1 of 37

SYSTEM INTEGRATION

FOR

LASER ADDITIVE MANF.

MECH 527: Mechatronic System Design Project II

AASHKARAN DHILLON, M.ENG (MECHATRONICS DESIGN)

2 of 37

INTRODUCTION

Laser Additive Manufacturing:

Rapidly Growing Industry

Create complex metallic 3D prototypes

Laser Metal Deposition vs Selective Laser Melting

Purpose:

Use Laser Metal Deposition Method
Components Available:

4-axis Motion Platform
Laser Generator
Laser Deposition Head
Powder Feeder
Chiller

Figure 1: Laser Metal Deposition System [1].

[1] Additive manufacturing. LASERLINE. (n.d.). https://www.laserline.com/en-int/laser-additive-manufacturing/.

2

3 of 37

INTRODUCTION

Figure 2: Layout For Additive Manufacturing [2].

[2] Picture Credit to A.O.

3

4 of 37

INTRODUCTION

Tasks Accomplished:

Assembly and integration of sub-systems

Setup, Testing and Interfacing using schematics�layout drawings & manuals

Troubleshooting

Reliability of Hardware
Z-axis Leadscrew Fix

Motion Control

System Identification
Controller Design
Simulation and Experimental Validation

Misc.

Control Desk GUI
MATLAB Code, Documentation

Figure 3: ATS Machine in Lab with Laser Deposition Head.

4

5 of 37

PRESENTATION OF WORK

System Integration for Additive Manufacturing: Laser Generator

Laser Power:

500W Class 4 Laser in Deposition Head
Proposed method for interfacing Laser Generator to control system

Figure 4: Laser Generator 24-Pin Connector Interface [3].

[3] YLR-Series User Guide. (2015). IPG Photonics

5

6 of 37

PRESENTATION OF WORK

System Integration for Additive Manufacturing: Laser Generator

Interfacing:

Pins 1-4 Safety Interlock
Pins 10-11 Remote Start
Pins 12 & 14 Analog Control Input
Pins 18-24 Monitoring Status, Pin 17 Guide Control

GUI w/ Ethernet
LCD on Machine

Connections:

24 Pin Connector
18AWG shielded cables

To Breakout Board
To dSPACE DAC
To Safety Interlocks

Figure 4: Laser Generator 24-Pin Connector Interface [3].

[3] YLR-Series User Guide. (2015). IPG Photonics

6

7 of 37

PRESENTATION OF WORK

System Integration for Additive Manufacturing: Laser Generator

Safety:

7-Pin Connector to pair ATS Machine & Laser Generator E-Stops

Pair Pins1-2 with Soft-Estop on Front Panel
Pair Pins 3-4 with Hard-Estop on ATS Machine

Use Logic IO Box Relays for Connection

ATS E-stop triggers relays to disengage Motor Power

Figure 5: Laser Generator 7-Pin Connector [3].

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 6: Relays in Logic IO Box for Motor Power.

7

8 of 37

PRESENTATION OF WORK

System Integration for Additive Manufacturing: Laser Generator

Interfacing with Controller Software:

Pins 10 Remote Start

Replace IO11 wire for Dispenser on

Terminal Block w/ Laser Remote Start wire

Pins 12 Analog Control of Laser Power

Use DAC channel on dSpace board w/ BNC for

1-10V voltage input to Laser Generator

rundirectraj.m for printing desired Trajectory

alltrajs.mat: X,Y,Z,Dispenser On/Off, Duty Cycle

Dispenser On/Off => Laser Remote Start
Dispenser Duty Cycle => Laser Power

createtrajs.m: generates trajectory from gCode

Figure 7: Laser Generator 24-Pin Connector and Breakout Board.

