Final Research Update - Bimodal Tesla Coil

Transcription

1 Final Research Update - Bimodal Tesla Coil Collin Matthews Advisor: Dr. Jovan Jevtic 5/21/2013 Abstract The Tesla coil, invented in the 1890s, has found applications in areas as varied as radio-technology, medical and industrial X-rays, electromagnetic pulse weapons, lightning simulators, education and entertainment. Since its creation many evolutions have occurred: spark gap, rotary spark gap, vacuum tube, and solid-state designs. When used as lighting simulator Tesla coils are only capable of producing current pulses of short duration as would naturally occur in cold lightning. However, the most damaging form of lighting, hot lightning is a continuous current of much longer duration. We are proposing the first bimodal Tesla coil. In our design we will simultaneously excite the Tesla coil at two frequencies. The purpose of the first frequency is to excite the conventional mode of the Tesla coil, to establish a high voltage that can initiate the breakdown of the spark gap. Unique to our design is the second frequency, which excites a higher order mode of the Tesla coil resulting in a continuous current flow once the high voltage terminal has been grounded though the spark. Our hypothesis is that the addition of a higher order mode will extend the spark duration, allowing us to add a continuous current component to the discharge of a Tesla coil. We have run simulations based on laboratory measurements of a Test coil. The resonant frequency and driving impedances with the discharge terminal open and short circuited to ground were measured and recorded for the first three resonant frequencies. Using the data from the measurements, we built a power supply capable of producing two independently adjustable frequencies to drive the coil. 1 of 20

2 Table of Contents Abstract...1 Project Timeline and Updates...3 Basic principles and operation of a rotary spark gap coil...4 Why is a Rotary Interrupter Spark Gap Important?... 4 The Oberbeck Theory...5 Initial Multisim Simulation...6 What I have learned from helical resonator research...9 Helical resonators and a bimodal coil Physical Coil Characteristics DC Resistance Physical Coil Characteristics Coil Inductance Measured Resonant Frequencies Matlab Calculated Resonant Frequencies Matlab Projected Voltage and Current Distributions Bucket Coil Basic Physical and Electrical Properties Inverting Coil Driver Simulation and Design Inverting Coil Driver Current State Final Remarks of 20

3 Project Timeline and Updates The Table Below contains the timeline originally set in the submission for the research project. Background information was compiled and multiple simulations have been done using Multisim based on accepted theories of Tesla coil operation. Most of the winter term was spent designing and simulating a power supply capable of driving a coil with two independent frequencies. We expected to be testing our theory once the power supply is finished. Unfortunately to date, we have only verified that a non-conventional standing wave can be excited on a coil with the top grounded. We are currently working out unforeseen feedback issues with the supply that has prevented us from testing both modes at once. The National Conference on Undergraduate Research (NCUR) held at UW-Lacrosse was attended in mid-april and a live demo was given of both the conventional and non-conventional mode. The future plan is to continue the work on the power supply this summer in order to test bimodal operation. Suggested Timeline UR-4981 (Fall ) Part 1: Background and Literature Search Part 2: Preliminary Simulation and Testing UR-4982: (Winter ) Part 3: Power Supply Design and Construction Part 4: Testing of the Bimodal Tesla Coil UR-4983: (Spring ) Part 5: Thesis and Publication 3 of 20

4 Basic principles and operation of a rotary spark gap coil 1. Charge flows into the capacitor (C1) on the primary side. a. This charges the capacitor until a critical threshold is reached. 2. At a critical voltage, the air in the spark gap of the rotary wheel has been ionized to the point of breakdown, causing an arc to form. a. This essentially short circuits the transformer and capacitor bank through the primary coil causing a large oscillating current to flow through the primary coil. b. Due to the inductive coupling between the primary and secondary coils, energy is transferred into the secondary circuit. 3. The LC circuits exchange energy until the average voltage or current drop below another critical level, or the rotary spark wheel extends the path of the arc to a point that the arc becomes extinguished. 4. The spark is extinguished thus separating the primary inductor and capacitor. a. From the secondary s point of view, there is no longer any resonant circuit coupled to it and the energy is trapped there. b. Then energy resonates in the secondary slowly ringing down, assuming another path was not established for the energy to flow from. This could be in the form of a streamer discharge into air or a spark to a grounded device. 5. The process repeats as the tank capacitor begins to charge again. Why is a Rotary Interrupter Spark Gap Important? Rotary interrupters can handle a large amount of instantaneous power; to date solid state technology is only capable of driving smaller coils. You can control the duration of the arc better than with a standard spark gap, due to the ability to control the speed of the contacts passing the primary arc zone. You can be assured that an arc will not be maintained and cause your HV transformer to be shorted for an extended amount of time. 4 of 20

