What is the globe size ? If they are fairly big compared to the lamps, they can contain a bit more than 6..8W heat source inside without temps going up way too much. You might be able to find something on the order of 10W, or use double sockets and 2 x 6W for 12W



If there is fair budget to spend (more than cost of retrofit lamps), consider doing better performing conversion : PL-Ts with PUSH controlled dimmable ballasts (pushbutton switched 120V line controls on/off and dimming. No dimmers used - everything is inside the ballasts)

The ballasts can be installed at the ceiling height (thats on the limit of wire length that Electronic ballast can handle, but i'd try with one first and i think they'd not show any problems). Install the ballast in some sort of decorative box above the stick holding the lantern, then inside the lantern is only the lamp itself - Then you can up the wattage well above 10W without overheating any electronics

The closest PL-Ts that put out above 1300Lm are 1800Lm 26W and 2400Lm 32W. They use the same base and ballast (atleast with some ballasts) so you'd be able to switch between them in the future

Additional benefit is that the PUSH switch wiring can be wired to multiple switches in parallel (really parallel, not 3way wiring), that is, the lighting can be switched or dimmed from multiple locations

The ballasts can be installed at the ceiling height (thats on the limit of wire length that Electronic ballast can handle, but i'd try with one first and i think they'd not show any problems). Install the ballast in some sort of decorative box above the stick holding the lantern, then inside the lantern is only the lamp itself - Then you can up the wattage well above 10W without overheating any electronics

Regadring the wire lengths, you should never exceed the limit the ballasts are rated for. It may seem to work, but it may radiate the RF energy way over the allowed limit (and disturb radio reception all around). In fact it would mean the ballast certification won't be valid anymore.
And generally the length limit is barely covering the needed length inside the common fixtures (lamps just beside the ballasts).

But I think as well, the result you are looking for will need purpose made lights (fixtures made specifically for the given light source - would that be either fluorescent or LED), I doubt any retrofit would work well.

With a 5 core cable (4 lamp pins + PE) all the currents that cancel eah other's fields are in the same cable. Then it is also inside a (supposedly ferrous metallic) tube. Is this going to emit any much RF out ?

With a 5 core cable (4 lamp pins + PE) all the currents that cancel eah other's fields are in the same cable. Then it is also inside a (supposedly ferrous metallic) tube. Is this going to emit any much RF out ?

Yes, e.g. via the input power wires. The RF is generated inside of the ballast, but flows via stray capacitance to the (supposedly) grounded tube. If that capacitance is large, so the current becomes large as well. It has to return somehow.
That current has to return somehow and that happens via the phase and neutral wire inputs (as that is, where the RF source is connected), so it has to go there from the Ground (the tube connection). That means it will force the disturbance between the Power wires and the Ground (PE). Because these can not be directly connected together near the ballast, it means an RF power is carried by the input wires out from the ballast.

Normally there is supression for such currents, but the point is, this suppression is designed to reduce that emission for the rated capacitive load, but there is not that much margin (cost, reliability and a swell ground leakage safety specifications limit the achievable filtering suppression). If you exceed the currents the input filter is designed for, the radiation to the input then exceeds the limits.

But there are tons of other mechanisms, many not that visible. Controlling EMC is to a big extend a ghost working: You do something, the results change, but you may explain only part of it. The reason is, there are too many variables in play...


Therefore legally when you violate the conditions for which the certification is granted, it means the certification gets voided. So if you would like to use it in that way (you think you have solved the problem by other measures), you would have to let the thing measured and acquire the required certification for your setup. That is usually the case for larger equipment: You violate many EMC or safety related aspects of the individual components, but you cover them by the final assembly (all machine is covered by grounded metal,...). With such machines the certification would be necessary anyway, so the internal installation is made according to the actual needs (of course, that includes the safety concept, emissions and so on) of that machine design and not the general rules.

I don't think a church lighting is something worth that approach...