Focus Fusion-1 achieves much greater repeatability, closer to theoretical predictions
Major improvement in repeatability of fusion yield and beam production. Repeatable fusion yield is now within a factor of 4 of predictions. Clues found from data and simulation on improving filamentation, ending the early-beam problem and boosting yield up to predictions.
LPP researchers have made major advances in reducing the variability of Focus Fusion-1 (FF-1)’s performance and increased average fusion yield. Using our new unified database that puts together data from our neutron and X-ray detectors with that from our current-measuring Rogowski coil and the basic inputs for each shot, Lerner found that the shot-to-shot variability of FF-1 had changed dramatically over time. For most of its life, FF-1 shows the same large variability that afflicts many DPF devices, with a nearly 10 to 1 range in fusion yield for identical input conditions. But for a short period around March 30 of last year, variability was much less, dropping to around ±15%. At the same time, fusion yield rose dramatically, matching our theoretical predictions. So the mystery was—what was so different back then and how do we replicate it? To deepen the mystery, we learned from colleague Chris Hagen of NSTec, which has an even more powerful DPF, that they too had a similar short period of low variably and high yield—so this was not unique to us. Experiments in the past month found no clear evidence for two possible causes of variability, (1) changes in the synchronicity of switch-firing or (2) narrow peaks in the efficacy of the Axial Field Coil—resonances that produce much more efficient functioning.
However, we did find that any small deviation from symmetry greatly reduced yield and repeatability. When we changed the number of capacitors firing from 10 to 8, variability dropped dramatically, with the range of fusion yields dropping first to 3 to 1 and then to ±15% (around a yield of 5×10^10 neutrons). While 10 capacitors are not symmetrically arranged and 8 are, the current spreads out to make asymmetries quite small, so this effect told us that we could still improve the symmetry of the initial conditions of firing.
One possible source of asymmetry is the presence of tiny particles of copper and other materials deposited on the electrodes and insulator by the previous shot. To try to clean the particles up, as a first shot of the day, we fired a “cleaning shot” with a pressure that we figured would be too low to produce a pinch, and thus would not create more particles. The main shot after this produced a yield of 10^11 neutrons, double the amount we had been seeing and, for the 850 kA peak current, within a factor of 4 of our theoretical predictions. To confirm our idea, the second following shot (after the presumably “dirty” big pinch) was much lower in yield. When we repeated the process, the low-pressure shot did manage to pinch, so was not an ideal cleaner, but we again got 0.8×10^11 neutrons. Of course, we need to further confirm these few results with many more shots, but our ability to predict shot-to-shot results is very encouraging.
Another clue occurred when we fired after testing the trigger alone a few times. The yield on the subsequent shot was extremely low. The trigger has such low current that it can only affect the region around the insulator and the base of the cathode, including the tungsten pins that help to initiate the plasma sheath, so this is an indication that this region may be the one affected. Since the high yield of March 30 stopped after we had opened the chamber, and the early beam started in June of last year after we had done some work on the insulator and cathode, these are clues that changes in those areas must be corrected to eliminate the early beam. The chamber has just been opened again and indeed we found that the tungsten pins were no longer the same height as each other and there was clear evidence of asymmetrical sheath formation. This month we’ll be figuring out the best way to eliminate this important source of asymmetry so as to allow higher compression and more yield.
Finally, we noted last month that the early beam problem seems to be due to poor formation of filaments. Our continuing work on a simulation of the filament formation, with the able help of LPP consultant Dr. John Guillory and FFS volunteer Dr. Warwick Dumas, indicates that if the filaments form too early, with the current too low, they can be heated and expand before the current gets big enough to pinch them together. So small changes in the pins, for example, could make this race between heating and pinching go one way or the other, possibly leading to poor, too-fat filaments. Again, we should be able to fix this with adjustments to the cathode plate. So our simulation, unfinished as it is, may be able to contribute insight to our practical challenges.