[Home][2019 Index] [TCML] EMI filters for Tesla Coils [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[TCML] EMI filters for Tesla Coils

Here are some of my thoughts on EMI filters that I posted in reply to a question on Coilsmiths several weeks ago.

EMI filters for Tesla Coil usage

This may be a good EMI filter for your application,but look at the possible peak current problems. When measured in a 50-ohm system it will give you about 30 dBattenuation at 100 KHz, and your system likely will have an operating frequencyin this region, and thus the strongest emissions in this region. For voltageratios, 20 dB = factor of 10 attenuation, 30 dB = factor of 31, and 40 dB =100. Your system will not have 50-ohm input and output impedances, but this ishow filters are measured, and your real-life performance will be very roughly similarto these attenuations.

There are 2 types of EMI signals to be concerned about.Common Mode (symmetrical) signals have the same signal on each power line asmeasured with respect to ground. Differential Mode (asymmetrical) have thesignals measured between the 2 power lines. Different parts of the filter areused to attenuate the two types of signals. With EMI suppression  you are trying toattenuate the signals down to a level where they do not cause problems, youcannot totally eliminate them.

Many EMI filters are designed to start attenuating above500 KHz, so find one that has adequate attenuation at your fundamentalfrequency of operation.  At yourfundamental frequency of operation (probably around 100 KHz), I like to have aminimum of 20 dB attenuation, and preferably 30 to 40 dB. You can easily add severaluF of line to line capacitors (or several caps in parallel) to the filter toimprove low frequency Differential Mode attenuation (but this component willnot improve CM attenuation). The voltage rating should be 250 VAC or 600 VDCminimum, and it is best to use a “X” type safety rated cap which is designed tosafely operate on an AC line that has voltage transients. Keep the leads asshort as possible to minimize the lead inductance you add, as this will cause aresonance at higher frequency and the attenuation of the cap rapidly degradesabove that resonance.

As frequency goes up, the attention of the filter usuallygoes up, at least until you get to the 10’s of MHz region where internal resonancesshow up and cause dips in attenuation. Your coil will generate and emit energyat higher frequencies, especially at harmonics of the operating frequency. Atoroid to ground discharge can excite powerful resonances in the MHz and 10’sof MHz region, and these higher frequencies will radiate easier and can causeinterference in other electronics.

In a resistive load the peak current is 1.414 timesthe RMS current, and many filters are designed and rated for a resistive load.   A rectifier feeding a large capacitor loadcan have the peak current (crest factor) 2 to 3 times the RMS current,occasionally higher. When you overcurrent an inductor it will saturate themagnetic core on which it is wound and the inductance almost disappears, andthe attenuation almost disappears. The inductor in this filter is a common modetype which has two windings with the line current thru one, and the returncurrent thru the other, so it should not saturate. But in the real world theseinductors are often made with unequal windings so they will also have somedifferential mode inductance, which means they can saturate. 

It would be safer to have a 30A filter for a 15 Arectifier/capacitor load. The defining parameter in my mind for a Tesla Coil EMI filter isthe low frequency or 100 KHz attenuation. Significant attenuation at 100 KHzmeans large (physical and inductance) inductors and large capacitors. These arelarge and heavy, so as a rule of thumb a filter that is light in weight  doeslittle to attenuate the low frequencies, which are usually the strongest. Makesure that you have several inches of separation between the input and theoutput wires of your filter, as small amounts of capacitance between the wireswill cause very high frequency signals from the dirty side of the filter to leak into the cleanside. 

Almost all filters have line to ground caps which areneeded for common mode attenuation, and these also will inject several ma of 60Hz (50 in Europe) current into the ground connection, so make sure the power lineground wire is connected to the case of the EMI filter. Y type safetycapacitors are designed for line to ground operation, as a shorted Y cap willcause a dangerous situation with line voltage on your chassis. These are ratedat 250 VAC, but are usually made with caps rated at 3000 to 4000 volts toincrease the safety margin. Y caps are usually in the 2nF to 10nF range, aslarger caps will inject too much current into your ground system.

As a good practice I like to add a 20mm (or larger)diameter MOV (Metal Oxide Varistor) to the power section as a line to linecomponent. Use a 150 or 175 VAC part on a 120 VAC line, or a 250 or 275 VACpart on a 208 to 240 VAC power line. Do NOT run these above their rated voltageor they will massively overheat. These act like a big Zener diode and will takea large transient (like a secondary arc to your power section) and clamp itdown to a medium transient. 

It is also a good design practice to add anappropriately rated (voltage, current, and interrupting current) white ceramictube fuse (not a clear glass tube), Littelfuse 314 (fast blow) or 326(slow-blo) or equivalent, to your power section just past your main on/offswitch. The ceramic tube fuses have a sand filling which gives them a much higher interrupting current rating. Fuses have a much faster opening time for a large overload or shortcircuit, and this means the breaker in your building branch circuit will notblow. This will help keep other people in the building happier if some thinggoes wrong in your coil, and will limit the fault current and thus the damageto your system.

I hope that I have added some useful data to theconversation without adding too much confusion, but EMI is a very complexsubject. Do some reading on EMI when you have a chance.  I am a retired EMC engineer so this is thereason for the data dump.  It is nicehave a big parts box, and to now be able to design a circuit without having tobe cost effective in the design.

Don Anderson

Tesla mailing list