8

Coming back to Laser Generator Control. As mentioned, for the 24-pin connector, we utilize only safety interlock pins, remote start and analog input pins. Pin 10 for remote start will replace the existing wire for dispenser switch in the Terminal Block on Breakout board. Since the dispenser is not being used anymore. The terminal block has IO1-IO11 pin connections that are part of Digital IO signals that are used for communication between machine and dSpace using DB50 cable. The pin IO11 is already setup in Simulink Model for turning dispenser on/off and can be relabeled for Laser switching.��Pin 12 is used for 1-10V analog control of Laser Power. 1V being no power and 10V being 100% power. This analog voltage signal will be delivered using a BNC cable from the DAC channel on dSpace. BNC cable will be spliced open with the power wire going into pin 12 and ground into pin 14. And to transmit the analog voltage signal from chosen DAC channel, the DS1103DAC_C1 block will be need to be initialized like the other DAC channels being used to send analog signals for motors. ��On the software side, rundirecttraj.m file is used to execute a trajectory. It reads alltrajs.mat file with 5 columns for each layer that are X,Y,Z points, Dispenser on/Off signal and duty cycle for Dispenser. The Dispenser On/Off will be replaced by Laser Switching and Duty Cycle will correspond to Laser Power Output from the DAC channel. The rundirectraj file allows to override PWM value to constant value, which can be used.�These matrices for trajectories are generated using matlab file createtrajs.m that converts gcode for any toolpath shape into a quintic spline trajectory w/ optimized federate.

9 of 37

PRESENTATION OF WORK

Hardware Troubleshooting: Mechanical

Problem: Z-Axis Lead Screw

Unresponsive to Input Commands
LS/ECB fault on Amplifier Card => Current Surge
Lead Screw Jammed

Limit Switch Didn’t Trigger => Rammed into bearing mount
Concerns: lead screw bending & nut damage�

Diagnosing:

Took out the motion stage for inspection
Disassembled eBrake and carriage
Identified the source of problem was bearing�misalignment.

Figure 11: Amplifier Card Z-axis Fault

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 12: Lead Screw Z-axis Motion Stage

9

Over the course of the project, there were roadblocks faced due to the electrical and mechanical problems that hampered the functionality of the machine. These issues had to be diagnosed, identified and addressed at the root cause level to make the machine reliable to work with and increase its lifecycle. Here the mechanical troubleshooting is discussed first.

While doing the preliminary testing on each axis to move the motion stage from Point A to Point B, it was observed that the Z-axis wasn’t responding to the input current commands. First, I thought that the motor might have malfunctioned, but upon checking the amplifier card, a low-speed electronic circuit breaker fault (LS/ECB) was observed, indicating that motor was drawing excessive current due to mechanical loading and shorting out

It turned out that Z-axis leadscrew was jammed and needed to be freed. Assumably, a previous user drove down the Z axis leadscrew and smashed the carriage to the very bottom, causing it to get stuck and jammed. This happened despite the integrated safety mechanism in form of limit switches because the flag on the carriage was loose and probably didn’t trigger the bottom limit switch, allowing the Z-axis motor to keep driving down and ram the carriage on the bottom bearing mount.

To diagnose, the Z axis motion stage was taken apart to analyze the extent of the damage, the eBrake and carriage were disassembled to expose the leadscrew and nut and it was identified that the problem was from the bearing misalignment.

10 of 37

PRESENTATION OF WORK

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 12: Lead Screw Z-axis Motion Stage

10

11 of 37

PRESENTATION OF WORK

Hardware Troubleshooting: Electrical

Problem: Signal Integrity

DB50 Cable for I/O signals b/w Controller and Machine

dSpace Digital IO ⬄ Breakout board�Interfaces controller with machine
Random disruption of signals

Breakout Board for routing IO signals

Encoder Signal Flickering => Noise
Y&Z Axis Encoder RS232 pins broke�

Diagnosing:

Inspected Digital IO connections into DB50 connector
Identified Broken Encoder Pins on Breakout board
Worn out/ Poorly soldered joints the underlying issue

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 13: DB50 Cable for Digital IO signals.

Figure 14: Broken Encoder RS232 Connection Pins

11

For hardware troubleshooting of Electrical parts, work was done towards ensuring the signal integrity and increasing the hardware reliability of the machine.

