reducing conducted transients in automotive windshield wiper motors
DESCRIPTION
Reducing Conducted Transients in Automotive Windshield Wiper Motors. Robert Langdorf, Shuvra Das, Mohan Krishnan University of Detroit Mercy. Project Objectives. Study the causes of conducted transients and develop a low-cost design solution to reduce them - PowerPoint PPT PresentationTRANSCRIPT
Reducing Conducted Transients in Automotive Windshield Wiper Motors
Robert Langdorf, Shuvra Das, Mohan Krishnan
University of Detroit Mercy
SAE 2006-01-02972
Project Objectives
Study the causes of conducted transients and develop a low-cost design solution to reduce them
Apply knowledge and skills obtained during other university coursework
Gain additional understanding of automotive motors and their electrical/mechanical interrelationships
SAE 2006-01-02973
Problem Description
When an electric motor is switched off, a large amount of energy (measured as a negative voltage) can be emitted to the main power net and can often be damaging to other devices.
For current design motor, transient emissions of >200V are possible. Customers desire no more than 100V (even less for some customers).
SAE 2006-01-02974
Design Considerations
Cost (there is already a very costly solution using varistors)
Packaging/Space Constraints Use of standard components
Effects on Other Electrical Requirements Other Conducted Emissions (radio
interference) Conducted Immunity Radiated Emissions
Effect on Motor Performance
SAE 2006-01-02975
Problem-Solving Approach
1. Create a working circuit model2. Perform some hand calculations on the
2nd-order system3. Perform PSPICE simulation4. Apply DOE principles to find optimum
solutions using PSPICE, Minitab & Excel5. Build and test physical samples to
validate results
SAE 2006-01-02976
Background InformationThe current design:
Capacitors
Inductors
Terminal Connections to Cover Assembly
Printed Wiring Board
SAE 2006-01-02977
The active components during a “switch-off” function are:– Two 0.47 F Capacitors– Two 5 H Inductor Coils– Motor (including inherent induction properties)
Background Information
SAE 2006-01-02978
Background InformationThe circuit used for simulation and analysis:
SAE 2006-01-02979
Assumptions
The high speed part of the circuit was neglected - there is no current flowing through it.
Relay was assumed to have a switching time of 0.5s. (Ford spec is <1s)
Motor armature inductance was measured at approximately 970 H.
Motor resistance, including armature and brushes was measured at approximately 0.5, but was assumed lower due to magnetic effects.
SAE 2006-01-029710
Assumptions, cont..
The rotational load on the motor (~10Nm) was accounted for with a 25 resistance from motor ground to source ground.
There are 2 different grounds in the system
Line resistance was assumed to be 1.25 between each side of the power source and the motor brush card terminals.
Note: these two assumptions were derived empirically by changing values until a solution was found that approximates the result of a typical experiment.
SAE 2006-01-029711
Comparison of solution to test resultProduction part test result:
SAE 2006-01-029712
Comparison of solution to test resultPSPICE Result:
Time
0s 0.2ms 0.4ms 0.6ms 0.8ms 1.0msV(Vout)
-400V
-200V
0V
200V
SAE 2006-01-029713
Comparison of solution to test result
The previous voltage responses exhibit: Voltage peaks of similar magnitude Similar dampening characteristics
SAE 2006-01-029714
Ground-to-ground issueFor a production motor, the motor ground to source ground was captured:
SAE 2006-01-029715
Ground-to-ground issueThe PSPICE model produces a similar result:
Time
0s 0.5ms 1.0ms 1.5ms 2.0ms- V(Vmg)
-80V
-40V
0V
40V
80V
SAE 2006-01-029716
Hand Calculations
Hand calculations were done using the same model as used in PSPICE.
