝑥
𝑥 𝑥=𝑥̅
Design-Stage Uncertainty
Multiple Variables: 𝑅 ′ = 𝑅̅ ± 𝑢𝑅 (𝑃%)

𝜕𝑅
𝑅̅ = 𝑓(𝑥̅1 , 𝑥̅2 , … , 𝑥̅𝐿 ) | 𝑢𝑅 = ±√∑𝐿𝑖=1(𝜃𝑖 𝑢𝑥̅ 𝑖 )2 | 𝜃𝑖 = (𝜕𝑥 )
𝑖 𝑥=𝑥̅
∆𝑅
Principle of Equal Effects: ∆𝑢𝑖 =
𝜕𝑅
𝑛( )
𝜕𝑥𝑖

Measurement Uncertainty: 𝑢𝑥 = 𝑡𝑣,𝑃 √𝑏𝑥̅ 2 + 𝑆𝑥̅ 2 (𝑃%)

Random Error Uncertainty:

𝑆𝑥̅ = √(𝑆𝑥̅ )12 + (𝑆𝑥̅ )22 + ⋯ + (𝑆𝑥̅ )2𝑘

(𝑆𝑥 )𝑘
(𝑆𝑥̅ )𝑘 =
Multiple Measurement Uncertainty √𝑁𝑘

(∑𝐾 2 2
𝑘=1(𝑆𝑥̅ )𝑘 )
𝑉𝑠 = 𝑣𝑠𝑘 = 𝑁𝑘 − 1
(𝑆𝑥̅ )4𝑘
∑𝐾
𝑘=1 𝑣
𝑠𝑘
Systematic Standard Uncertainty:
𝑏𝑥̅ = √(𝑏𝑥̅ )12 + (𝑏𝑥̅ )22 + ⋯ + (𝑏𝑥̅ )2𝑘

Combined Standard Uncertainty: 𝑢𝑥 = √𝑏𝑥̅ 2 + 𝑆𝑥̅ 2

2
(∑𝐾 2 2
𝑘=1(𝑆𝑥̅ )𝑘 +(𝑏𝑥̅ )𝑘 )
Expanded Uncertainty: 𝑣 = 4 4
(𝑆𝑥̅ )𝑘 (𝑏𝑥̅ )𝑘
∑𝐾
𝑘=1 +∑𝐾
𝑘=1
𝑣𝑠 𝑣𝑏
𝑘 𝑘

1
Propagation of Uncertainty: 𝑢𝑅 = 𝑡𝑣,𝑃 [𝑏𝑅2 + 𝑠𝑅2 ]2 (𝑃%)
 1 1
2 2 2 2
𝑏𝑅 = (∑𝐿𝑖=1[𝜃𝑖 𝑏𝑥̅ 𝑖 ] ) 𝑠𝑅= (∑𝐿𝑖=1[𝜃𝑖 𝑠𝑥̅ 𝑖 ] )

2 2 2
(∑𝐿𝑖=1(𝜃𝑖 𝑠𝑥̅ 𝑖 ) + (𝜃𝑖 𝑏𝑥̅𝑖 ) )
𝑣= 4 4
(𝜃 𝑠 ) (𝜃 𝑏 )
∑𝐿𝑖=1 𝑖 𝑥̅𝑖 + ∑𝐿𝑖=1 𝑖 𝑥̅𝑖
𝑣𝑠𝑖 𝑣𝑏𝑖

System Response:
First Order System 𝜏𝑦̇ + 𝑦 = 𝐾𝐹(𝑡)

Step Function Input

Time response (for 𝑡 ≥ 0+):
−𝑡
𝑦(𝑡) = 𝐾𝐴 + (𝑦0 − 𝐾𝐴)𝑒 𝜏
Steady response + Transient response
𝐴𝑈(𝑡)
0, for 𝑡 ≤ 0−
={
𝐴, for 𝑡 ≥ 0+

𝐹(𝑡) = 𝐴𝑠𝑖𝑛𝜔𝑡 (Periodic input signal)

Periodic Function
Input
System Equation:
𝜏𝑦̇ + 𝑦 = 𝐾𝐴𝑠𝑖𝑛𝜔𝑡

General Solution:
−𝑡
𝑦(𝑡) = 𝐶𝑒 𝜏 + 𝐵(𝜔) sin(𝜔𝑡 + 𝜙(𝜔))