5 The Oberbeck Theory Oberbeck created the first well known theory of how Tesla s Coil operated; He mathematically concluded that Tesla had nothing more than two resonating circuits loosely coupled. The mathematical description is a transient analysis of an inductively coupled system. Although his theory has a limited application and does not model the standing-wave phenomena that we intend to use, it is a simplified way to analyze the actions of a Tesla coil for basic use and does provide us with three important conclusions and equations that have been tested and found to be accurate. 1. There is a splitting of the resonant frequency directly proportional to the coupling coefficient. Normally you would have a single resonant frequency, but in this case there are two peaks that deviate from the expected resonant frequency depending on the coupling of the coils. f kf o Where f o Is the resonant frequency of the RLC secondary and k is the coupling coefficient. 2. An estimation of the voltage at the output, or the ratio of input vs. output voltages. C 1 V2Max V1 C 2 Where C 1 is the tank capacitor and C 2 is the total secondary capacitance. 3. The time required for the energy stored in the primary tank capacitor to transfer to the secondary and vice versa. T T o 2k In the following section we will compare calculated gains and resonances with the ones obtained by using the above three equations. 5 of 20

6 Initial Multisim Simulation In order to develop a more complete understanding of Tesla coil operation and to verify Oberbeck s theory, a simple lumped element circuit was created, as shown on the schematic. The system was given an initial condition of the primary tank capacitor being charged to 1 V, simulating the moment before the spark gap forms and completes the circuit. Then the Spark Gap switch is closed and remains closed for 33uS before opening again. This allows the Primary circuit to transfer its energy into the secondary coil. Not noted in the diagram is that the coupling coefficient of For some of the calculations in this section, the spark gap switch was removed from the circuit and the resistor (R1) was connected directly to the primary coil. This allowed the analysis of a transfer of energy over an extended period of time. In the transient analysis shown below, the Secondary coil voltage was divided by the ratio of capacitances, just as the second point of Oberbeck s theory suggests the voltage step up from the primary to the secondary is approximately equal to the following equation: VSec VPri CPri CSec Keeping in mind that the system has losses due to the resistances placed in the series LC circuit. The above output circuit voltage normalized to the expected equation output vs. an input of 1v. This demonstrates that the second point of Oberbeck s theory is applicable for the estimation of output voltages. You clearly see the energy transfer initially into the secondary, up to t=33us (when the spark gap opens). Once the Spark Gap opens on the primary side, the secondary is no longer coupled to the primary and energy transfer between the LC circuits stop. Now the energy resonates in the secondary circuit slowly ringing down due to internal losses from the resistor, independently of the primary coil. If the circuit was lossless, the voltage graphed should reach approximately 1v and stay at that level indefinitely. It reaches 6 of 20

7 approximately 73% of the predicted value in the modeled circuit. This demonstrates that this part of the theory is applicable to spark gap circuits and can be used to roughly estimate the final output voltage. It is worth noting that this is applicable to spark gap coils and not continuous wave coils such as the one we will implement. Next, to show an idealized example, using the circuit with the switch removed, when an oscilloscope was attached to the input and output of the circuit, the image shown below was observed. With an input of ~1V, the output was ~ 32V. Using this modified equation: we see that Oberbeck s theory holds exactly in a lossless example. VSec VPri CPri CSec 1 20nF 20 pf 31.6 The waveforms show a lossless case example of the ability to use the capacitance ratio of the primary vs. the secondary to predict output voltages. The input (Blue) is 1V and we can see the output (Red) rises to approximately 32V. Using the above simulated waveforms, we can also examine the third important point of Oberbeck s theory. The time it takes for the energy to transfer from the primary circuit to the secondary is dependent on the coupling of the inductors and the resonant frequency. With a coupling coefficient (k) of 0.05 and a f o of 300kHz when substituted into the equation: To 300k It shows the time from one peak of the primary to the next T 33.3us 2k 2(0.05) adjacent peak of the secondary is exactly as shown, 33.3us. The time scale as shown is 20us/Div, you can see starting at the 2 nd division from the left the primary (in blue) is at 0V and the secondary wave (in red) is peaking. As we follow the cycle forward, we see the primary rise to its peak and the secondary fall from its peak to 0. This demonstrates an exchange of energy between the primary and the secondary. If one counts out the sub divisions you will find the difference of time for the above process is ~33us, just as was estimated from Oberbeck s equation. 7 of 20

8 Finnaly we can examine if a frequency split occurs at the resonant frequency as Oberbeck s theory predicts with the following equation: f kf o In our model, the switch that represents the spark gap was removed in order to allow a spectrum analyzer to examine an uninterrupted flow of energy between the two coupled LC circuits. The resonant frequency of our model f o is 300kHz, which was found with the following equation: 1 o 2 LC With a coupling coefficient (k) of 0.05 and a f o of 300kHz, we should expect each peak to be ±7.5 khz from the center frequency. As illustrated below, there are two peaks that form, one on each side of the 300 khz fundamental frequency as predicted. The above spectrum analysis centered around the 300kHz resonant frequency shows the split predicted by Oberbeck s theory. In this case, about 7.5 khz from resonance. From the above comparisons, it was found that the simulated values in Multisim match the calculated values obtained from Oberbeck s theory. This justifies their use for any further analysis that does not strictly depend on the effects of standing wave resonance, which will be covered in the following section. 8 of 20

9 What I have learned from helical resonator research As it turns out, a properly constructed Tesla coil is acting as a helical resonator and in most cases, as a quarter wavelength helical resonator. I have found the easiest way to explain this phenomenon is to think of the electrical current you apply to the coil as a sound wave and the coil as a long hall. Sound, as you know, can cause destructive and constructive interference. Now imagine this hall as being ¼ of the wavelength long, and having a rigid sound reflective wall on the far side. What happens if you keep sending a sine wave down the hall? The sound pressure at the wall will increases due to the constructive interference between the incident and reflected waves. At the same time, the sound pressure at the source would be cancelled by the destructive interference between the incident wave and the reflected sound wave which is delayed by a roundtrip phase of 180 degrees. The wave will reach its maximum pressure at the end of the hall while the source end will maintain zero pressure. The only limit to the amplitude in this case would be the reflective and sound propagation losses. Now let the sound pressure be the voltage and the hall the wire of the Tesla coil. The end that is connected to the source will maintain a low voltage with a high current flow, but the end of the wire or top of the coil will have high voltage due to a constructive interference in a standing wave. This will cause the voltage to continually build until a point of breakdown occurs or the resistive losses in the coil dampen the reflections to a quasi-steady state maximum. As Corum put it there is no fundamental limit in classical physics on the voltage rise. It is also worth noting at this point the current is essentially zero at the end of the coil. That being said, one of the biggest losses is the internal resistance of the coil. These effects can be limited but it is very difficult if not impossible to eliminate all losses. The discussed analogy may or may not work better with a Solid State Tesla Coil (SSTC) because of its ability to produce a highly accurate pulsed wave form, but for coils of higher power, there is no suitable solid state hardware that can withstand the intense power spikes produced during operation. We will assume the setup below to be a rotary spark gap coil, bearing in mind that it may not be as effective. First, consider the operation of a coil assuming it is a spark gap coil. The tank capacitor on the primary side is charged until the arc is created. Energy is then transferred to the secondary side as described in the operation of a Tesla coil previously discussed. When the primary spark gap is extinguished, the energy trapped in the secondary coil begins to oscillate between the capacitor and the inductor at a frequency for which the length of the inductor is approximately ¼ of the wavelength. A unique phenomenon happens; a slow wave VSW pattern is established. This standing wave builds to a maximum and then slowly rings down due to losses in the coil such as resistance, capacitive coupling to the outer world and coronial discharge to list the top three contributing factors. Next, assuming you have a primary and secondary coil (For pulsed single coil design such as a SSTC this point would not be applicable.) The Rate of energy transfer from primary to secondary is dependent on coupling factor k. So a tighter coupling results in faster energy transfer. Unfortunately, the higher the coupling, usually the closer in proximity the two coils are. If the primary and secondary become to close, an arc can establish between the coils. This can damage not only the coil but possibly other components of the circuit. 9 of 20

10 Next, while there is no exact solution for the transient fields in a helical resonator, assuming a quasisteady state operation we can look at the peak output voltage as a simple sum of the forward input voltage and the addition of the reflected past forward voltages that build during operation. Unfortunately it s only applicable to continuous wave excitation. Instead an engineering approach has been taken to this problem. Unfortunately these types of calculations and coil physics are outside the scope of this project but Dr. Jovan Jevtic has created a Matlab simulation to calculate the different driving frequencies or resonances of a coil and the voltage current waveforms on the coil by separating the problem into electrical and magnetic analyses. First, if one looks at the coil as a purely electrostatic problem, this is possible because at points in time the current flow will be zero and you will have a separation of charge. This is the instant with the highest potential as well. Then the coil can also be treated as purely magnetic problem because once the current flow is at its peak, the distributed charges will all be equal to zero. This allows one to examine the flux build up, which in turn can again be used to find the total potential that can be achieved during the collapse of the flux. I will continue to learn more in the upcoming weeks on the specifics how this program creates these estimations. Helical resonators and a bimodal coil Normally a Tesla coil produces a brief electrical discharge which may appear to last long because it repeats many times per second. The reason is that the standing wave is depended on the coil parameters and the circuit being properly tuned. This is true for both theories mentioned previously. Once an arc is created, the fundamental circuit is altered and the standing wave form can no longer be produced. Or if you re looking at it as a lumped element, the element is no longer what it was. The spark adds capacitance and inductance to the circuit, so the element is no longer tuned to the driving source. A simple way to think of this is that essentially the same thing happens if you tune a radio and then bring an object near the antenna; the radio goes out of tune. The reason the helical resonator theory is so important is that it models what happens when the circuit is driven with more frequencies than just the fundamental frequency. You can drive it with any number of wavelengths that result with a peak appearing at the end of the wave guide or wire. If we could drive the coil with two frequencies simultaneously, one frequency to establish the maximum voltage rise at the terminal and another frequency to allow maximum current transfer once the spark is established and the fundamental circuit parameters are altered. This could allow us to maintain an arc for an extended period of time. In order to take advantage of this theory, we need to generate as close to the required frequencies as possible. In order to do this we will use a solid state driver instead of a spark gap. In doing so we will reduce the maximum amount of power that the coil can be driven with to 1kW. Keeping in mind this is a proof of concept design, the power reduction also makes it easier because anything more than that would make it difficult to use a standard 120V outlet to power it for demonstrations. 10 of 20

11 Physical Coil Characteristics Before we could design a power supply to drive the coil, we first had to get coil characteristics. Initially we were given a beautiful coil created by Dr. Fennigkoh. We designed a test plan to gather both physical and electrical characteristics. DC Resistance We measured the DC resistance of the coil using two different methods. 1. An Agilent 34401A DMM was used. The resistance with leads shorted was first measured and then the system was measured. The result was: =26.032Ω. 2. We used a constant current supply and measured the voltage across the terminals of the coil. The result was: V/0.1999A=26.142Ω. The difference of these two measurements was 0.4%, considering these instruments are in the ±1% class of accuracy this is acceptable. We will assume Ω. Physical Coil Characteristics The only physical characteristics we can accurately measure the outer diameter of the coil being in and there are 108 turns of wire in 3 inches. The pitch of the winding was determined by taking a digital photograph of a winding between two marks of known separation and by counting the number of turns after the image was enlarged on a computer. But using this along with the measured DC resistance and number of turns per inch, we can find the gauge of wire, the length of wire, and the number of turns on the coil. 1. Finding the wire gauge and length: 1in 36 *92% copper in wire dia. turns As it turns out, this diameter is almost exactly correct to be #22 AWG, so we assume that it is. This wire size has a resistance of 16.14mΩ/ft. So using the Measured DC resistance, we can find the length ft m m / ft Now that we have the wire length, we can also find the number of turns based on the diameter m 1017turns m / turn One item worth noting is that the diameter we measured is the outer edge of the wire on top of another piece of plastic, so it is reasonable to assume that the coil probably has a few more turns, but we will use this value for the remaining calculations. Another note is that the coil has a sizable metallic rod at the top with a jack to insert a wire; this will add a noticeable capacitance to the coil in respect to its self-capacitance. 11 of 20