The following calculation is done to find the approximate magnitude of the negative transient spike
Finding the decay takes considerably more calculation
SAE 2006-01-029717
Steady State SolutionCurrent through motor at t=0 is 4.737Avc1 = 7.588V, vc2 = 5.921V
SAE 2006-01-029718
Initial conditionsdi/dt = 9.184 A/s at t = 0+
SAE 2006-01-029719
2nd Order Differential Equation
The following equation can be derived:
dt
di
Li
LCdt
di
L
R
dt
id LR
RmR
25.2612
2
The following parameters can be calculated:
57.1782
L
R 9.465941
0 LC
SAE 2006-01-029720
2nd Order Differential Equation
The response is underdamped and the natural frequency can be expressed as:
6.46954220 d
The natural response can be expressed as:
tBtAeti ddt
n sincos)(
SAE 2006-01-029721
2nd Order Differential EquationThe forced response, which will be neglected for now, is
expressed as:
dt
di
Lti L
f
25.26)(
This is neglected because I do not have an expression for iL related to iR
The parameters A & B in the natural response equation are calculated by applying the initial conditions:
737.4A 018.0B
SAE 2006-01-029722
2nd Order Differential EquationThe expression for current with all of the
constants applied becomes:
ttedttiC
tv tt
RC 6.46594cos65.16.46594sin76.216)(1
)( 57.178
01
The expression for voltage across the capacitor C1 becomes:
tteti tR 6.46594sin018.06.46594cos737.4)( 57.178
SAE 2006-01-029723
2nd Order Differential Equation Solution
Plot of voltage across C1 versus time:Voltage across capacitor based on hand calculation
-214.2 V @ 32s
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012
Time (s)
Vo
ltag
e (V
)
SAE 2006-01-029724
Simulation ResultPSPICE Result:
V = -219.2 V @ t = 23.5 s
Time
0s 0.2ms 0.4ms 0.6ms 0.8ms 1.0msV(Vout)
-400V
-200V
0V
200V
SAE 2006-01-029725
Experimental Design
Comment on inductors, L1 & L2: Changing the values of the external inductors
has very minimal effect on the transient solution. Inductors in series simply add and these 5mH coils are negligible compared to the 970mH motor inductance.
These coils only will significantly effect the RFI filtering.
For the purpose of these experiments, the coils will be left unchanged.
SAE 2006-01-029726
Experimental Design
Using PSPICE & Minitab, a DOE was performed, modifying only the values of the capacitors, C1 & C2.
Each capacitor was simulated at 5 levels: 0.047F, 0.1F, 0.47F, 1F, 4.7F
SAE 2006-01-029727
Experimental Design
Using Minitab’s response surface feature, regression equations were formulated to help solve for the expected minimum and maximum peak voltages
SAE 2006-01-029728
Main Effect Plots
SAE 2006-01-029729
Main Effect Plots
SAE 2006-01-029730
Interaction Plots
SAE 2006-01-029731
Interaction Plots
SAE 2006-01-029732
Regression Equations
Minimum voltage peak:
Maximum voltage peak:
Note: C2 is insignificant in the min. voltage equation and the interaction C1xC2 is insignificant in both equations.
211 1.1061.6014.561 CCV
222
211 5.243.1459.1068.6051.435 CCCCV
SAE 2006-01-029733
Regression Solution
This yields as an optimum solution: C1 = 1.18F, C2 = 2.97 F Vmin = 0V, Vmax = 84.7V
When tested in PSPICE, the result is: Vmin = 145.5V, Vmax = 122.7V
?????This means there must be some other
relationship – try using the log of the capacitance values
SAE 2006-01-029734
Log Regression Equations
Minimum voltage peak:
Maximum voltage peak:
Note: C2 is insignificant in the min. voltage equation and the C2
2 is insignificant in both equations.
