Kylowave Control Systems Platform

Model-Based Design Process:

A Linear Systems Course Perspective

By Julio Pimentel

University of Ottawa

November 9th, 2015

Objectives & Takeaways

Present a quick summary of model-based

design (MBD) process

Discuss why linear systems course is

important to MBD

Present a quick introduction to Kylowave

Teaching Platforms and K-CSP (Kylowave

Control Systems Teaching Platform)

Demonstration:

An embedded motor controller

A real-time virtual model with motor emulation

What is the problem?

Today, there is a growing trend for the automotive, aerospace, mechatronics and energy industry, among others, to blend: Mechanical

electro-mechanics

digital controllers

power electronics

communication and software

Coordinating this multidisciplinary and disparate engineering disciplines involved on a single development process presents a set of hard demanding challenges for technology companies to overcome [Aberdeen08:mbd].

Traditional Design Process

Also known as Text-Based Design Process

Modeling and simulation may be used but separated on each subsystem

Difficult to find problems early in the development cycle

Specifications

Design and Implementation

Integration and Test

Control Design

Control Algorithms

C/C++

Embedded Software

MCAD/MCAE

Mechanical Components

EDA

Electronic Components

Research Requirements

Model-Based Design Process

Modeling and Simulation is at the heart of the MBD methodology

Verification and validation is done continuously through all phases

Much easier to find bugs at the very beginning of the development cycle

C/C++

Design

Real-Time Testing

ContinuousVerification &

Validation

Requirements

Hardware

Environment Models

Timing and Control Logic

Electrical Mechanical

Algorithms

VHDL/Verilog CAD

Rapid PrototypingHardware-In-The-Loop

Test Environments

MBD V-diagram

ProjectInitiation

PreliminairyEngineering

Plans, specs and estimates

ConstructionProject

CloseoutOperations and

maintenance

Life cycle time line

Concept of operations

System Requirements

Project Architecture (High Level)

Component Level Design (Detailed )

Software CodingHardware Fabrication

UnitTesting

Subsystem Integration & Verification

System Integration & Verification

System Validation

Operations & Maintenance

Systems Engineering Management

Plan framework

Decision Gate

Example of companies providing

SW and HW tools for MBD Matlab from Mathworks

SytemVision from Mentor Graphics

Labview from National Instruments

Rockwell Collins

Esterel Technologies

Xilinx

Altera

Example of companies adopting

MBD as their design flow Boeing

Airbus

General Motors

FORD

Discussion Panel

Why is linear system modeling and

simulation important to model-based

design process?

Kylowave laboratory products

ONE platform – ALL your courses

K-CSP high-level architecture

K-CSP – Kylowave Control Systems Teaching

Platform

A new way to provide a teaching platform for

control systems courses

Disabled Lead-Lag #1

Lead-Lag#1

Disabled Lead-Lag #2

Lead-Lag#2

Disabled Controller

InternalDigital

Controller

ExternalController

PowerDrive

DC Motor

Reference

Feedbackopen-loop (1)

Disabled low-pass filter

FeedbackLow-pass

Filter

K L2 (S t Z2 + 1)(S t P2 + 1)

K e(S t M + 1)

K L1 (S t Z1 + 1)(S t P1 + 1)

K F(S + w F)

Controlleropen-loop (2)

SpeedSpeed error

PositionPosition error

Note: 1) Signals in red can be visualized on the GUI waveform window 2) Signals in green are internal signals and can not be visualized in the waveform window 3) The position gain maps +/- 2π radians to +/- 5V (V = gain * 0.79577 Angle) 4) Print a WARNING if one of the signals crosses/hit one of the saturation limits

K-CSP High-Level Architecture

K-CSP Kylowave Control Systems

Teaching Platform

K-CSP benefits to undergrad

students Useful in courses from introductory to

senior levels

Develop experience in analog and digital PID, PI-PD, Lead-Lag controllers

Develop better link between theoretic knowledge and practical hands-on experience

Provides for more practical hands-on skill development for what the engineer will face in industry

K-CSP Graphical User Interface

Used to configure the experiment

Procedure

1. Upload the embedded sketch to K-CSP

2. Connect K-CSP to the proper USB port

3. Setup the data file destination path

4. Setup the experiment runtime

5. Setup the sampling time to 50 ms

6. Setup the visualization time to 50 ms

7. Setup signal generator to PULSE GENERATOR (Initial Value = -3.0V, Amplitude = 6.0V, Period = 30s, Duty cycle = 50%, Delay = 0.0ms)

8. Assign Channel 0 and Channel 1 to “Raw speed” and “Raw speed error” respectively

9. Assign Graph3 and Graph 4 to “Reference in mV” and “Control Output in mV” respectively

10. Click on “START” to start the experiment or “STOP” to abort it

K-CSP GUI and Waveform Viewer

Configuration: Embedded controller PI-

PD position control

K-CSP GUI and Waveform Viewer

Configuration: Real-time simulator for

PID speed control

System level architecture of modern

speed controllers

PI Controller

PWM Modulator

Actuator Process

Sampling TimeClock

Generation

Feedback Sensor System

DC-DC

ConverterDC Motor

Optical Encoder

First-Order Low Pass

Filter

Shaft

Speed

Reference

Speed

Measured

Speed

Filtered

Speed

+

-

Error

w=dq/dtq

Km

1 + s tm

1

sKi

s

Kp

wref

VcontVerr

Kv

1 + s tv

Low-Pass Filter

DC Motor Model

PI Controller+

+

+

-

Vcont= (Kp + Ki/s) * Verr

Experiment Connection Diagram

K-ECS

Inv3Ph_A

Digital_7

Reserved

Digital_6Digital Control

Pin14

K-MCK

Inv3Ph_B

DG

ND

DG

ND

Three-Phase

Inverter pin2 DC_Mot-

OptEnc_X Digital Control

Pin9

OptEnc_A

OptEnc_BDigital Control pin29

pin30

ADCVin_0

An

alo

gIO

_1

Exte

rnal

sig

nal

gen

erat

or

(0 t

o +

5V

)

