Transcript Imaging System of a Bose-Einstein Condensation Experiment, and

U.C Berkeley
Physics department
Stamper-Kurn’s group
Imaging System of a Bose-Einstein
Condensation Experiment,
and its Automation
Fabien Lienhart
August, 29th 2003
Plan
• Recent steps forward in the BEC
experiment
• The Imaging System
• Automation of the imaging system:
Visual Basic in WinView
Improvements and breakthroughs
on the BEC experiment
•
•
•
•
•
Ultrahigh Vacuum fixed
MOT laser brighter and monomode
Polarization Gradient cooling set
First efficient magnetic trappings
Next (last?) step…
Back to a Ultra-High Vacuum
Problem: water leaking from a coil in the vacuum chamber
Open the chamber, fix the leak, close the chamber and Back to the vacuum
Torr
•Rotary pump
10-3
•Turbomolecular pump
10-7
•1 week baking-out,
monitored by a Rare Gas Analyser
10-11
•Gauge limit
MOT Laser
How things work…
Problem
Ugly beam which could be stronger
Solutions
•Tapered amplifier: up to 300 mW
•Polarization maintaining
optical fiber
Results
•After the fiber: - 70 mW
- monomode
- good shape
•Up to 3 billion atoms trapped
Optimizing pg-cooling
How things work…
Easy case of lin –lin light
Improvements
• Better beam shape
• Transition between the two
pg-cooling steps found:
optimization of T and N
Results
30 mK reached
(Time of flight measurement)
First efficient magnetic trappings
How Ioffe-Pritchard trap works…
Results
• More than 600 million atoms trapped
• Lifetimes ranking from 15s (low Bias-Field) to 50s
• Adiabatic compression achieved
The Imaging System
•
•
•
•
The constraints
The experimental device
Characterization of the system
Results and future work
The Constraints
• Requirements of the system
- Resonant light has to be used: 2-to-3 beam
- Top-bottom axis
- Keep the polarization of the light
- Three very different magnifications: ~ ½, 5 and 16
• The difficulties
- Top-bottom axis is already crowded!
- Only a few gold mirrors can be used at 45o
- Precise magnification needs to be known (quantitative
imaging) and no way to place a fine object in the center of
the trap!
The Experimental Device
Characterization of the System
Results
Mag ½
Magnification:
0.510  0.015
Resolution:
80 mm (close to Camera limit)
Not to sensitive to the position of the MOT
But sensitive to the angle of the last lens (distortion)
Mag 5
Magnification:
4.60  0.01
Resolution:
8.7 mm (close to diffraction limit)
No distortion
Resolution very sensitive to the position of the cloud
Mag 16
Magnification:
12.0  0.6
Resolution:
14 mm
Position of the camera affects the magnification (.35/cm)
without really changing the resolution
Very sensitive to the angle of the last lens (distortion)
Automation of the Imaging
System
• How WinView works
• Adding buttons with WinView
• Example of routine: rotating the images
• Future work
How WinView works
•WinView controls the camera
•Automating WinView:
Using Macros
Easy,
but limited and buggy
Or…
Adding buttons with Visual Basic
1. Write your script in VB
Object Oriented Programming
Classes which implement WinView
2. Transform it into a DLL
3. Register the DLL
4. A new icon should appear in
WinView’s taskbar
The implemented buttons
• Close All - closes all the windows
• AutoSave -
saves all the windows with the date
• QuickASCII - saves the image as a text
• AbsorptionLoop - cycles absorption pictures
• RotateFrame -
rotates the pictures
Example: the rotation
Problem
Matrix index must be integer numbers
Which is not the case after rotation
Solutions
Implementation of various algorithms
1. Closest neighbor
2. Gaussian interpolation
3. Bicubic interpolation
4. b-Spline method: 2 ideas
leading to the best results
Idea 1 - Three-pass rotation
Idea 2 – Efficient 1D interpolation
Cubic b-Spline
Third order
Basis
Piecewise polynomials
Advantages
- Normalized contributions
- Compact support -> local control
- Interpolated function is C2
- Fast implementation: z-transform of the convolution gives
an efficient recursive algorithm
Performances of the different methods
A.
B.
C.
D.
E.
Original
Clothest neighbor
Gaussian interpolation
Bicubic interpolation
b-Spline method
Time (in s) taken for a rotation
6
5
4
3
2
1
0
B
C
D
E
Conclusion
Work achieved
- Imaging system: characterization and limits of the system
- WinView: add-ins
In the next month
-Imaging system:
- way to easily calibrate the system
- try different lens for mag ½ and 16
- WinView: gaussian fit of the profile of the cloud