212
121 loglog3.45log4.174log5.51log0.1855.83 CCCCCV
2122
211 loglog8.18log1.19log9.171log0.1746.144 CCCCCV
SAE 2006-01-029735
Log Regression Solution
This yields as an optimum solution (with minimum peak-to-peak voltage): C1 = 3.81 F, C2 = 4.7 F
Vmin = -85.6V, Vmax = 51.8V
When tested in PSPICE, the result is: Vmin = -86.6V, Vmax = 65.5V
This is a much better model!!!
SAE 2006-01-029736
Log Regression Solution
Based on feedback from the supplier, it is not recommended to pursue use of 4.7F capacitors due to the high cost of materials. 3.3F capacitors are relatively less expensive.
Using 3.3F as a limit, the log regression model is re-optimized to yield: C1 = 3.3 F, C2 = 3.3 F Vmin = -90.4V, Vmax = 49.0V
PSPICE yields: Vmin = -93.8V, Vmax = 64.4V
SAE 2006-01-029737
Other Possible SolutionsSeveral other possible solutions exist to fix the transient
spike problem: Bridge capacitor (Y-type) Voltage suppressor Diode
These devices are placed in the circuit in this location:
Since these are much more capable of fixing the problem than only capacitors, the capacitance used in conjunction with these items can be reduced (thereby reducing cost).
SAE 2006-01-029738
Experimental Design #2
Another designed experiment was run to simulate the effects of the various solutions: No change Bridge capacitor (0.47F) Voltage Suppressor (Vishay TPSMA27A) Diode (D1N4184 from PSPICE library)
Each option was run at 3 levels of matched C1 & C2 (matched may be better to suppress RFI): 0.47F, 0.047F, 4.7nF
SAE 2006-01-029739
Dotplots of PSPICE Results
D1
N4
14
8
TP
SM
A2
7a
0.4
7
No
ne
0
-1000
-2000
C3
Min
Dotplots of Min by C3(group means are indicated by lines)
SAE 2006-01-029740
Dotplots of PSPICE Results
D1
N4
14
8
TP
SM
A2
7a
0.4
7
No
ne
2000
1000
0
C3
Ma
xDotplots of Max by C3
(group means are indicated by lines)
SAE 2006-01-029741
Analysis of Dotplots
It is evident from these plots that one of the recommended solutions may have a major impact.
Data means for each solution: None - min = -974.2, max = 829.8 0.47F Cap - min = -188.4, max = 175.6 TPSMA27A - min = -30.1, max =
1.0 D1N4148 - min = -3.7, max = 0.5
SAE 2006-01-029742
PSPICE Result for 0.47mF Bridge Capacitor
Time
0s 0.5ms 1.0ms 1.5ms 2.0msV(Vout)
-200V
-100V
0V
100V
200V
SAE 2006-01-029743
PSPICE Result for Voltage Suppressor
Time
0s 0.5ms 1.0ms 1.5ms 2.0msV(Vout)
-30V
-20V
-10V
0V
10V
SAE 2006-01-029744
PSPICE Result for Diode
Time
0s 0.5ms 1.0ms 1.5ms 2.0msV(Vout)
-4.0V
0V
4.0V
8.0V
SAE 2006-01-029745
Motor Build and Test Plan2 sets of parts have been built and tested:
Motors with the current capacitors (3x3 full factorial DOE)
C1 = 0.47F, 1F, 3.3F C2 = 0.47F, 1F, 3.3F
Motors with smaller capacitors and 2 of the voltage reduction solutions previously mentioned (3x2 full factorial):
C1 & C2 = 0.47F, 0.047F, 4.7nF C3 = 0.47F bridge capacitor, TPSMA30A Voltage
Suppressor
SAE 2006-01-029746
Comments on Build Plan
Cost is a serious consideration: 0.047F ~ $0.025 0.47F ~ $0.046 1F ~ $0.092 3.3F ~ $0.13 TPSMA30A ~ $0.16 Diode ~ too expensive to
seriously consider
SAE 2006-01-029747
Motor Test Plan
All motors were subjected to CE 410 (conducted emissions)
DOE principles are applied to analyze testing results
They will also be subjected to CE 420 (RFI emissions). However, timing did not allow such testing to be completed during the scope of this project
SAE 2006-01-029748
Test ResultsFollowing shows how Minitab outputs
analysis results:
General Linear Model: Min versus C1, C2Factor Type Levels Values C1 fixed 3 0.47 1.00 3.30C2 fixed 3 0.47 1.00 3.30Analysis of Variance for Min, using Adjusted SS for TestsSource DF Seq SS Adj SS Adj MS F PC1 2 1980.0 1980.0 990.0 2.21 0.138C2 2 16958.5 16958.5 8479.3 18.96 0.000C1*C2 4 3101.1 3101.1 775.3 1.73 0.187Error 18 8050.3 8050.3 447.2Total 26 30090.0
P = 0 translates to virtually 100% confidence that the factor is significant.