Inte

rnal

1kΩ

Po

ten

tio

met

er

An

alo

g

Ch

ann

el 1

Analog

ControllerController Output

Three-Phase

Inverter pin1

Digital Control

pin13

Digital Control

Pin7

PotV_0

Current Sensors Pin1

VSU

PP

LY

Vin = -5V to +5VVout = -15V to +15V

An

alo

g

Ch

ann

el 0

DACFltSym_1 DACFltSym_0

DACPWM_0DACPWM_0

DACPWM_1

Analog Connector

pin29

Analog Connector

pin30

VSU

PP

LYThree-Phase Inverter

Pin5 pin6DC Motor pin3

DC_Mot+DC Motor pin2

Digital Control

Pin10

Digital Control

Pin11

Digital Control

Pin30

DACPWM_1 Digital Control

Pin29

Current Sensors

Pin1RESERVED

Pin7

Analog Connector

pin33

Analog Connector

pin34

Analog Connector

pin24

MainFeedback

(Error Signal 1)

Three-Phase

Inverter pin4

Vin = -5V to +5V

AuxFeedbak (Error

Signal 0)

Error 0 (digital)

Error 1 (digital)

Analog Connector

Pin18

Analog Connector

Pin28

Controller output (sensored)

Analog Connector

pin6

AG

ND

Analog Connector

Pin6

AGND

SenseV_S0

Analog Connector

Pin5

AGND

INOUT

Analog PI controller with OpAmps

Analog Controller circuit

The Op. Amp. component is OPA4171 from

TI. All pin numbers are w.r.t. K-MCK analog

connector

+

-

VIN

R2=240kΩ

R1=220kΩC1=0.67µF

+15V

AGND

4

11

2

3

1

From Pin 33

-15V

-

+

R3=10kΩ

R4=10kΩ

To Pin 286

5

4

11

+15V

-15V

AGND

To Pin 5

To Pin 5

C2=1nF

7

Procedure to run the analog

controller experiment Load KcspAnalog sketch to K-CSP

Setup the PI controller power supply voltage to +/- 15V

Configure the GUI as explained in a previous slide

Setup run-time to 60 s and run the experiment

Use K-CSP to save ALL IMPORTANT experimental data as well as the waveforms

Repeat the procedure with power supply voltage set to +/- 5V. What is the difference? Explain.

What do we need to modify in the system to obtain a different closed-loop response? Is it easy to do it?

Procedure to run the digital

controller experiment Load KcspDigital sketch to K-CSP

Configure the GUI as explained in a previous slide

Setup run-time to 60 s and run the experiment

Use K-CSP to save ALL IMPORTANT experimental data as well as the waveforms

Do we need an external power supply? What do we need to modify in the system to obtain a different closed-loop response? Is it easy to do it?

In the sketch, set Ki to zero and increase Kp until the response approaches oscillatory behavior. Then, increase Ki gain to reduce the overshoot

What is the physical effect of Kp and Ki? Explain.

Feedback from the students

They liked to use K-CSP because it is compact, beautiful, flexible and allow them to experience realistic (non-ideal) behavior

The demo helped them to better link theory to practical experience by allowing them to: ”feel” the effects of the controller (Kp, Ki, controller counter-act

reaction and loop delay)

easily try other scenarios on their won that were not part of the lab manual but that triggered their curiosity

Add these additional experiments In open and closed –loop - Put pressure on the wheel to observe

the speed slow down, compensation by the controller and quickly release the pressure to observe the effect of the loop delay in overshooting the closed-loop response

Students suggested to fast finger tap the wheel to simulate load disturbance noise to understand the controller robustness to noise

Design an experiment to help them to visualize the effect of the loop delay in limiting the controller from achieving “zero error” but oscillating around it (tracking the reference)

Feedback from the professor

If the Lab has space limitation, the setup

with K-MCK on top of K-ECS plus

Kylowave’s expansion connector is a

clean, compact and beautiful solution

If the Lab can afford more space, a setup

with K-ECS besides K-MCK would

provide additional visual feedback to the

students by allowing them to see what is

happening inside the two products

Quiz questions asked during the

presentation

With Ki = 0, what is the steady-state error? Can it ever be zero? And if Ki ≠ 0?

What happens to the motor speed when we put pressure on the wheel? Why?

And what happens when we release pressure from the wheel? Why?

Name three realistic effects not usually accounted for in ideal simulations? And explain what is their effect on the controller response

In layman’s terms, what is the controller loop delay?

Q&A

Thank you

Literature:

[1]

[2]

[3]

Contact:

Kylowave Inc.

www.kylowave.com

jpiment@kylowave.com