To demonstrate the principles behind Optical Power Extinction Tomography we designed an experiment consisting of two coherent collimated beams that intersect within an object of interest. One beam is held fixed while the other rotates around the first (control) under computer control; the sample angle can also be controlled. Mechanical control of the angular separation of the beams is limited to a two-dimensional (2-D) plane, thus the current system is only capable of imaging 2-D objects.

As shown in the following figures, the source for this experiment consists of a HeNe laser which is split using a non-polarizing beam splitter and finally coupled into two polarization preserving fibers.

What is not shown in the figure that is in the picture is a quarter wave plate used to create circularly polarize light coming from the laser; the laser is linear.  This was done to save the cost of adding fiber rotators to the fiber alignment stages, it was assumed that the polarization preserving fibers would pick their preferred state of polarization from the circularly polarized beams. In hindsight this was a bad idea since not only do the the fibers pick their preferred state, they also allow other states to propagate randomly, which translates into fringe visibility fluctuations in the interference pattern. It is planned that the fiber rotators will be added in the near future to avoid these problems.

The fibers are routed to two beam towers as shown below. The paths each fibers takes is designed to stress this fiber as little as possible.

While both of the two beam towers serve to recollimate the light coming from each of the two fibers and deflect the collimated beams horizontally across the sample, they also serve unique functions. The stationary tower (the outer tower or Arm I) has a piezo translation stage that is used to change the path length difference between the two beams and is controlled with the computer. The rotating tower (the inner tower or Arm II) is  mounted on a Newport rotation stage allowing the angular separation to be adjusted between the two beams. At first small angles were to be ignored because of the occlusion caused by the towers, but it became clear that very large angles would make it very challenging to acquire data. The towers were then modified by replacing the lower mirrors with beam splitters to remove the occlusion at small angles. However, there are still certain angles were occlusion occurs but with the current system there is very little that can be done to completely eliminate it. The following two figures show the general structure of a beam tower as well as the general assembly of the rest of the experiment.

Note that the sample holder is also mounted on an identical Newport rotation stage that is positioned below and coaxial with the first since this is the precise location where the two beams intersect. Not shown in the schematic is a second beam splitter at the end of Arm II that allows light from Arm I to pass through for small angles and be detected with a photo detector.

There are 4 photo detector assemblies that make up our system. They include a reference detector and signal detector on each arm. Measuring the signal both before and after the sample allows for true power extinction measurements to be  made. Each detector assembly consists of lens, a pinhole, and a photo diode. The lens and pinhole (~5microns) effectively form a spatial filter allowing only the DC component of the forward propagating beam to be detected. Each detector is connected to a channel on a Tektronics (5???) mainframe oscilloscope, which is in turn connected via GPIB to a data acquisition computer.

To control this experiment we are using an Intel based PC running Mandrake Linux 9.0. The PC has a Computer Boards PCI-DDA02/16 data acquisition board as well as a Computer Boards PCI GPIB board. The following figure shows the general control layout for the system.

The software used for control is an application written at the University of Illinois called Imagekitchen.