If you suffice with step dimming (and you do not need to have it continuous), a series capacitors with the ballast inductor would do the trick. But you may need some momentary switch for the initial arc ignition - I doubt the 120V would be enough to strike it, even with hot electrodes, but you may try. For a "100%-50%-25%-14%" you need a "3-way" switch, one 2uF and one 1.2uF capacitor (both rated for at least 250VAC). The switch has one lead marked as "Line in", one as "Low" and one as "High" and two contacts, one between the LineIn and High (On at "medium" and "high" positions), one between LineIn and Low "ON at "low" and "high" positions. You probably know these, but I wanted to define it's terminal names I would be using in this descriptions, so you may link that to your switch. So connect the 2uF capacitor between "LineIn" and "Low", the 1.2uF one between the "LineIn" and "High". Then connect this contraption in seris with the ballast choke via the "High" and "Low" wires. What it essentially means: When both contacts are ON (at "High" switch position), both capacitors are bypassed, so the lamp is ballasted just by the series choke. When at "Medium", the 2uF capacitor becomes in series with the choke (the 1.2uF bypassed). If my calculations were correct, the 2uF should have about 3x higher reactance than the choke, so that means the total ballasting impedance would be 2x to the original ballast, hence the 1/2 current.
When the 3-way switch would be at "Low", the 1.2uF cap becomes in series with the choke, (the 2uF bypassed). That means capacitive reactance about 5x the choke reactance, so total it is 4x to the choke alone, so hence the 1/4 of the rated lamp current.
And when the 3-way would be in "Off", both capacitors would be in series, so forming 0.75uF total. That means 8x higher reactance than the choke, so together it means 7x higher ballast impedance, so 1/7 of the rated lamp current.
You may play with the capacitor values, but I would keep the ratio of the brightness levels about factor of two - that I would see as quite a good compromise between the brightness differences (don't forget the eye is logarithmic, therefore the ratios should be about the same) and the total control range.
|