D.I.Y. Electronic Control Gear

This article examines the scope for reusing electronic control gear from dead Compact Fluorescent Retrofit lamps. When Compact Fluorescent Retrofit lamps first appeared, they were the cheapest source of Electronic Control Gear for small fluorescent lamps. Indeed, such control gear purpose designed for discrete fluorescent lamps could be hard to find at any price which meant in some cases it was worth buying a new Compact Fluorescent Retrofit lamp just to get the Electronic Control Gear. This is less true today, but the possibility of reusing control gear which still works from an otherwise dead lamp is still an interesting possibility.

Warning

A word of warning is prudent here. The electronic control gear in these lamps is perfectly capable of delivering lethal electric shocks. The control gear internally runs from rectified mains, and DC at this voltage level is particularly dangerous. Unless you are familiar with working with mains voltages and DC levels above mains voltages, you should not attempt to copy any of this experimentation.

A selection of Compact Fluorescent Retrofit lamps. From left to right:
Top row: GE BIAX 20W, Philips Ecotone 9W, GE Genura 23W;
Bottom row: IKEA 4W, IKEA 7W, IKEA 11W, Philips 11W, B&Q 15W, Philips 20W, IKEA 20W, Philips 23W.

These each contain high frequency electronic control gear, which in most cases can be reused after the integrated fluorescent tube itself has died. One exception here is the GE Genura, which is an electrodeless induction lamp - the control gear in this lamp is not suitable for reuse, and in any case, the failure mode of this lamp is usually the control gear itself, since other than the phosphor, there is nothing to wear out in the glass part of the lamp.

This diagram shows the control gear within the base of a Compact Fluorescent retrofit. In this case, the glass tube has been carefully removed and disposed of, and the plastic peeled away from around the control gear.
You can see the high frequency ferrite cored transformer in the foreground, next to one of the switching transistors. A second PCB contains further components, and is joined with 3 interconnecting wires in this particular lamp.

Four short connection wires are visible from the edge of the rear PCB. These are the connections to the fluorescent tube.

The following picture shows the first lamp I drove from the reclaimed control gear out of a dead compact fluorescent retrofit. The lamp being driven is a 13W T5 tube, and the control gear was extracted from an 11W compact fluorescent retrofit. The whole circuit was checked with a true power meter and found to be drawing 15W. I estimate the power dissipation in the control gear to have been about 1W, which means the lamp was probably being slightly overrun at 14W. The control gear was also delivering more power than it was designed to, probably due to a higher tube voltage than the original integral tube it drove. This did not seem to matter, indeed because the control gear is now remote from the lamp, it will be running very much cooler than it was. This lamp was used as an under-cupboard lamp for some 8 years in a kitchen, and the tube has never been replaced in that time, and still works today, although it is no longer installed (neither is any of the rest of that kitchen;). At the time it was constructed, electronic control gear for fluorescent lamps was very expensive, and in this size, and very difficult to find at all.

The following diagram shows a closeup of the control gear fitted into a new box. Disassembling the two circuit boards in order to lay them flat side by side (rather than keeping the back to back as in the original lamp) was more difficult than I thought. The main problem is the fragility of the copper tracks on the PCB - they are much thinner than normal for PCB tracks, and they come unstuck extremely easily whilst being soldered. The other thing that was done was replacement of the original electrolytic capacitor. The original, although still working, was getting hot in use and probably not much longer for this world. This is the component in the control gear least able to survive high temperature in the lamp base for extended periods of time, so this is no surprise. In another conversion, I did have one of these capacitors spew its contents out, so replacement is probably no bad thing. Once the control gear is remoted, a new capacitor should last indefinitely as it won't be running at such a high temperature.

The next project was to refit some downlighters which were designed to use regular 60W GLS filament lamps. Retrofit compact fluorescents were tried but hung too low in the fittings, resulting in excessive glare visible at a wide angle, and the light source being in the wrong position for optimal downward directing by the reflector. The other issue with the downlighters was that the plastic lampholders were falling apart after 20 years due to heat from the GLS lamps, even though the units had only been used with 40W lamps. However, the units were in good decorative order with good reflectors and the ceiling already had the appropriate holes obviously, so it was decided to try converting them to use compact fluorescents.
The compact fluorescent retrofits tried were Philips PL 9W ones. The electronic ballasts from these were to be reused in the conversion, together with 4-pin 10W compact fluorescents (also pictured below). Suitable lamp holders for these were sourced with the same 40mm external thread as the original bayonet cap lamp holders used.

One lesson I learned from doing the original 13W T5 fluorescent lamp is that the circuit boards in the lamp bases of compact fluorescent retrofits are really too fragile to be able to be removed and reconnected, without having to repair a number of broken tracks. This time, I would go for a scheme which didn't require removal of the control gear, but instead used it in-situ in the lamp base, with just the necessary connections brought off to connect to the remote tube. Another factor in this case was that the units would end up out-of-site in the loft, so I didn't want something which might burst into flames. This required they were totally enclosed in metal boxes. Cooling ventilation should not be an issue since the control gear was designed to run in an even more enclosed space in the lamp base, right next to the tube which is a significant source of heat. However, the loft insulation material was moved back from the boxes anyway. In practice, the boxes barely warm up at all when the lamps are operating.

The picture below shows the construction of the ballast box. The box is drilled at one end to take a baynet lamp holder into which the base of the control gear is plugged, both for electrical connection and for mechanical support. 4 core flex is used to interconnect the ballast with the downlighter. In this case, I used screened cable and the screen is clamped under the brass ring holding the lamp holder in place (and serves a secondary purpose of preventing the 4 core flex being accidentally yanked out, damaging the PCB connections). The 4 flex cores are soldered to the connecting wires to the original tube. The soldering is done very quickly to reduce the chance of the heat conducting back, melting the PCB tack glue and breaking the track. Enough slack is provided in the 4 cables to enable easy fitting and removal without putting any strain on the delicate PCB connections. The mains supply used contains an earth wire and the brass lamp holder has a suitable connection terminal for it.

The results from this conversion have been very successful. The 3 fittings converted produce high quality light on the work area under them, and no one would guess to look at them from within the room that they had been designed quite differently from the way they are now being used. The capacitor on one unit failed as previously mentioned, although the unit carried on operating from unsmoothed DC with the lamp noticeably dimmer. I swapped it out for another base, and will replace the capacitor in due course to give me another working spare in case any of the others go the same way. The metal box contained the capacitor debris, which must have been ejected with some force from one end of the capacitor looking at the way it was plastered over the inside of the box.