UNCOOLED IR DETECTOR TEMPERATURE CONTROL Performed by : Shimon Amir Yogev Ben-Simon Instructor : Arie Nakhmani 1 TECHNION - ISRAEL INSTITUTE OF TECHNOLOGY.

UNCOOLED IR DETECTORTEMPERATURE CONTROL

Performed by : Shimon Amir Yogev Ben-Simon

Instructor : Arie Nakhmani

1

TECHNION - ISRAEL INSTITUTE OF TECHNOLOGY

Department of Electrical Engineering Control And Robotics Laboratory

Control And Robotics Laboratory

Control And Robotics Laboratory 2

OUTLINE PRESENTATION

• PROBLEM• SYSTEM STRUCTURE• PROJECT GOALS & OPTIONAL SOLUTIONS• CHOSEN SOLUTION• RESULTS• FINDINGS • SUMMARY• CONCLUSIONS

Control And Robotics Laboratory 3

PROBLEM

• Detector “BIRD 384” is IR detector type uncooled .

• This kind of detector requires stable substrate temperature .

• The requirement for temperature deviation is +/- 10mK .

• During the characterization after manufacture  the detectors pass series of tests ,which include settling to several set points. This procedure takes a lot of time , therefore we need to reduce the detector settling time.

• The initial tests showed that there is variance between the time response of the detectors .

Control And Robotics Laboratory 4

SYSTEM STRUCTURE

Detector Controller & Acquisition

Mechanical Stand

active element: TEC

active element: TEC

Measurement : RTD PT100

Measurement : RTD PT100

Microcontroller embedded on FPGA

Microcontroller embedded on FPGA

24 bitADC 24 bit ADC

20 bitDAC 20 bit DAC

linear power amplifier

linear power amplifier

Heat sinkHeat sink

Thermal Coupling Thermal Coupling

Uncooled IR Detector

Control And Robotics Laboratory 5

TEC

AluminaRTD

Thermoelectric Cooler (TEC)

Control And Robotics Laboratory 6

PT100 RTD Platinum RTDs ( Resistance Temperature Detectors ) are recognized as the most reliable standard available for temperature measurement . The PT100 RTD is described by the following generic equation , which makes a obvious nonlinear relationship between temperature and resistance:

Since the B and C coefficients are relatively small, the resistance changes almost linearly with the temperature.

7Control And Robotics Laboratory

2 3 0 0T 0

2 0 0T 0

00

3 0 1

7 0 2

12 0 4

R R [ 1+AT+BT +C(T-100)T ] ( 200 0 )

R R [ 1+AT+BT ] (0 850 )

: R 100 ( 0 )

3.9083 10

5.775 10

4.183 10

C T C

C T C

where at C

A C

B C

C C

T 0R R [ 1+AT]

Control And Robotics Laboratory 8

Mechanical Stand

Gap Filler For Thermal coupling

Electronic card & Detector

Control loopPT

-100

I=1m

A

G=30

ADC

REF

FPG

API

DR

EGU

LATO

R

DACTECShunt

TEC CURR. MEAS.

SUBSTRATE TEMP

PT-1

00

I=1m

A

G=30

ADC

OFF

SET,

GAI

NM

EAS.

Vsubstr.

Vcase

ADC_GAIN

ADC_OFF

Vref

Vref

Vref

Vref

Linear AMP Linear AMP

Diff. AMP

A/D

9Control And Robotics Laboratory

PROJECT GOALS & OPTIONAL SOLUTIONS

The project goals :

1. To reduce settling time .

2. To improve the robustness .

3. Identify the reasons for the variance between the settling time of detectors .

Optional solutions :

1. To improve the control algorithm .

2. To improve the physical parameters of the system/detector such as :

– Thermal impedance between detector package and the heat sink.

– Heat sink.

– “Inside of the package”, e.g. internal PT100 sensor location.

10Control And Robotics Laboratory

Control And Robotics Laboratory 11

CHOSEN SOLUTION

• System modeling .

• Algorithm : PID + switch mode - full power until getting to set-point area ,initializing the integral with a desired value , and then continue with closed loop to convention.

• PID tuning using software tool and manual adjustment.

• Mechanics : improve the thermal coupling to the heat sink.

System Modeling• The chosen model is second order system : one pole for the TEC and

another one for the temperature sensor.

• Even though we know that there is delay in thermal system , in our system the delay can be neglected.

• Initial system modeling using step response of the open loop system , and set the system parameters using software tool .

12Control And Robotics Laboratory

0 200 400 600 800 1000 1200 14000

10

20

30

40

50

60

70

T

empe

ratu

re [

oC

]

Time [ sec ]

Plant Simulink Model

13Control And Robotics Laboratory

td

k

b

a

Transfer Fcn 1

t1.s+1

k1(s)

Transfer Fcn

k2

t2.s+1Step

Scope 2

Scope 1

Scope

k1

Output Constraint

k2

FromWorkspace

simin

t2

t1

Abs

|u|

Modeling Results

The results for the second order estimation :

• DET 1(OK) :

• DET 2 (SLOW):

HIGH VARIANCE

26.15.165.3 21 KSECSEC

14Control And Robotics Laboratory

31.1234 21 KSECSEC

PID tuning using sisotool

Control And Robotics Laboratory 15

10-2

10-1

100

101

-180

-90

0

Frequency (rad/sec)

-100

-50

0

50Bode Editor for Closed Loop 1 (CL1)

