Posted on Leave a comment

12v CFL Lamp Failure Analysis

On the boat I have installed custom LED lighting almost everywhere, but we still use CFL bulbs in a standing lamp since they have a wide light angle, and brightness for the size.

I bought a couple of 12v CFLs from China, and the first of these has been running for over a year pretty much constantly without issue. However, recently it stopped working altogether.

12v CFL
12v CFL

Here’s the lamp, exactly the same as the 240v mains versions, except for the design of the electronic ballast in the base. As can be seen here, the heat from the ballast has degraded the plastic of the base & it’s cracked. The tube itself is still perfectly fine, there are no dark spots around the ends caused by the electrodes sputtering over time.

Ballast
Ballast

Here’s the ballast inside the bottom of the lamp, a simple 2-transistor oscillator & transformer. The board has obviously got a bit warm, it’s very discoloured!

Failed Wiring
Failed Wiring

The failure mode in this case was cooked wiring to the screw base. The insulation is completely crispy!

Direct Supply
Direct Supply

On connection direct to a 12v supply, the lamp pops into life again! Current draw at 13.8v is 1.5A, giving a power consumption of 20.7W. Most of this energy is obviously being dissipated as heat in the ballast & the tube itself.

Ballast PCB
Ballast PCB

Here’s the ballast PCB removed from the case. It’s been getting very warm indeed, and the series capacitor on the left has actually cracked! It’s supposed to be 2.2nF, but it reads a bit high at 3nF. It’s a good thing there are no electrolytics in this unit, as they would have exploded long ago. There’s a choke on the DC input, probably to stop RFI, but it doesn’t have much effect.

Supply Waveform
Supply Waveform

Here’s the waveform coming from the supply, a pretty crusty sinewave at 71.4kHz. The voltage at the tube is much higher than I expected while running, at 428v.

RFI
RFI

Holding the scope probe a good 12″ away from the running bulb produces this trace, which is being emitted as RFI. There’s virtually no filtering or shielding in this bulb so this is inevitable.