𝐾𝐴
Steady Response Amplitude: 𝐵(𝜔) = ;
√1+(𝜔𝜏)2
−1 (𝜔𝜏)
Phase Shift: 𝜙(𝜔) = − 𝑡𝑎𝑛

𝜙(𝜔)
Time delay 𝛽 = ; if 𝛽 < 0, time shift is a delay
𝜔

𝑜𝑢𝑡𝑝𝑢𝑡 𝑎𝑚𝑝. 𝐵(𝜔) 1

Magnitude ratio 𝑀(𝜔) = 𝑖𝑛𝑝𝑢𝑡 𝑎𝑚𝑝. = 𝐾𝐴 = √1+(𝜔𝜏)2
Second Order System 1 2𝜁
𝑦̈ + 𝑦̇ + 𝑦 = 𝐾𝐹(𝑡)
𝜔𝑛2 𝜔𝑛
𝜔𝑛 =
𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 ;
Characteristic Equation:
1 2 2𝜁
𝜆 +𝜔 𝜆+1=0
𝜔2
𝑛 𝑛
𝜁 = 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑟𝑎𝑡𝑖𝑜 Roots of the equation: 𝜆1,2 = −𝜁𝜔𝑛 ± 𝜔𝑛 √𝜁 2 − 1

Step Function Input

For 0 ≤ 𝜁 < 1, underdamped system
𝜁
𝑦(𝑡) = 𝐾𝐴 − 𝐾𝐴𝑒 −𝜁𝜔𝑛 𝑡 [ sin (𝜔𝑛 𝑡√1 − 𝜁 2 ) + 𝑐𝑜𝑠 (𝜔𝑛 𝑡√1 − 𝜁 2 )]
√1 − 𝜁 2

Ringing frequency 𝜔 = 𝜔 √1 − 𝜁 2
𝑑 𝑛

For 𝜁 = 1, critically damped system

𝑦(𝑡) = 𝐾𝐴 − 𝐾𝐴𝑒 −𝜔𝑛𝑡 (1 + 𝜔𝑛 𝑡);

For 𝜁 > 1, overdamped system

𝜁+√𝜁 2 −1 2 −1) 𝜔 𝜁−√𝜁 2 −1 (−𝜁−√𝜁 2 −1) 𝜔 𝑡
𝑦(𝑡) = 𝐾𝐴 − 𝐾𝐴 [ 𝑒 (−𝜁+√𝜁 𝑛𝑡 − 𝑒 𝑛 ]
2√𝜁 2 −1 2√𝜁 2 −1

Rise time Time to achieve 90% of step input

Time to achieve 90% of (𝐾𝐴 − 𝑦0 ) ;

Settling time Time to settle ±10% of steady value

Periodic Function Response:
Input 𝐾𝐴
𝑦(𝑡) = 𝑦ℎ + 𝑠𝑖𝑛[𝜔𝑡 + Φ(𝜔)]
2 2 2
𝐹(𝑡) = 𝐴𝑠𝑖𝑛𝜔𝑡 √[1 − ( 𝜔 ) ] + [2𝜁𝜔 ]
(Periodic input signal) 𝜔𝑛 𝜔𝑛
2𝜁𝜔
−1 𝜔𝑛
Phase shift: Φ(𝑤) = − tan ( 𝜔 2
)
𝑦ℎ : homogeneous 1−( )
𝜔𝑛
solution
𝐾𝐴
𝑦𝑠𝑡𝑒𝑎𝑑𝑦 (𝑡) = 𝑠𝑖𝑛[𝜔𝑡 + Φ(𝜔)]
Steady Response
2 2 2
√[1 − ( 𝜔 ) ] + [2𝜁𝜔 ]
𝜔𝑛 𝜔𝑛

= 𝐵(𝜔)𝑠𝑖𝑛[𝜔𝑡 + Φ(𝜔)]

𝐵(𝜔) 1
Magnitude ratio for a 𝑀(𝜔) = =
𝐾𝐴
2nd-order system 2 22
√[1 − ( 𝜔 ) ] + [2𝜁𝜔 ]
𝜔𝑛 𝜔𝑛

Resonance Frequency 𝜔𝑅 = 𝜔𝑛 √1 − 2ζ2 ; if 𝜁 > 0.707 , system won’t resonate

Transfer Functions 𝐺(𝑠); 𝑓𝑜𝑟 𝑠 = 𝑖𝜔, 𝐺(𝑖𝜔) = 𝑀(𝜔)𝑒 𝑖Φ(𝜔) = 𝑀(𝜔)∠Φ(𝜔)