12 In summary, the coil has m of #22 AWG, resulting in approximately 1017 turns. It has an approximate height of m. Coil Inductance This value was experimentally found in 3 ways. 1. We measured it using an LRC meter, and ran it at 4 different frequencies, below are the results. The average of the first three measurements was taken because at 100 khz the meter no longer gave a series resistance, and it was assumed not to be accurate. The averaged inductance value was found to be 30.28mH with a series resistance of 27.05Ω. We feel this resistance is slightly higher due to the driving frequency. (normal level) total system Freq L R 100Hz 30.30m kHz 30.30m kHz 30.26m kHz 29.3m na* 2. Using a known capacitor value placed in parallel with the coil, the coil was excited using an external field coil wrapped around it and the resonance was measured on a second external field coil also wrapped around the body of the main coil. The 1 st field coil was driven with a signal generator and manually swept through resonant frequency that was estimated to be around 8 khz. The second field coil was attached to an oscilloscope to allow viewing of the exited current. The peak value was found, and then the two -3dB values were found. The distance between the -3dB values was calculated. Using this information we placed the resonant frequency f o centered between the two -3dB values. The calculated coil inductance was found to be 30.60mH. In the equation below, the 8.044kHz is the resonant frequency based off the two -3dB frequencies Res Freq khz w L 2 LC w C +3dB khz 1-3dB khz 30.6mH 2 9 (2 *8044 hz) 12.8x10 F Res from 3dB khz 3. Using the formula for the inductance of an air core solenoid (L = µ₀n²a/ln). This can be expected to have a relative degree of accuracy because the length vs diameter ratio is large making it a more ideal infinitely long coil. This value will not be averaged with 1 and 2 because of all the added previous error, we assume the above two methods are more accurate. This value does prove that everything is fitting together and no major mistakes have been made. (4 x10 )(1017) ( ) m m 32.01mH 7 H m 2 In summary, averaging the values from the above first two methods, we found the inductance to be 30.44mH. 12 of 20

13 Measured Resonant Frequencies Using a network analyzer, the coil was positioned on top of a wooden table and connected. The first three resonant frequencies with the top open and grounded were found. This was done by looking for a maximum in the conductance. The Q value was gathered by looking at the -3dB values of the maxima on a log magnitude scale, and finally the series resistance was gathered by taking 1/max conductance. Resonant Freq [khz] Ungrounded Top Q Series R [Ω] Grounded Top G Network Analyzer G Max=1/R Coil f Matlab Calculated Resonant Frequencies Using the GUI program written by Dr. Jevtic, the first 3 resonant frequencies for an open and grounded top were calculated. They are in the table below; one thing to notice is that the program has higher frequencies than the ones measured, by 5% to 10%. This can be accounted for when you assume the metal rod and wire insert at the top of the coil must have some self-capacitance which would in turn drive down the resonant frequency. It is also worth noting that the program correctly calculates the inductance of the coil within 1% of our experimentally measured value. Top Open Res freq [khz] 1st nd rd Top Grounded Res freq [khz] 1st nd rd 1240 Inputs Calculated d=.154 m h=.7175 m Nt=1017 n=500 L=30.86mH C=203.97nF 13 of 20

14 Matlab Projected Voltage and Current Distributions Using the program created by Dr. Jevtic, the estimated current and voltage distributions were graphed based on the vertical position on the coil. Coil with Open Top (1 st Eigen-mode) Coil with Grounded Top (1 st Eigen-mode) Coil with Open Top (2 nd Eigen-mode) Coil with Grounded Top (2 nd Eigen-mode) 14 of 20

15 Bucket Coil Basic Physical and Electrical Properties Unfortunately after examining the above data, we found that Dr. Fennigkoh s coil has to high of a resonant frequency in order to efficiently drive it with the power electronics components we have. The switching losses of the mosfet module we have on hand would be excessive, and designing a better high speed driver is out of the scope of this project, we decided to use a coil that Dr. Jevtic possesses. It is smaller, but as it turns out has 5x the inductance which resulted in almost 1/3 rd the resonant frequency. So the above tests were repeated for his coil and the results were a DC resistance of 256.2Ω and an inductance of mH. The table adjacent shows the open and grounded top resonant frequencies. It Resonant Freq [khz] Ungrounded Top Q Series R [Ω] Grounded Top can be seen that the frequency of the 1 st eigenmode with grounded top ( khz) is more than a factor of 2 lower for the bucket coil then for Dr. Fennigkoh s coil (566 khz) which will allow us to use slower power electronics components to excite this mode. Our generator should still be able to drive the conventional mode of Dr. Fennigkoh s coil which resonates at khz. Inverting Coil Driver Simulation and Design Now that we know the specific coil parameters, we can design a power supply capable of creating two independently tunable frequencies to simultaneously drive the coil with. In order to do this we will implement a center tap DC rectifier to produce ±170V DC and filter it. Then the power is routed to a full bridge MOSFET module. Our design uses one half of the bridge to drive each frequency. The frequencies are created with a NE555 astable square wave generating circuit to allow precise tuning via a 10 turn potentiometer. The square wave generator drives a MOSFET driver that in turn takes care of driving the MOSFETs gates with a sufficiently large current. The driver provides an appropriate blanking time of 0.4us, just short enough for our frequencies, and includes the circuitry for driving the high-side of the MOSFET bridge via a boost capacitor. To date we have completed the simulation of the supply driving a coil. One key piece of information we learned from the simulation is the interaction of the two different driving frequencies causing increased power dissipation inside the FET module to the point of us needing to possibly add a duty cycle of approximately 50% with a period of 100s of milliseconds in order to keep the average power below 150W from being dissipated by the bridge when the coil is consuming 1KW of power. Below is the Multisim simulation with a switch used to flip between the two different coil models. One model is the coil with the top open and the lower model is with the top grounded (simulating when an arc occurs). The high voltage sides of the coupling transformers are connected in series to form a series connection of two square-wave voltage sources operating at two different frequencies. The turn ratios of the coupling transformers were adjusted to deliver approximately 1kW of power to each mode. The coil is never resonant at both frequencies at once, so the current which flows thru both transformers is very nearly sinusoidal due to the filtering effect of the Tesla coil resonator. 15 of 20

