The conclusion from analyzing the betatron motion data was that a significant vertical electric field was imparting an angle to the ion beam as it was being launched from the chimney. Please see Ion Source Studies - Part II: Simulations and Measurements to get the fine details. Modeling with PSF and SIMION has shown that the vertical electric field came from the slight vertical asymmetry in the design of the first copper chimney, previously described on the Early Ion Source web page.
Great effort has gone into designing an electrically symmetric chimney, including metalizing the ceramic boat on which the chimney sits. During the design it was decided to thin the wall near the aperture to increase the extraction field at the plasma sheath.
Additionally, we decided to incorporate "pullers" (extraction electrodes) mounted on the face of the dee to further enhance the extraction field. This has proven to be very successful, but has brought about it's own troubles. The vertical alignment of the "pullers" with the chimney aperture is critical to the initial launch angle (see figures 6 & 7). A dee positioner is being built so the vertical alignment can be adjusted during operation.
During operation of the cyclotron, the internal pressure was brought quite high (1E-4 Torr) for short periods of time (~ 1 minute). Under the high pressure conditions the beam can be seen via recombination. In the following two images, the RF power was about 700 watts - creating 10kVp-p on the dee. The initial first half of the ions' revolution can be seen spiraling as expected, but also downward (this comes from the poor vertical alignment of the pullers). The downward slope is so great, and at this early stage of the ions path the magnetic focusing effects are very weak, thus the ions are just lost to hitting the bottom of the chamber.
Only after the new chimney was designed, did the author find the following images of existing ion sources. We were pleased to find that our own, independent, evolution has brought us to the same construction as the Cyclotron Pioneers (Cyclotroneers) had developed.
(scanned from E. Segre's Methods of Experimental Nuclear Physics, vol. III)
(scanned from the proceedings of the Sixth International Cyclotron Conference, Vancouver, 1972.)