Sunday, March 17, 2013

Tesla coil completion & "first light" tests

During the last week, I finished 'My Last Tesla Coil', at least insofar as it can be operated and is not an ugly mess to wire up. I was moved to push hard and get it working before Saturday March 23 (six days from today) which is the scheduled date for two grand opening parties for two businesses run by friends of mine.

The owner of The Concoctory wants the coil to be present and running as an attention-getter near the store.

Down the street and around the corner is a great little café called Cafe Crescendo, who are also opening that day. I gather that between them, they are trying to get the other local businesses to join in a sort of block party to draw attention further south to all of the existing and new businesses which are opening in that (newly renovated) south Broadway neighborhood.

Clever, fun, smart people are behind both operations and I hope I can add some sizzle and excitement to their opening day.

Yesterday, I set up the coil in the parking lot of Denhac, Denver's very own local hackerspace, and fired it up. After several extra trips to various places for forgotten parts and to put gas in the generator, and then perhaps fifteen minutes of the usual first-time tuning of the primary tap, adjusting the safety gaps in the power panel, and adjusting the main spark gap (an air-blast or "sucker" gap, the first of at least three gaps to be built for this coil) I obtained some reasonable results; approximately 36" streamers from the toroid.

I was expecting and hoping for closer to 60" streamers, however two problems rapidly became apparent during yesterday's tests:

1) the alleged "3,000 watt" generator (the only one readily available at no cost to me) is either failing to deliver its full rated power; or perhaps it is getting some RFI or spikes from the coil, causing its regulator to reduce the RPM; or -- and this seems the most likely cause to me -- the reactive load is causing a low power factor, resulting in more current being drawn from the generator than necessary and less "real power" being delivered to the load.

2) the spark gap was not quenching reliably (or at all). I don't have E/H field antennas for an oscilloscope which would confirm that theory, but I have enough experience with Tesla coils that I know a badly operating spark gap when I see and hear one. This gap (described here) uses an air blast to encourage the arc to extinguish (quenching the primary circuit) immediately after it has established. This is the desired "opening switch" operation of a spark gap, and it is much harder to achieve than the initial "closing switch" operation which precedes it in each firing cycle.

[aside: "opening switch" is a term of pulsed power jargon which refers to
switches -- there are a variety of methods -- which are designed to
interrupt current flowing in a circuit as quickly as possible. Opening
switches are frequently used to transfer energy between stages in a pulsed
power system.

A brief pulse of current is made to flow from a charged energy store having a
relatively long time constant into a following stage having a shorter time
constant, by way of passing electric current through an already-closed opening
switch. When most of the energy has been transferred into the next stage, the
switch is opened, causing the current to cease very abruptly. This "traps"
the transferred energy in the next stage, whether the energy is being stored
in an electric field (capacitance) or a magnetic field (inductance).

When the energy is again transferred out of this second, "faster" stage into
either another power amplification stage or the load, the smaller reactance of that stage causes the pulse to be delivered in a shorter time. Because the same amount of energy is now being transferred in a shorter period of time,
the peak power increases proportionately, obeying conservation of energy.]

While I have experimented with air blast gaps in the past, this gap was my first experience with the "sucker" design, which pulls air into holes in the electrodes, creating an axial shearing flow across the faces of the gap. While I was tuning up the coil, I noticed that the gap operation was not being affected much by the amount of air flow.

The gap did not work well at the original electrode spacing. With a conventional gap, if you want to increase the voltage at which the gap fires, you open the gap spacing. In a sucker gap, once the gap spacing exceeds the cross sectional area of the gas ports in the electrodes, the air speed across the faces drops dramatically, which spoils the quenching action.

The solution would seem to be opening the holes in the electrodes a bit (which I did after yesterday's test run) and reducing the gap distance, relying on the air flow to quench the gap rather than a long arc distance, which merely increases resistance in the gap and thus wastes power badly.

Some finessing may be required on one or both electrode faces as well.

The coil tuned up (the system resonates around 105 KHz) at 9 turns on the primary. I was shooting for 10 turns and there are 14 turns available, so that was agreeable. It would not hurt to have a slightly lower capacitance in the primary tank capacitor, which would allow (require) a few more turns on the primary. Counter-intuitively (because turns ratio in a Tesla coil doesn't matter as much as one might expect) increasing inductance in the primary tank circuit acts to reduce the peak current through the gap. Because gap losses increase as the square of the current, a small reduction in primary circulating current results in a large reduction in lost power. This improvement in efficiency completely swamps the small change in output which might otherwise result from reducing turns ratio.

So, before next weekend, I've got to find a slightly beefier generator. It's probably that some PFC capacitors on the line side of the HV transformer will reduce the amount of current being drawn from the 120V supply, but that's unlikely to happen in the next six days.

I also need a better-working spark gap. I have already modified the existing spark gap, but it will take a little experimentation before I can confirm whether the changes make enough of a difference.

It might actually be possible to build a Richard Quick -style multi-gap before next weekend, especially since I started work on one over a year ago; I have all the electrodes fabricated (but not yet drilled) and I probably have a suitable muffin fan in stock...

No comments: