Summary:

When used in Tesla coil primary circuits, neon sign transformers often fail due to high-frequency voltage and current stresses. In order to protect the transformer from these stresses, various filter networks can be used to block destructive signals from reaching the transformers secondary windings. This paper describes testing that was done using a simple RC filter network. The before and after affects of adding this filter are shown along with details of the filter’s construction.

Construction:

The filter consists of three parts. There is a single capacitor that is placed across the output of the transformer. This capacitor is a 2.7nF 30KV doorknob style ceramic capacitor. The value of this capacitor is not critical and this part was already at hand. There are also two 1000 ohm 250 watt power resistors that are placed between the 2.7nF capacitor and the primary circuitry. One resistor is in each of the high voltage leads. These resistors were sized to account for the high voltages they may encounter. Further testing may show that the voltages across these resistors does not reach sufficient levels to justify the large parts used here. Figure 1 shows the filter network as it is connected to the primary circuit and the transformers output.

Figure 1. Filter network and connected in primary circuit.

Setup:

The experimental setup is shown in figure 2 and a block diagram of the setup is given in Figure 3. The primary system used for testing does not have a secondary inductor to limit the danger to the test equipment. The primary capacitor is a 17.25nF 30KV ceramic capacitor bank. The primary inductor is a 120uH linear coil inductor. The spark gap is a simple 0.02 inch gap that has holes drilled so that air can be pumped into the gap. The transformer is a standard 15KV 60mA neon sign transformer.

Figure 2. View of the test setup.

Figure 3. Block diagram of the test setup.

Results:

The test was run in both the transformer across the spark gap and the transformer across the primary capacitor configurations. The test was run with and without the filter network in place. The maximum voltages used in this testing was limited to about 3000 VRMS to prevent damage to the transformer.

The first set of pictures (Figure 4) shows the voltage and current waveforms of the transformer’s output. The transformer is placed across the spark gap without the filter network in place.

Figure 4. Voltage and current waveforms with the transformer across the spark gap. No filter network

Top Trace Voltage 2KV/div

Bottom Trace Current 50mA/div

Figure 5 shows the same situation with the filter network in place. Notice that the 90mA current spike has been eliminated.

Figure 5. Voltage and current waveforms with the transformer across the spark gap. The filter network is in place.

Top Trace Voltage 2KV/div

Bottom Trace Current 50mA/div

The test was repeated for the much higher stress case of the transformer being placed across the primary capacitor. Figure 6 shows the transformer voltage and current waveforms with no filter.

Figure 6. Voltage and current waveforms with the transformer across the primary capacitor. No filter network

Top Trace Voltage 2KV/div

Bottom Trace Current 50mA/div

Figure 7 shows the same situation with the filter network in place. The heavy 100mA oscillation has been greatly reduced. The voltage step has also been significantly softened. There is a concern that high frequency voltages will not evenly distribute across the secondary of the transformer. The high inductance of the transformers windings my stop the voltage in only the outer few layers of winding resulting in very high voltage across these outer layers resulting in breakdown.

Figure 7. Voltage and current waveforms with the transformer across the primary capacitor. The filter network is in place.

Top Trace Voltage 2KV/div

Bottom Trace Current 50mA/div

Conclusion:

The filter network presented effectively reduced high frequency voltages and currents on the transformers secondary windings. In the case of the transformer being placed across the spark gap, 90mA currents spikes were eliminated. In the case of the transformer being placed across the primary capacitor, 100mA current oscillations were heavily attenuated. The high voltage oscillations were also practically eliminated and the very fast voltage step was significantly softened. If this transformer was operated at 20KV peak voltage instead of the 4KV peaks used in this experiment, the high frequency currents across the transformer may have reached 500mA. This is probably the mechanism by which so many neon sign transformers have been destroyed when they are used to power Tesla coils.