Obtained by taking the
Laplace transform; 1
𝐺(𝑠) = ; 1st order transfer function
assuming 𝑦(0) = 0 𝜏𝑠 + 1

1
𝐺(𝑠) = 2) 𝑠2
; 2nd order transfer function
(1⁄𝜔𝑛 + (2𝜁 ⁄𝜔𝑛 )𝑠 + 1

Multiple-Function If 𝐹(𝑡) = 𝐴0 + ∑∞
𝑘=1 𝐴𝑘 𝑠𝑖𝑛𝜔𝑘 𝑡 ,
Inputs
∞
𝑦𝑠𝑡𝑒𝑎𝑑𝑦 (𝑡) = 𝐾𝐴0 + ∑ 𝐵(𝜔𝑘 )𝑠𝑖𝑛[𝜔𝑘 𝑡 + Φ(𝜔𝑘 )]
𝑘=1

where 𝐵(𝜔) = 𝐾𝐴𝑘 𝑀(𝜔𝑘 )

Operational Amplifiers:
Inverting Amplifier
𝑣 𝑅
closed loop gain is: 𝑣0 = − 𝑅2
𝐼 1

Miller Integrator
1 𝑡
𝑣0 (𝑡) = −𝑉𝑐 − 𝑅𝐶 ∫0 𝑣1 (𝜏) 𝑑𝜏

Laplace:
𝑉0 1
= −
𝑉1 𝑗𝜔𝐶𝑅
Differentiator:
𝑉0
= −𝑗𝜔𝐶𝑅
𝑉1

𝑑𝑣𝐼 (𝑡)
𝑣0 (𝑡) = −𝐶𝑅
𝑑𝑡

Weighted Summer 𝑣0 = −𝑖𝑅𝑓

𝑣
𝑖𝑛 = 𝑅𝑛
𝑛
𝑉 𝑉 𝑉
𝑣0 = −𝑅𝑓 (𝑅1 + 𝑅2 + ⋯ + 𝑅𝑛 )
1 2 𝑛

Non-Inverting Op Amp 𝑅
𝑣0 = 𝑣𝐼 (1 + 𝑅2)
1

𝑉 𝑅
𝐺𝑎𝑖𝑛 = 𝑉𝑜 = 𝑅2 + 1
𝐼 1

Difference Amplifier
𝑅 𝑅
𝑣0 = 𝑅2 (𝑣2 − 𝑣1 ) = 𝑅2 𝑣𝐼𝑑
1 1

𝑉𝑜 𝑅4 𝑅 𝑅
𝐺𝑎𝑖𝑛 = 𝐴𝑐𝑚 = = (𝑅 ) (1 − 𝑅2 𝑅3)
𝑉𝑖 4 +𝑅3 1 4

Instrumentation Amplifier
𝑅 2𝑅2
𝑉0 = 𝑅4 (1 + ) (𝑣2 − 𝑣1 )
3 𝑅1

𝑅 2𝑅2
Gain: 𝐴𝑑 ≡ 𝑅4 (1 + )
3 𝑅1
 Frequency Dependence of open loop gain: 𝑉𝑜𝑢𝑡
|𝐴| = 20 log (| |)
𝑉𝑖𝑛

𝑓𝑡 unity gain bandwidth

𝑓𝑏 3𝑑𝐵 break frequency
𝐴0
Finite gain: 𝐴(𝑗𝜔) = 1+𝑗𝜔/𝜔
𝑏

𝐴0 𝜔𝑏
For 𝜔 ≫ 10𝜔𝑏 : 𝐴 (𝑗𝜔) = 𝑗𝜔
Unity gain bandwidth at 0dB frequency 𝜔𝑡
𝜔 𝑓
|𝐴(𝑗𝜔 )| = 𝑡 = 𝑡 𝑑𝐵 = 20 𝑙𝑜𝑔(|𝐴|)
𝜔 𝑓

Loading Effects and Impedance Matching:

Impedance 𝑉(𝑠) 𝐸𝑓𝑓𝑜𝑟𝑡 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒
Z(s) = =𝐼(𝑠) 𝐹𝑙𝑜𝑤 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒
1
𝑍𝑅 (𝑠) = 𝑅, 𝑍𝐿 (𝑠) = 𝑠𝐿, 𝑍𝐶 (𝑠) =
𝑠𝐶
Impedance buffering For minimum loading error:
1
Voltage : 𝑍𝑂𝑈𝑇 (𝐴) ≤ 10 𝑍𝐼𝑁 (𝐵)
Current : 𝑍𝑂𝑈𝑇 (𝐴) ≥ 10 𝑍𝐼𝑁 (𝐵)