10-3

10-2

10-1

100

101

-180

-135

-90

P.M.: 45.2 degFreq: 0.231 rad/sec

Frequency (rad/sec)

-60

-40

-20

0

20

40

60

G.M.: InfFreq: InfStable loop

Open-Loop Bode Editor for Open Loop 1 (OL1)

-0.4 -0.3 -0.2 -0.1 0-0.4

-0.2

0

0.2

0.4Root Locus Editor for Open Loop 1 (OL1)

Step Response

Time (sec)

Ampl

itude

0 5 10 15 20 25 30 35 400

0.2

0.4

0.6

0.8

1

1.2

1.4

System: Closed Loop r to yI/O: r to yTime (sec): 7.17Amplitude: 0.922

System: Closed Loop r to yI/O: r to ySettling Time (sec): 29.4

RESULTS

16Control And Robotics Laboratory

The results of the PID tuning using sisotool :• P = 15• I = 0.008

and after manual adjustment :• P = 7• I = 0.02

SETTLING TIME RESULTS Measurements were taken from 8 detectors that tested in 3

configurations :

1. Original system - “Old”.

2. Improved controller and manually PID tuning – “New”

3. Improved controller , heat sink and thermal impedance - “New +Pad”

17Control And Robotics Laboratory

IMPROVEMENTS

 OldNewNew + padNew vs OldNew vs OldNew + pad

vs OldNew + pad

vs Old

 Avg time

[sec]Avg time

[sec]Avg time

[sec]improvement

[sec]improvement

]%[improvement

[sec]improvement

]%[

25 6082.56047.222.527.335.342.8

60 25 101.867.237.534.73464.363.2

25 3562.536.5352641.627.544

35 2576.348.734.227.736.242.255.2

25 1591.761.334.830.333.156.862

15 2576.735.336.241.353.940.552.8

25 573.75333.820.728.139.854.1

5 2564.232.233.73249.930.547.5

18Control And Robotics Laboratory

Comparison Between Detector #1 Results For The Jump 25 60

Det 1 “Old ”

1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950

58.5

59

59.5

60

60.5

61

61.5

62

62.5

63 Det 1 “New+Pad ”

1885 1890 1895 1900 1905 1910 1915 1920 1925 1930 1935

59.98

59.985

59.99

59.995

60

60.005

60.01

60.015

19

Det 1 “New ”

50 60 70 80 90 100 110 120 13055

56

57

58

59

60

61

108 110 112 114 116 118 120

59.99

59.995

60

60.005

60.01

30 40 50 60 70 80 90 100 110 120 130

52

54

56

58

60

62

75 80 85 90 95 100 105

59.992

59.994

59.996

59.998

60

60.002

60.004

60.006

fig. 2 : zoom on the relevant area in fig. 1 fig. 4 : zoom on the relevant area in fig. 3 fig. 6 : zoom on the relevant area in fig. 5

fig. 1 : Det 1 “old” settling time to +/-10mK fig. 3 : Det 1 “New” settling time to +/-10mK fig. 5 : Det 1 “New+Pad” settling time to +/-10mK

0( 10 ) 35[sec]settlingt m K 0( 10 ) 55[sec]settlingt m K 0( 10 ) 80[sec]settlingt m K

MEASUREMENT QUALITY• Temperature sense readout circuits accuracy and quality is very

important.

• Testing the accuracy and noise of the temperature sense circuits was done by replacing the RTD with a regular resistor of 118.2 ohm [~ 46.5 deg’].

• Readout circuits noise is ~10mK pk-pk.

• Impact of readout circuits temperature drift is less then 20mK for 10 deg’ change of surroundings.

20Control And Robotics Laboratory90 95 100 105 110 115 120 125 130

46.61

46.615

46.62

46.625

46.63

~10mK

FINDINGS • There are two main kinds of temperature stability

problems:1. “Detector slow response”

2. “Detector can’t cool down”

• Slow response : the problem is mainly TEC-alumina and alumina-RTD contacts.

• Not cooling : the problem is mainly TEC-package contact.

21Control And Robotics Laboratory

Control And Robotics Laboratory 22

RTD ASSEMBLY

23Control And Robotics Laboratory

RTD SURFACEALUMINA-TECTEC - PACKAGE

GLUINGS

MECHANICS

• The current design includes a gap filler pad as thermal coupling to the heat sink .

• It has a thermal conductivity of 2 . and with 2mm thickness it’s thermal impedance is 1.3 .

• Compression achieved using material elasticity and socket friction .

• Possible improvement can be achieved by using a thermal pad with a better thermal impedance and a copper heat sink. – will require mechanical adjustments.

• Tested – 0.3 mm thermal pad with aluminum film to avoid pull-out (pad stick to the detector’s package) effect, thermal conductivity of 17 and thermal resistance of 0.06 , copper heat sink .

mK

W

2C in

W

mK

W

2C in

W

24Control And Robotics Laboratory

SUMMARY

Control And Robotics Laboratory 25

• Settling time improvement of 30%-40% has been achieved.

• Reasons for variance has been identified .

• Manufacturing process is being modified .

CONCLUSIONSPoints of view for future work :

• To build a better model , linear or non-linear ,and check the influence of adding derivative .

• To explore robust control methods .

• To explore adaptive control methods .

• Check the influence of adding derivative .

• Change the location of the temperature sensor.

• Thermal coupling improvement.

• Thermal mass reduction.

• Better heat sink.

Control And Robotics Laboratory 26