Adiabatic Demagnetization Refrigerator (ADR) Control
Program
Olin College Professor Holt and Professor Pratt in collaboration with
the XRS team at Goddard Space Flight Center have put together a summer
intership for ten Olin College students.
One of the five projects is to modify the control program for an Adiabatic
Demagnetization Refrigerator (ADR). It currently uses a basic Proportional
Integral Derivatieve (PID) control algorithm. The temperature input
occasionally has a data point that is physically impossible, so the
new program should take a wider range of data points into account to
lessen the effects of the error. There are also other noise issues within
the system. The goal of this project is to reduce noise and errors by
a factor 10.
Calendar
June
Monday 
Tuesday 
Wednesday 
Thursday 
Friday 

1 
2
Begin NASA study 
3 
4
Trip to Goddard 
7 
8 
9
Recieved code
& data 
10 
11 
14 
15 
16
Analyze data 
17 
18 
21 
22 
23
(Joe is gone) 
24
(Joe is gone)

25
Model ADR 
28
Get new data &
analyze 
29 
30



July
Monday 
Tuesday 
Wednesday 
Thursday 
Friday 



1 
2
Model system 
5 
6 
7

8 
9

12 
13 
14
Finish & send code 
15 
16
Refine & send back 
19
[Slack time] 
20
[Slack time]

21
[Slack time] 
22
[Slack time] 
23
[Slack time] 
26 [Slack
time] 
27 [Slack
time] 
28
[Slack time] 
29
[Slack time] 
30
End of NASA study 
Timeline in PERT Form
1. Get current ADR control code and some data results from Goddard.
2. Convert log files into Matlab data.
3. Create graphs in Matlab from data results.
4 . Figure out the magnitude of the noise so it can be reduced.
5 . Make further connections with data and what needs
to be changed.
6. Understand the physics of the adiabatic demagnetization refrigerator
(ADR).
7. Work through the mathematical equations for the ADR.
8 . Understand Proportional Integral Derivative (PID) control and find
where it fits in the system.
9. Understand Simulink to model ADR equations.
10. Analyze the new data and figure out exactly where noise is coming
from.
11. Model the slow demagnetization step of the ADR
using Simulink.
12. Fit the ADR model to the data to find the constants.
13. Figure out good smoothing algorithms.
14. Figure out good algoithms for ignoring outliers.
15. Model the exact system we are using to control
the ADR.
16. Become familiar with previous ADR control code.
17. Become familiar with C programming language.
18. Code the best smoothing and algoithms in C.
19. Use code in the model to simulate the new control algorithms.
20. Finish writing the code.
21. Contact Scott to tell him when the code will be there.
22. Email the code to Scott.
23. Get new data from ADR test with new code.
24. Analyze new data.
25. Identify any problems or things that need to be improved upon.
26. Fix Them!
27. Send them back for further tests.
28. Repeat steps 23 to 27 each week until done.