The capacitor is wired in parallel with the inductor to compesate the lag that takes up the power loss through the circuit. I just had a class lecture and lab from Circuit II yesterday which I am attending at the college.

Ultra efficient design! Do you have any photos of how that internal starter is mounted in the case? I have two Advance ballasts for 1 F30T8 lamp that are preheat with internal starters in my bug zapper. Does yours here drive the lamp at full power/brightness?

I agree John. This is probably the most efficient method of driving a 22 watt circline that I've seen so far. BTW, in your case, the best thing to do would probably be to use a setup with a single F30T8 lamp and a glow starter and test sets of 2 wire leads. If there is one white lead, that is probably one of the ones you will want to go to the lamp. However, the only way to really know is to test them one by one, being careful to assess which ones will obviously not work. In a smaller choke like ballast such as this one, the glowbottle may be easily seen when the ballast is opened up, or it may be hidden under the tar. It could be either way.

Mine are working fine now and would be a PITA to get out of the bug zapper! I am actually surprised they used that design, because in my experience, bug zappers are terribly neglected when it comes to relamping. They have lamps left in @EOL often for long periods of time, especially if only one lamp fails. They won't be noticed till the whole thing goes dark. Mine works perfect, I had just never seen another ballast with the internal starter. If mine ever fails to start, I will sure open it up!

My Kill a Watt meter finally arrived and I tested a number of ballasts but this one probably had the most astonishing results I have ever seen. This ballast pulled 0.18 amps with an older US made 22 watt circline lamp (Ace branded cool white made by GE) and 0.21 amps with a newer 22 watt circline lamp made in Thailand (Philips made daylight) and the wattage results were even more strange. The older US made lamp pulled about 13 watts on the Kill a Watt meter while the newer Thai made lamp pulls about 17 watts, yet the lamp did not appear underdriven in either case, as they both pulled a pretty decent brightness. However, neither of them passed the bright neon starter test. That BTW is my system of testing whether or not a lamp is underdriven using a bright neon starter. If the lamp is underdriven, the starter will have a dim glow after the lamp is lit, and in both cases, we got the dim glow. Another interesting thing is that the presence of a capacitor lowers the WATTAGE drawn by the entire setup as well. I tried this with a normal power factor 22 watt preheat circline fixture and then added a capacitor in parallel with the line and neutral leads and saw that the wattage also decreased, albeit with no change in lamp brightness. Now that is pretty cool!

Where are you placing your capacitor at? A series capacitor or a capacitor after the ballast can affect the wattage, but one parallel to the input would have no effect.

It will reduce VA, but not W.

Interesting about the underdriven lamps and the neon starter! Is it doing that (glowing) because the arc flowing through the tube is not at the designed "power"? I know different starters act different ways. Some will allow a lamp to rectify without flashing @EOL (with the starter glowing dimly as you describe)and some will violently flash the lamp until they stick closed.

If cap is in series, it depends on cap value as to the overall lamp wattage. A 4 ufd cap will run lamp at higher wattage than a 3.5 ufd. But be careful...a too big of cap will destroy lamp and/or ballast rather quickly so only increase in small increments and monitor ballast temperature.

That is the funny thing. In this case, I placed the cap in parallel with the line and neutral power leads. It was a 3.5 uF cap with a discharge resistor attached. Although it would have no effect on a ballast with an autotransformer, apparently this setup does effect simple reactor ballasts in more ways than just power factor. I also ran a bunch of fixtures with preheat chokes through the kill a watt meter. I have made the discovery that not only do different brands/makes drive the same lamp at different wattages and amperages but that a simple reactor ballast actually uses less watts if you use the highest wattage lamp it is rated for. I was told by Rick Delair aka Ricksbulbs about simple reactors running more cool if you use the highest wattage bulb they are rated for. On the same make of ballast that drove a F15T12 at 18 watts, pulls 17 watts from a F20T12. The F20T12 also pulled a lower amperage than the F15T12, hence why the ballast runs cooler.

As for cap values, I'm aware of things that can go wrong from experimentation. Too large a cap value in series with the lamp will make the ends glow bright and the ballast buzz loudly. Too large a cap value in parallel with the line and neutral leads on a simple reactor will actually pull more amps and watts than without it, thus actually making the ballast less efficient than before. Too low a cap value in series with the other lamp and the lamp won't light or will have problems lighting. I haven't tried too low a cap value in parallel with the line and neutral leads yet largely cause I worry about the cap getting destroyed. For 14-20 watt fluoros, I think 2.5-5 uF is a safe range. I might try some 2 uF caps as well and see how well they work.

If the capacitor is placed BEFORE the ballast, it does not affect the ballast at all. It is just like putting a capacitor across anywhere along the power line. Are you certain that you're not reading the VA value? The Kill-A-Watt toggles between VA and W by the same button.

Pure reactor ballasts have a poor regulation. The constant-wattage type HID ballasts as well as HPF rapid start ballasts have a capacitor in series with the lamp. The capacitor provides improved variation when there is a variation in load and source power.

The lamp power swing in percent is less than or about equal to the variation in line voltage for the series capacitor type. Significantly more swing than voltage variation for pure inductor. (you can test this out using a variac)

I just uploaded a datasheet for choke ballast meant for F20T12, F15T12 and such. It overdrives the F15T12, and underdrives F20T12 relative to its respective ANSI reference ballast.

F15T12: 21.2W, 113% BF: 800lm x 1.13 = 904
F20T12: 20.6W, 93% BF: 1200lm x 0.93 = 1116 * 53/60 = 982

53/60 cancels out the lamp efficacy difference. Here, you can see that F20T12 will still get more output even if its efficacy was reduced to that of F15T12, which means that despite the same wattage, higher lamp voltage reduces the ballast loss.

http://www.lighting-gallery.net/gallery/albums/userpics/10556/43643534612.pdf

I have noticed that it has no effect on ballasts with autotransformers but in the case of this particular preheat choke, since its a simple reactor it did have an effect. They were in parallel with the line and neutral leads and Im certain that it was watts I was reading because I checked watts, amps, VA and power factor. In this particular ballast shown in the picture, the capacitor was also wired in parallel with the line and neutral leads before the ballast, as seen in this shot of how the ballast was wired straight from the factory. I suspect that for simple reactor and for HX autotransformer ballasts, the results are different. However, I did check both watts and VA so I'm pretty sure. Plus I've also seen a fixture that came wired the same way as seen here.

I attempted capacitor across power downstream of power meter with F15T8 pre-heat choke ballasted light as a load. It affected the VA and PF, but wattage difference was not duplicated.