The null-balanced condition

𝑅1 𝑅3 − 𝑅2 𝑅4
𝑉𝑂 = 𝑉
(𝑅1 + 𝑅4 )(𝑅2 + 𝑅3 ) 𝑅

𝐼𝑓 𝑉𝑂 = 0 => 𝑅1 𝑅3 = 𝑅2 𝑅4

(watch out which configuration is used to use appropriate formula)

Filtering-Noise Characteristics-Grounding
Low Pass Filter (Capacitor): 1
𝐻(𝑠) = 𝑅𝐶
1
𝑠 + 𝑅𝐶
1
|𝐻(𝑗𝜔)| = 𝑅𝐶
√𝜔 2 + ( 1 )2
𝑅𝐶
 𝜃(𝑗𝜔) = −𝑡𝑎𝑛−1 (𝜔𝑅𝐶)

1
𝜔𝑐 =
𝑅𝐶

Low Pass Filter (Inductor): 𝑅

𝐻(𝑠) = 𝐿
𝑅
𝑠+𝐿

𝑅
|𝐻(𝑗𝜔)| = 𝐿
2
√𝜔 2 + (𝑅 )
𝐿

𝜔𝐿
𝜃(𝑗𝜔) = −𝑡𝑎𝑛−1 ( )
𝑅

𝑅
𝜔𝑐 =
𝐿

High Pass Filter: 𝑠

𝐻(𝑠) =
1
𝑠+
𝑅𝐶
𝜔
|𝐻(𝑗𝜔)| =
2
√𝜔 2 + ( 1 )
𝑅𝐶

𝜃(𝑗𝜔) = 90° − 𝑡𝑎𝑛−1 (𝜔𝑅𝐶)

1
𝜔𝑐 =
𝑅𝐶

Band Pass Filters: Center Frequency: ω0 = √𝜔𝑐1 𝜔𝑐2

𝑉0 (𝑠) (𝑅⁄𝐿)𝑠
=
𝑉𝑖 (𝑠) 𝑠 2 + (𝑅⁄ )𝑠 + (1⁄ )
𝐿 𝐿𝐶
𝑅
(𝐿 ) 𝜔
|𝐻(𝑗𝜔)| =
2 2
√[(1⁄ ) − 𝜔 2 ] + [𝜔(𝑅⁄ )]
𝐿𝐶 𝐿

(𝑅⁄𝐿 )𝜔
𝜃(𝑗𝜔) = 90° − 𝑡𝑎𝑛−1 [(1 ]
⁄𝐿𝐶 )−𝜔2
 𝜔0 = √1⁄𝐿𝐶

|𝐻(𝑗𝜔0 )| = 𝐻𝑚𝑎𝑥 = 1

𝑅 2
𝜔𝑐1 = − + √(𝑅⁄2𝐿) + (1⁄𝐿𝐶 )
2𝐿

𝑅 2
𝜔𝑐2 = + √(𝑅⁄2𝐿) + (1⁄𝐿𝐶 )
2𝐿

𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ: 𝛽 = 𝜔𝑐2 − 𝜔𝑐1 = 𝑅⁄𝐿

𝜔0 𝐿
𝑄𝑢𝑎𝑙𝑖𝑡𝑦 𝐹𝑎𝑐𝑡𝑜𝑟: 𝑄 = ⁄𝛽 = √ 2
𝐶𝑅

First-order LPF: 1
𝑍𝑓 𝑅2 || 𝑆𝐶
Transfer function: 𝐻(𝑠) = − = −
𝑍𝑖 𝑅1
𝜔𝑐
𝐻(𝑠) = −𝑘 ;
𝑠 + 𝜔𝑐
𝑅2 1
𝑤ℎ𝑒𝑟𝑒 𝑘 = ; 𝑎𝑛𝑑 𝜔𝑐 =
𝑅1 𝑅2 𝐶

