How can you use ballast-powered flouresecent lamp outputs to power LEDs?

Question

We have a 2G11-socket 36W germicidal UVC lamp in our air ducts (Honeywell UC36W1006), and since UV-C LEDs are readily available, I thought maybe I could make a PCB with a bunch of UV-C LEDs on it as a longer-lasting lower-power replacement.

This study shows that a dose of 50mJ/cm2 will kill 90% of bacteria and viruses (1 log reduction), so very few LEDs would be required in the plenum (assuming a 30-second exposure).

I could just run DC to the new boards, but wondered if it would be feasible to use the existing twist-lock enclosures with a pin-compatible connection to the existing fluorescent driver outputs. The PCB would plug directly into the blue socket below in place of the fluorescent lamp:

I've looked around quite a bit, and while there are many places selling 2G11 bulb replacements, I can't find any technical documentation on the pinout. Here's what I've found so far:

• It looks like 2G11 is also implemented as "PL-L 4P" (Philips Linear Compact Fluorescent Lamps).
• This video shows lighting it with a ballast that has a 430V output.

There are 4 pins, two pairs to illuminate two elements. The tube is a loop at the far end (not pictured) so (I think) the left and right elements would be the same as those at each end of a linear fluorescent tube:

Question:

How can I use those 4 pins to get ~6V DC to illuminate UV-C LEDs?

Others have made 2G11 LED lamps (but not UV), so clearly it can be done...but what components are involved in stepping down the lamp input to a DC level for LEDs?

Answer 1

Before you go further, consider the cost to replace the Hg-vapor lamp with LED's. The data sheet for LED in your link states output power is ~4 to 7 mW. Specifications for a similar 36W lamp states its radiant output at 12 W, and this Osram lamp claims 10.8 W. To get the same amount of UV-C radiation would require 1,500 to 3,000 LED's.

That said, there might be a similar lamp that fits that socket at *far less than the OEM price.

As for using the ballast to get DC, that would not be practical. In the days of the magnetic ballast, some (newer) had three windings on the ballast transformer: two for filament power, which might be 12 VAC or so, and a third for ignition voltage, initially ~1,000 VAC, which drops to ~120 V on each arc (well, glow discharge) ignition. Others used a single resonant inductor/capacitor circuit to get those voltages. Your simple "adapter" would have to accommodate those voltage regimes.

However, an electronic ballast is far more complex, with circuits to switch voltage to filaments and across the glow discharge. These voltages would be continually changing.

It would be simpler and far more efficient to tap into the AC mains before the ballast.