SAE 2006-01-029749
Test Results – Experiment #1
Assumptions of normality, independence of the testing order, constant variance and independence from other variables are deemed adequate based on analysis of residuals.
For the minimum peak voltage: The value of C1 is ~86% significant The value of C2 is 100% significant The interaction is ~81% significant
The effect plots (next slide) show that the optimum condition is when both capacitors are 3.3 F, similar to the simulation results.
SAE 2006-01-029750
Test Results – Experiment #1Optimum Settings at C1, C2 = 3.30
SAE 2006-01-029751
Test Results – Experiment #1
For the maximum peak voltage: The value of C1 is 100% significant The value of C2 is ~80% significant The interaction is ~99.4% significant
The effect plots (next slide) show that the optimum condition is when both capacitors are 3.3 F, similar to the simulation results.
SAE 2006-01-029752
Test Results – Experiment #1Optimum Settings at C1, C2 = 3.30
SAE 2006-01-029753
Test Results – Experiment #1
Using the optimum settings from this experiment (C1 & C2 are 3.3 F): The negative peak is about -110V The maximum peak is about +50V
Compared to the current capacitor design: Negative peak ≈ -200V Positive peak ≈ +85V This is a significant improvement
SAE 2006-01-029754
Test Results – Experiment #2
Assumptions of normality, independence of the testing order, constant variance and independence from other variables are deemed adequate based on analysis of residuals.
For the minimum peak voltage: The effect of the matched capacitors is not
statistically significant. The effect of the suppression device is 100% The interaction effect is ~93% significant.
Based on the effect plots (next slide), the ideal solution is the combination of 4.7 nF capacitors and the voltage suppressor.
SAE 2006-01-029755
Test Results – Experiment #2
Optimum Settings at C3 = TPSMA30A & C1C2 = 0.0047
SAE 2006-01-029756
Test Results – Experiment #2
For the maximum peak voltage: The effect of the matched capacitors is
~92% significant The effect of the voltage suppression
devices is 100% significant The interaction is ~97% significant
The effect plots (next slide), show that the optimum solution is the combination of 4.7 nF capacitors and the voltage suppressor, similar to the simulation results.
SAE 2006-01-029757
Test Results – Experiment #2Optimum Settings at C1C2 = 0.0047 & C3 = TPSMA30A
SAE 2006-01-029758
Test Results – Experiment #2
Based on analysis, the overall optimal solution includes: Matched Smaller Capacitors ~4.7nF Vishay Voltage Suppressor (TPSMA30A or similar)
Using these optimum settings: The negative peak is ~ -90V The maximum peak is ~ +15V.
Compared to the current capacitor design: Negative Peak ≈ -200V Positive peak ≈ +85V This is a significant improvement, even better than the
optimized design as determined in Experiment #1.
SAE 2006-01-029759
Conclusion
Cost comparison: Optimal solution
3 x $0.025 + $0.16 $0.235 per motor. Current cost
3 x $0.046 $0.138 per motor Varistor solution
$0.38 per motor
SAE 2006-01-029760
Conclusion
Questions