The Lamp Room

What's the rule with this?
Does the bulb have to be the same wattage, more or less or is there a curve difference in something as they go up and down?
Is there any material on this anyone's done?

You are building a series circuit with 2 "resistors", both of which are non linear, each in its own way

If it would be normal (constant Resistance) resistors, they would act according to Ohms law : Voltage across the resistor = Current * Resistance. Change the Voltage by some factor and the current will change accordingly

When connecting the resistors in series, the voltage splits proportionally to the resistance (as can be evident if you write down the equations for the current in the series circuit, since it must be the same in the 2 series connected resistors)

Discharge lamps (when the dischrage is allready struck) behave approximately as "constant Voltage" devices : The Voltage across the discharge will stay the same regardless of the current



Once you have your Voc (say 230V) and the arc V of the lamp of choice (103V for 40W T12), the rest of voltage (230 - 103 = 127V) will drop on your ballast resistor

Some notes :

1. It isn't really right to do what i did here. The voltage across the lamp is squarewave (or in our case here, something like discontinuous squarewave due to longer off times at zero crossings), you can't just add/subtract that with a sine wave RMS voltage. But it is close enough for a first estimate...

2. The calculation is different for resistive and magnetic ballasts :

Resistive ballasts - Assume that there is no Phase shift (also not true, but nevermind it) : Vballast = Voc - Varc

Magnetic ballasts - Assume that the Phase shift is 90deg : Vballast = sqrt( Voc^2 - Varc^2 )



Now, we have to take a ballast resistor that will conduct the desired current (430mA in 40W T12 example) at the voltage it will be getting

127V / 0.43A = 295 Ohm



If we want to use an Incandescent lamp as the resistor, we have to take into account also the behavior of the lamp :

The filament resistance is fairly low when cold and goes up by several times when the lamp is up to temperature (the effect is immediate, as fast as the filament heats up or cools down, it is not delayed to seconds or more)

So : Take for example an 100W 230V lamp

At 230V, we can calculate the resistance of the lamp :

100W / 230V = 0.434A
230V / 0.434A = 530 Ohm

At 0V (cold), we can measure the lamp with a multimeter. So the max current that the lamp could pass is

230V / (measured resistance)

We know that at voltages inbetween the resistance will be inbetween. There probably is an approximate formula to calculate it which is used by lamp manufacturers, but i don't know it

Actually, it makes sense to think of it in current, not going to the resistance figures at all. Which gives some more insight :



We know that 100W lamp at 230V will take 0.434A (near perfect for a 40W T12... But the lamp won't be getting 230V, so the current won't be 0.434A)

If the lamp would be a constant Resistance device, then at 127V, it would take proportionally :

127/230 * 0.434 = 0.240A

However, since the filament is colder, it's resistance is less and therefore current will be more than 0.240A

So without knowing much more, we can expect that 40W T12 + 100W GLS in series on 230V will provide somewhere between 0.240A and 0.430A (and most likely well around the middle between those values, and not near the ends)

This is close enough estimate to take the lamps and try.. And then experiment from there



Take into account that during starting/preheating, near full Voc is applied to the ballast lamp. So for example, 130V GLS lamp cannot be used as a ballast (at 127V), since it will be getting something near 220V during preheating

Also, with Resistive ballast, it is generally a rule that Voc must be atleast 2x lamp arc V (not precise rule), and that differnt gas fills will act differently around the zero crossing - So what works for 40W T12 won't necessarily work for 36W T8

I'm printing that out!
Gonna sort through my tubes and have a think about combinations.