DB50 cable connects the dSpace board and the breakout board to communicate the digital IO signals between the controller and the machine. Working with the system, digital IO signals would randomly get disrupted and DB50 connector would need to be routinely adjusted at the breakout board side to restore signals. This was an inconvenience that needed to be fixed for reliable operation.

Next, few RS232 pins on breakout board for Y and Z encoder broke while relocating the Breakout board and dSpace board under ATS machine. Even before that., there was a persistent issue where the encoder signals to flickered on the dSpace board, which at times made the encoder readings questionable due to the signal noise.

For diagnosing, the Digital IO wiring into DB50 connector was inspected using multimeter. The breakout board had broken pins and it was taken to technician for repair. For both, the underlying problem was worn out or poorly soldered joints that hampered the signal communication with dSpace.

12 of 37

PRESENTATION OF WORK

Hardware Troubleshooting: Electrical

Solution:

DB 50 Cable

Resoldered loose wires into DB50 connector
Replaced DB50 Cable with Spare Part
Crimped +5 VCC & GND connections

5V +VCC powers Logic IO box (Dispenser removed)
GND shared with Amplifier & Logic IO GND

Breakout Board

Got PCB investigated by Technician

Multiple Loose Joints => Soldered
New RS232 male connectors added

Conclusion:

Hardware made reliable to work with
Signal Integrity Preserved

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 15: Old DB50 Cable with twisted VCC & GND connections

Figure 16: New DB50 Cable Connection Setup

12

How these problems were solved are as following.

The loose wires were resoldered into the DB50 connector and that fixed the I/O signal fluctuation problem. Later this DB50 cable was replaced with a spare one that was sturdier with properly soldered connections and heat shrink on each wire. Through the DB50 connector, as you can see, +5VCC and GND is shared via the red and black wires which are used to power the Logic IO box and the ground acts as a common ground between dSpace, Amplifier box,and Logic IO box. Further, the +5V and GND connections were crimped for stable connectivity. These were before twisted wiring connections that would abruptly loosen and cause IO signals to toggle.

For the breakout board, the technician analyzed PCB and identified multiple loose joins all across that were resoldered. The broken RS232 connectors for encoders were removed and replaced by new ones.

In conclusion, through the diagnosing and repair, the electronics hardware was made reliable to work with and the signal integrity was restored.

13 of 37

PRESENTATION OF WORK

Hardware Troubleshooting: Electrical

Problem: Amplifier Box

Unresponsive Reset for Amplifier Cards
Power Loss to Encoders, Motors, Switches�

Diagnosing:

Amplifier Box

Checked IC connections in PCB
Tightened cable connectors behind Amplifier Box
Minimize Compression Force from Cable Slack
Ensure Ground Connections

Logic IO Box

Examine Fuses of OPTO22 Modules
Tracked down wiring of relays and ensured�match with Logic IO Box Schematic
Document information on function of relays

Useful in planning Laser E-stop incorporation

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 17: Logic IO Cabinet

13

There was an issue with the amplifier box where the reset push button would not reset the amplifier cards, thus stalling any flow of current from amplifier cards to the motors. Also, an instance occurred where suddenly the power to the motor, encoders and limit switches was lost. Troubleshooting was performed to locate the source of these problems.

The unresponsive reset button problem was first addressed by cross checking all the IC chip connections inside of the amplifier box, tightening all the cable connections to the amplifiers, the slack on all the cables behind the amplifier cabinet was minimized to prevent compression force from cables onto the connector ports. Next, the joint where all the ground connections merge was tightened to make sure all the ground connections overlapped each other. Following these steps fixed the reset issue, and the amplifier now resets dependably. It’s safe to say that the problem emerges from loose connections. Glenn, the technician, suggested that there could be oxidation as the machine is old, which makes connections act unpredictably.

