LPPFusion’s efforts to reduce troublesome oscillations in our FF-2B experimental fusion device have started to make progress. On October 21, we cut the highest-frequency oscillations in half, and increased fusion yield by 30% over our best previous shot with FF-2B. This puts fusion energy yield at 1/8 joule (one joule is one watt-second).
Is 1/8 joule a lot or a little? Compared to other private fusion efforts, it’s a lot. Other than LPPFusion, TAE Technologies is the only fusion company that has reported any fusion yield at all, and in March of this year they published a yield of 40 microjoules, which is 3,000 times less than our Oct. 21 shot. They have not published the energy input of their machine, but it is definitely more than FF-2B’s.
Our Oct. 21 shot is also pretty good in terms of LPPFusion’s recent work. It quadrupled the yield we got at the same conditions before we had reduced the oscillations (the highest previous yield had been at different conditions). It ended a 70-shot drought of new fusion yield increases. It also surpassed, by a bit (6%), any fusion yield that we achieved in 2017-2018 using tungsten electrodes.
Figure 1 gives an idea of the progress made. These graphs show the rate of change of the current in the device in shot 1 of August 9 and shot 1 of Oct 21. The early oscillations on Oct 21 are slower, and they damp down much faster than on Aug. 9. The pinch—the big dip in current—is much bigger on Oct 21. This shows energy moving much more quickly into the plasmoid where the fusion takes place, indicating a smaller, denser plasmoid. That is what produced four times as much fusion energy as on Aug. 9.
Fig. 1 August 9, shot 1 (left) shows the high frequency oscillations in the current at the start of the shot and the subsequent small dip at the pinch, when the fusion reactions take place. After we adjusted the switches, in October 21, shot 1, (right) the initial oscillations decreased in frequency and damped out quicker, leading to a much deeper current dip at the pinch and four times as much fusion energy. Vertical axis measures the rate of change of the current in units of 10 A/ns and the horizontal axis measures time in ns.
But 1/8 J is little compared to our near-term goals. To get a new record for our project we have to exceed the ¼ J that we achieved with FF-1 in 2016, using longer electrodes. To get closer to net energy (more energy out of the device than we put in) than any other fusion project in the world, we need to exceed 0.4 J. Beyond this, to fully validate our theories of how the plasma focus device works, we need to get to 10 J fusion yield. So we still have a ways to go, but our new understanding of the oscillations will help to get us there fairly quickly.
We knew back in September that the oscillations were caused by something in the external circuit of our machine, not inside the vacuum chamber (see “Tracking Down FF-2B’s Problematic Oscillations”). We found that the circuit was oscillating at four frequencies: 16 MHz, 28 MHz, 38 MHz and 40 MHz. Like detectives eliminating suspects, we started to rule out sources by disassembling the circuit and seeing when each frequency disappeared. LPPFusion Chief Scientist Eric Lerner and Research Scientist Syed Hassan could test the circuit safely by firing the trigger pulses. This pulse, which triggers the capacitors when they are charged, is too small to do any damage, but is big enough to set off the oscillations.
As we had previously thought, the lowest frequency, 16 MHz, was caused by current sloshing back and forth along the aluminum plates that connect the switches to the electrodes. That was not our main worry, as that frequency had always been present. The 28 MHz frequency came from the “ringing” of the trigger cables that carry the trigger pulse to the switches. The 38 MHz oscillation was definitely coming from the switches themselves.
That left the 40 MHz “tone” as the mystery. On the suggestion of LPPFusion’s Electrical Engineer Fred Van Roessel we checked out the grounding of the bottom aluminum plates, which were attached to a ground plate on the floor. Fred suggested that only one connection should be between the two plates. They were separated by just the right distance to explain the 40 MHz signal. Indeed, when two of three grounds were removed, the 40 MHz dropped a lot.
So the switches were the culprits—their 38 MHz oscillations started the ringing at other frequencies. But what caused the switches to start oscillating? (The “motivation” for the crime.) The fact that the current in the switches was oscillating even when the capacitors were not charged was a big clue. When the capacitors are not charged, the high voltage (minus 50 kV) pulse from the trigger causes a spark to jump to the nearest ground—which is the top plate that the spark plug is attached to, separated by a Lexan insulator. But when the capacitors are charged, the additional electric field created by the charge of plus 40kV causes the spark to jump vertically, allowing current to flow out of the capacitor to the top plate.
This is what is supposed to happen. What instead was happening in our case was that the switches were breaking down horizontally, to the top plate first, even when the capacitors were charged. That caused the back and forth motion of the plasma formed within the switch. That sloshing plasma in turn caused the current oscillations.
Why did the oscillation get worse in 2019? It looks like it was because the switches were firing so close to simultaneously, making all the switches oscillate in unison. Before this year, the switches fired at different times within a 20-ns window, enough to smear out a 26-ns-long oscillation. But with the new ceramic inserts, the switches became sufficiently similar to fire within 4 ns, putting all the oscillations in sync, making them larger.
The solution is to adjust the spark plugs, which are mounted on screw-threads, to move closer to the capacitors, reducing the vertical gap and leading to the switches firing first vertically. This is tricky since we must also increase the pressure in the switches to prevent them from firing by themselves, before the trigger. It will take us a bit to get the adjustment right, but we have already seen the initial results. After initial adjustments, we saw the partial reduction in the oscillations and a rise in fusion yield. We expect that the complete elimination of the oscillations will result in more dramatic increases in yield.