Transfer function:
First-Order HPF:
𝑍𝑓 𝑅2
𝑇𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛: 𝐻(𝑠) = − =−
𝑍𝑖 1
𝑅1 + 𝑆𝐶
𝑠
𝐻(𝑠) = −𝑘 ;
𝑠 + 𝜔𝑐
𝑅2 1
𝑤ℎ𝑒𝑟𝑒 𝑘 = ; 𝑎𝑛𝑑 𝜔𝑐 =
𝑅1 𝑅1 𝐶

Inherent noise (Johnson): 𝑒𝑛,𝑅𝑀𝑆 = √4𝑘𝑅𝑇∆𝑓

𝑘 = 1.38 × 10−23 𝐽/𝐾 ; (𝑇 in Kelvin; 𝑓 is Hz)

 Inherent noise (Shot Noise): 𝑖𝑠𝑛 = √2𝑞𝑒 𝐼∆𝑓 = 5.7 × 10−4 √𝐼∆𝑓

In second equation, 𝐼 in 𝑝𝐴

Analog to Digital Conversion

Resolution 𝑉𝑚𝑎𝑥 − 𝑉𝑚𝑖𝑛
𝑉𝑄 =
2𝑘
𝑘 = number of bits
𝑣𝑖 −𝑉𝑅𝐿
𝑁 = round ( ) where 𝑁 = integer value for voltage 𝑣𝑖
𝑉𝑄
Where 𝑉𝑅𝐻 and 𝑉𝑅𝐿 are input voltage ranges (max and min)
∑𝑛 𝑏 2𝑖
𝑖
𝑣i = 𝑖=0 × 𝐹𝑆 + 𝑉𝑅𝐿
2𝑘+1
Quantization Error 𝑉𝐸 = 𝑉𝑥 − 𝑁𝑉𝑄 − 𝑉𝑅𝐿
Minimum number of bits allowable for error 𝑉𝑄 :
𝑉 −𝑉
log( 𝑅𝐻 𝑅𝐿 )
𝑉𝑄
𝑘≥ log(2)
Aliasing Error 𝑓
Distortion when: 𝑓 > 2𝑆 = 𝑓𝑁𝑦𝑞𝑢𝑖𝑠𝑡
Apparent aliasing frequency: 𝑓𝑎 = |𝑛𝑓𝑠 − 𝑓|
Where 𝑓𝑆 is the sampling frequency
To avoid aliasing: Nyquist Criterion
𝑓𝑆 > 2𝑓𝑚𝑎𝑥
Temperature Measurement: Relationship between R and T:

RTD equations: 𝑅 = 𝑅0 [1 + 𝛼(𝑇 − 𝑇0 ) + 𝛽(𝑇 − 𝑇0 )2 + ⋯ ]

Linear assumption:
𝑅 = 𝑅0 [1 + 𝛼(𝑇 − 𝑇0 )]
1 1
Thermistor: β( − )
𝑅 = 𝑅0 e T T0
Dissipation constant:
𝑃
𝛿 = 𝑇−𝑇
∞
where 𝑃 is power supplied to thermistor
Thermistors: 𝑇 is the temperature of thermistor
And 𝑇∞ is the ambient temperature

Metallic Strain Gauges Change in resistance:

𝑑𝑅 𝑑𝐿
= 𝐿 (1 + 2𝜈𝑝 + 𝜋1 𝐸𝑚 ) where 𝜋 is the piezoresistance
𝑅
coefficient
Gauge factor:
𝛿𝑅
𝐺𝐹 = 𝑅
𝜀

Single arm bridge:

Balancing condition: 𝐸𝑜 = 0 and all 𝑅’s equal
𝛿𝑅
𝛿𝐸𝑜 𝑅 𝐺𝐹𝜀 𝐺𝐹𝜀
= = ≈
𝐸𝑖 𝛿𝑅
4 + 2 ( 𝑅 ) 4 + 2𝐺𝐹𝜀 4
Multiple gauges:

𝑑𝐸𝑜 1
= (𝜀1 𝐺𝐹1 − 𝜀2 𝐺𝐹2 + 𝜀4 𝐺𝐹4 − 𝜀3 𝐺𝐹3 )
𝐸𝑖 4

Bridge constant:
𝑎𝑐𝑡𝑢𝑎𝑙 𝑏𝑟𝑖𝑑𝑔𝑒 𝑜𝑢𝑡𝑝𝑢𝑡
𝜅=
𝑚𝑎𝑥 𝑜𝑢𝑡𝑝𝑢𝑡 𝑠𝑖𝑛𝑔𝑙𝑒 𝑔𝑎𝑢𝑔𝑒