For the power loss to the limit switches, motors, encoders, it was first assumed that the switching in the Logic IO box was the source of problem. So, I checked all the Opto22 module fuses, I tracked down the wiring of the relays using multimeter and made sure the connections were in sync with the Logic IO box layout schematic. Even though this didn’t correct the issue, I was able to obtain significant info on the role of Logic IO box in machine operation, which was useful in documenting information and planning expansion like incorporating Laser E-stops to the relay to pair with ATS machine E-stops.

14 of 37

PRESENTATION OF WORK

Hardware Troubleshooting: Electrical

Solution:

Studied Amplifier Box Schematic => Power Routed Through Orange Terminal
Amplifier Reset Cleared as discussed

�

Conclusion:

Amplifier Box Functions Reliably
Common Sources of Problem Identified and Fixed

[3] YLR-Series User Guide. (2015). IPG Photonics

Figure 19: Amplifier Cabinet Inside

Figure 18: Amplifier Cabinet Behind

14

15 of 37

PRESENTATION OF WORK

Miscellaneous Troubleshooting:

Fuse Upsized from 5A 240V to 7A 110V
Measured Voltage Losses from Controller Output to Amplifier Input

Z-Axis 1V Input Command => 0.32V at Amplifier => 0.68V Offset
Compensated in Software

Software Changes

Discrepancies Corrected in Simulink Model & MATLAB Code

[3] YLR-Series User Guide. (2015). IPG Photonics

15

16 of 37

PRESENTATION OF WORK

Controller Design

Goal: Redesign Controller and Improve Upon� Stability & Robustness w/ added loading

Procedure:

System Identification of Plant TF

X-Axis only as Z-axis compensates weight�w/ pneumatic cylinders
Experimentally done

Controller Design
Simulate Performance
Experimental Validation

Figure 20: 3-Axis Motion Control with Weights Added to mimic Deposition Head

16

17 of 37

PRESENTATION OF WORK

Controller Design: System Identification

Procedure: Freq Domain Data Acquisition

Open Loop Frequency Sweep of X-axis w/ Loading

To Acquire Magnitude and Phase Output Data
Sinusoidal Input:1A amplitude at 1-100Hz for 100 sample points

Air Bearings to drive Motion Stage with Linear Motor

Non-linearities, disturbance from friction => small.
Can approximate plant TF w/ Matlab’s System Identification Toolbox

Results:

Figure 21: Frequency Response for X-Axis w/ Weight

17

18 of 37

PRESENTATION OF WORK

Controller Design: System Identification

Procedure: Estimating Transfer Function

Magnitude & Phase Data Fed into System Identification Toolbox

Dataset1 used for Estimation sliced to 1.5hz-100hz to remove noisy measurements
Dataset2 used for Validation

Predict 2^nd Order System

2-Poles 0-Zeros
Prevent Overfitting

Figure 21: Frequency Response for X-Axis w/ Weight

Figure 22: Screen capture from System Identification Toolbox

18

19 of 37

PRESENTATION OF WORK

19

20 of 37

PRESENTATION OF WORK

Figure 23: Bode Plot for X-Axis w/ Weight

20

21 of 37

PRESENTATION OF WORK

Controller Design: PI Controller + Lead Compensator

Results:

Increased B.W / Gain Crossover Freq ~30Hz

Improved Transient Response
Stability

Increased Magnitude at Low Freq

Better Steady State Accuracy, Sensitivity, �Disturbance Rejection.

Gain Margin (inf) and Phase Margin (~45deg)

Desirable
Robustness & Stability

Figure 24: Compensated vs Uncompensated Plant FRF

21

22 of 37

PRESENTATION OF WORK

Controller Design: Simulated Controller Comparison

Step Response:

Figure 25 a): 1mm Step Response Compared. OS 20% vs <10%

Figure 25 b): 3mm Step Response Compared. OS 50% vs 15%

22

23 of 37

PRESENTATION OF WORK

Controller Design: Simulated Controller Comparison

Step Response:

Observation:

Designed Controller produces a fraction of overshoot
Older Controller unstable at Larger Inputs, New Controller Stable
Slower Transient Response, a trade off for lower overshoot & stability.

