Digging in the Data—oscillations and x-rays

[DerekShannon]

From the February 2014 report, original item here (click for graph!).

The months that FF-1 is not running have given the LPP research team time to look more deeply at the hundreds of gigabytes of data that the experiments have been generating. One puzzle LPP Electrical Engineer Fred van Roessel looked at was the rapid oscillation in current and voltage that occurs when the machine fires and right after the pinch (see graph). These are almost always seen in plasma focus devices but have not had a satisfactory explanation. Van Roessel wrote programs to analyze the oscillations, concluding that they were due to current sloshing back and forth along the transmission plates that connect the capacitors to the electrodes. This hypothesis implied that the peak voltage the device generates during the pinch is proportional to the total drop in current at the same time. Sure enough, the data for the past year showed that the ratio of the voltage spike at the time of the pinch relative to the drop in current during the pinch is constant at 72 milliohms with a standard deviation of only 16%.

This result is important because it allows us to predict that the maximum possible voltage spike is 200 kV, well below the 1 MV estimated strength of the main Mylar insulators. It is even further below the 5 MV strength of the ceramic insulator that separates the electrodes.

In addition, we now have some part-time help with the data analysis—our new high school intern, Matthew Ho. Matt has been working with Van Roessel and Lerner to analyze the x-ray data which will give insight into the electrons in the plasmoid. Some tantalizing clues have already turned up. Stay tuned!

[asymmetric_implosion]

Alternative hypothesis on the ringing: The voltage probe is the victim of a ground loop. I’ve run across it before and by changing the way the ground is done can alter the amount of ringing.

The hypothesis on the voltage spike is right on. There are a few published papers relating the depth of the current bite with the voltage spike.

[Lerner]

No it can’t be a ground loop–the ringing is just the same in the Rogowski measurement of the current.

[asymmetric_implosion]

Are the rog coil and the HVP on the same ground?

[Francisl]

DerekShannon wrote: From the February 2014 report, original item here (click for graph!).

Van Roessel wrote programs to analyze the oscillations, concluding that they were due to current sloshing back and forth along the transmission plates that connect the capacitors to the electrodes.

A detector coil should pick up the oscillating magnetic field associated with the sloshing current.

[asymmetric_implosion]

A rogowski coil is a magnetic coil. It is used to measure the time derivative of the magnetic field and converts it to a voltage. When placed around the anode, the rogowski coil is related to the current in the anode.

My observation on previous PF devices is the rogowski coil is a pretty clean signal when properly shielded outside the detection region. High voltage probes are more problematic. Most HV probes require a local ground in addition to the digitizer ground. The local ground is really the problem. A few volts of noise on the ground can lead to artificial signal at the digitizer by affecting the ground rather than the primary signal you wish to measure. I’ve notice noised coming into the system when the HV probe is not attached. Depending upon the location of the probe the noise can vary greatly. Large PF systems have a harder time keeping the ground lead short so they tend to produce noisier HV probe signals. When the ground lead is very short the noise is minimized and the best measurement is realized.

[Francisl]

I am under the impression that the current is sloshing back and forth from the pinch area to the capacitors. If a detector coil is placed close to the cables connected to the capacitors it could determine if the current is oscillating back and forth at the same frequency as the voltage readings on the oscilloscope.

[Lerner]

Right–we have a Rogowksi coil built into the machine to monitor the current and it does indeed show the current oscillating at the same frequency as the voltage. Although all the grounds are connected, the Rogowski coil in grounded to the ‘scope while the high voltage probe is grounded to the support structure. The ratio of current to voltage oscillations and the frequency of the oscillations combine to give us evidence that the current is moving back and forth between the top and bottom transmission plates–they have the right inductance and capacitance for this. The energy is mostly radiated out the edges of the plates, creating a big RF noise source which we spent a long time shielding against the first year of FF-1’s existence.

[Francisl]

Is that oscillating energy lost as far as the plasmoid is concerned? Could that energy be used to partially recharge the capacitors if a switch could be designed to do that?

[Francisl]

Francisl wrote: Is that oscillating energy lost as far as the plasmoid is concerned? Could that energy be used to partially recharge the capacitors if a switch could be designed to do that?

I want to expand on that thought a little more.
1. The oscillating energy is lost as heat in the electrical equipment of the device and the Faraday cage of the room. Stopping that loss would keep things cooler.
2. Capturing the oscillating energy to partially recharge the capacitors would improve the energy input-output ratio.

I think this will become a design consideration in production machines or high repetition rate experimental machines. It would be good to start collecting ideas on how to address this problem.

[asymmetric_implosion]

Lerner wrote: Right–we have a Rogowksi coil built into the machine to monitor the current and it does indeed show the current oscillating at the same frequency as the voltage. Although all the grounds are connected, the Rogowski coil in grounded to the ‘scope while the high voltage probe is grounded to the support structure. The ratio of current to voltage oscillations and the frequency of the oscillations combine to give us evidence that the current is moving back and forth between the top and bottom transmission plates–they have the right inductance and capacitance for this. The energy is mostly radiated out the edges of the plates, creating a big RF noise source which we spent a long time shielding against the first year of FF-1’s existence.

OK, all grounds are connected. This by definition is a ground loop. Ground loops generate noise on signals by floating the grounds at different parts of the circuit leading to noise on the ground side rather than the power side. I don’t doubt you see oscillation in the current at some level but you also have to contend with ground loops.

IF you are seeing the current oscillating at a higher frequency than the initial frequency of the system (contributions from L, C and R for short circuit) then you have to have a transmission line like system. Typically, a ~1 us pulse system does not behave like a transmission line. You usually have to be much faster. One might argue that the fast behavior of pinch can do it but you see it on the trigger as well. The trigger should be reasonably clean as there is negligible current in the system. If you see oscillation at breakdown, it tends to be high voltage noise in my experience. It could be something else but I suspect if you can make the noise go away with a better selection of the ground point of the HV probe. It is a simple experiment that could tell the difference between a paper and a problem.

[Author]

Posts