16 The Multisim simulation of the proposed circuit driver, the switch differentiates between a coil with a grounded top vs. an open top parameters. On the following page you can see a result of the transient analysis when the coil has an unloaded open top. The voltages have been scaled by a factor of 10x so you can visualize the current and voltage wave forms on the same graph easier. This turned out as expected, one channel fed back into the other to an extent, causing a large power loss inside of the MOSFET. Then a power analysis was performed on the power dissipation inside of the MOSFET module. This gave us a picture of some of the instant power spikes our hardware would be enduring, sustained 3kW pulses for over 1us. If this image is zoomed in we witness a lot of transient noise, but we can examine average wattage using a watt meters in Multisim. 16 of 20

17 The Multisim transient analysis of the voltage and current output with an unloaded top, the voltage has been scaled by 10x to allow easier viewing. The Multisim transient analysis of the power dropped across the FET module, the average power was better found with watt meters, but this gives an idea of the large transient spikes. 17 of 20

18 Inverting Coil Driver Current State The spring quarter was spent constructing and testing the dual frequency driver. In initial startup tests performed with a load resistor revealed a lot of high frequency noise that resulted due to the switching, this caused distortion and high overshooting of 300% in the square wave along with a lot of ringing. Due to the high frequency s we are operating at, the simulation did not accurately show the full effect of the transients pictured in the previous page. Packs of ceramic capacitors were placed in parallel with the power rails and each rail to ground. This capacitive coupling to ground will dampen out the noise and to help stop the ringing that appeared on the output. This helped but the square wave still had a large resulting ripple. The unit operated like this, but once the transformers were connected to the switching output, even with the secondary s unloaded, the high inductance produced voltage spikes that would instantly fry the primary FET module. Many methods were examined to reduce this, but the oscillation on top of the square wave was a frequency with less than a magnitude difference from the switching frequency so this limited the effectiveness of common fixes like snubber filters. That being said, we conducted initial tests with the bucket tesla coil and successfully excited both a conventional and nonconventional mode one at a time. Unfortunately with the driver running on its edge of operational parameters, we could not run simultaneous bimodal operation. As of writing this, we are currently examining alternate options to improve FET life by reducing the reflected ringing voltages. The best option appears to be to remove the ferrite core transformers and create an air coupled transformer by wrapping wire directly around the coil body. This will give the circuit less inductance and thus the ringing and increasing the frequency of any remaining ringing to a point that may be easier to filter with an LC network. Alternate FET s have also been investigated with higher voltage ratings that may better withstand the transient spikes present when driving an RLC load. 18 of 20

19 Below are the basic operating flow chart and a picture of the finished product. High level overview of the driver. Picture of the top view of the driver: on the left you see the driver control board, on top is the FET module, and on the right are the two output transformers. 19 of 20

20 Final Remarks In the last 11 months I have learned a tremendous amount about Tesla, power electronics and high frequency switching applications. I have seen this project go from an idea to an operational prototype based on theory and research. The prototype successfully excited a non-conventional mode waveform on the bucket coil with the top connected to ground. Unfortunately the testing and operation of the prototype driver has fallen short of expectations, due to hardware malfunctions and larger than expected transient voltages frying the primary FET module. We are currently working on these problems in order to fully test the driver in a bi-modal configuration as discussed above. Although the class has officially ended, I expect to and am excited to continue working on this project with the assistance of Dr. Jovan Jevtic if available in order to verify that a bimodal tesla coil is realizable. 20 of 20