Preferred behaviour for Laser AM

Figure 25 c): 5mm Step Response Compared

23

24 of 37

PRESENTATION OF WORK

Controller Design: Simulated Controller Comparison

Ramp Response Steady State Error:

Figure 26 a): 10mm Travel @ 10mm/s Ramp SS Error

Figure 26 b): 10mm Travel @ 50mm/s Ramp SS Error

24

25 of 37

PRESENTATION OF WORK

Controller Design: Simulated Controller Comparison

Ramp Response Steady State Error:

�

Observation:

Both Old & New Controllers Eliminate Steady State Error
For Travel at demanding speed of 100mm/s:

Older Controller takes double time to creep up & Produces multiple Overshoots

Simulation promise improved robustness and stability. Important for working with Class 4 Laser!

Figure 26 c): 5mm Travel @ 100mm/s Ramp SS Error

25

26 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Step Response: 1mm Step Input Old (L) vs New (R)

Figure 26 a): 1mm Step Response Compared. OS: 20% vs <10%

26

27 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Step Response: 2mm Step Input Old (L) vs New (R)

Figure 26 b): 2mm Step Response Compared. 30% OS vs 10%

27

28 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Step Response: 3mm Step Input Old (L) vs New (R)

Figure 26 c): 3mm Step Response Compared. OS 50% vs 15%

28

29 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Step Response: 10mm Step Input Old (L) vs New (R)

Figure 26 e): 10mm Step Response Compared

29

30 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Ramp Response: 120mm Ramp Input at 10mm/s Old (L) vs New (R)

Figure 27 a): 10mm/s Ramp Response Compared

30

31 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Ramp Response: 120mm Ramp Input at 50mm/s Old (L) vs New (R)

Figure 27 b): 50mm/s Ramp Response Compared

31

32 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Ramp Response: 240mm Ramp Input at 100mm/s Old (L) vs New (R)

Figure 27 c): 100mm/s Ramp Response Compared

32

33 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Observation:

Transient Response from Experiments same as Simulated Plots

Validates System Identification Process => Estimated Transfer Function G(s) accurate

New Controller drastically reduces Overshoot & Stable at Large Step Inputs
New Controller performance for Ramp Input

More aggressively responds at faster speed, more stable, near 0mm tracking error.

33

34 of 37

PRESENTATION OF WORK

Controller Design: Experimental Validation

Goal: Test Performance for Custom Geometry

Procedure:

Trajectory quintic B-spline toolpath w/ optimized feederate

Output from createtrajs.m

rundirectrajs.m file executes Trajectory

Uses alltraj.mat & layertime.mat

Result:

Max Tracking Error ~0.27mm

Occurs when tool sharply enters oval shape

Average Tracking Error <0.05mm

Figure 28: Toolpath Tracking

34

35 of 37

CONCLUSION

Assembly, setup & testing of systems: Laser Generator, Powder Feeder, Deposition Head

Electrical Schematics, Layout Drawings & User Manuals

Proposed method for interfacing Laser Generator to Control System
Increased Hardware reliability of ATS machine by Troubleshooting

Z-Axis repair, Amplifier Box, Logic IO, Breakout Board Connections, Software debugging

Successful Controller Design

System Identification using acquired Frequency Domain Data
Controller Design: PI Controller + Loop Shaping for 30Hz BW 45deg PM
Simulation & Experimental Validation
System Robust & Stable

35

36 of 37

END OF PRESENTATION

THANK YOU!

QUESTIONS?

36

1 of 37

2 of 37

3 of 37

4 of 37

5 of 37

6 of 37

7 of 37

8 of 37

9 of 37

10 of 37

11 of 37

12 of 37

13 of 37

14 of 37

15 of 37

16 of 37

17 of 37

18 of 37

19 of 37

20 of 37

21 of 37

22 of 37

23 of 37

24 of 37

25 of 37

26 of 37

27 of 37

28 of 37

29 of 37

30 of 37

31 of 37

32 of 37

33 of 37

34 of 37

35 of 37

36 of 37

37 of 37