Because the LEDs are expecting AC current not DC.

I’m getting into the weeds here, but the actual LED bits themselves are DC components and each typically runs on 1 to 1.5 volts. It’s the flow of current that light them. How they are arranged, series or parallel or both inside the bulb assembly doesn’t matter to us. What matters is that inside the base is a miniaturized switching power supply (my brain keeps wanting to call it a regulator, my apologies) which expects to see relatively unregulated, unfiltered AC line voltage within a certain voltage range. These tiny devices are very efficient. I’ll go out on a limb and say that if the bulbs would light in series, it seems that you’d be increasing the current through the front end of the power supply of the previous bulb in the chain risking burnout of the initial components. I could be wrong but I don’t think you can daisy chain regular LED bulbs in series. I certainly wouldn’t expect the same result as you would with simple incandescent bulbs.

IIRC, in larger commercial lighting applications, LEDs assemblies are arranged in a series/parallel array and that array is powered by a controller. That system is even more efficient than the individual bulbs we use. But I digress….

Edison base LEDs are available that operate on DC. Here is an example of one that operates on 10 to 80 VDC. Note that it recommends against using in enclosed fixtures.

https://wattalight.com/collections/led- ... ht-e26-32v

A ballast resistor can help, but the problem is that the resistor must consume a significantly high amount of power compared the primary load to bring balance to the circuit. Even a 5 watt dissipation would be a lot of heat that would require a large resistor with good ventilation to stay cool.

Rewiring and adding converters would work, but in my mind, making as few changes as possible is the ideal plan.

Many LED bulbs will be just fine on DC power. Others won't light at all. The current though series devices is ALWAYS constant at any point in the circuit.

I'm tempted to build a 600vdc test board and try. I suspect at least some of the bulbs on the market will put on a light show and then go BANG. At 600vdc, you could start and sustain an arc over a significant distance.....

Aren't these bulbs still being made in other parts of the world, that haven't gone nuts?

In my electrical engineer's opinion, trying to use a string of consumer grade 120VAC LEDs, 5 in series, on a 600VDC supply circuit, is lunacy. It may appear to work, but there are multiple failure modes that could be problematic.

What is inside the typical retrofit LED bulb is a tiny switch-mode power supply which delivers constant current to a string of individual LEDs in series. LEDs have essentially a constant voltage drop and their light output intensity varies with current, within the safe operating range of the device. The junction voltage tends to be around 2 volts. In this regard, LEDs are similar to neon indicator bulbs, which have a fairly small voltage drop range of 70-90 volts over a wide range of currents.

So you could use a dumb ballast resistor in series with with say 30 LED elements and produce a device with 50% efficiency.

If you have access to both the high and the low side of the circuit, and someplace to stick a small electronics module, you could have a driver that feeds a fairly low voltage series string of 5 LED "bulbs" each of which has a number of LEDs in series internally, with no electronics in the base at all. But in typical wiring practice for trolley and interurban cars, the +600 feed is at one end of the string and the ground is at the other.

Would it be a big pita to rewire the circuit so the existing wiring (which becomes all positive) went to one terminal of the bulb socket, new negatives to nearby metal frame ground and supply the positive from a 600VDC to 12V or 24V or even 120VDC switchmode power supply for new led bulbs? The new power supply will be quite small and all the leds in parallel.

In the case of many old trolley cars with wooden interiors, it is a real PITA to do any modifications to wiring for lights.

Old trolley cars did not use conduit for those wires. Often the wires were installed in the framework which would hold the interior paneling by simply boring holes through framing members, or possibly in channels routed into framing members, and then the nice interior paneling was installed with just a hole for the two wires for a light socket. The only way to repair or modify that wiring is to remove the paneling, make the changes, and then re-install the paneling without damaging it. That is a PITA.

Also, in wooden trolley cars, there are no nearby grounds easily accessible.

Exactly. The typical streetcar lighting circuit is a single wire. It starts at the fuse or snap switch, runs to the first lamp socket, then the next wire runs from that socket to the next socket, and so on, the last wire runs to a ground. On a wooden car, often that means the wire ducks down inside a bulkhead or a hollow chase in the window post and wansocting and under the car to find a metal structural member.

So, you do not have two parallel wires running in the circuit. If you did, this discussion would be easy.

Inside an LED replacement light "bulb" there is a tiny switchmode power supply. The electronics topology of that varies. The best (hence most expensive) topology is the same as a typical PC power supply: at the 120VAC input, you have a bridge rectifier into a filter capacitor to the neutral. Then a switching device such as IGBT from the cap into the primary of a transformer, returning to neutral. And an isolated rectifier and filter on the secondary side feeding the DC output. In that topology, you could put all the power supplies in series: Assuming all the capacitors have equal nominal values, the voltages across them will divide more or less equally, and each bulb will function correctly. One key here is a full-wave rectifier on the input, not half-wave, because these series lighting circuits were often not wired with regard for polarity.

Many LED lamps that I've taken apart have a much cheaper topology without the isolating transformer. You have just the input voltage going right to the LED string through a switched element, often a triac or SCR, and fairly simply closed-loop feedback control of the duty cycle to obtain constant current. If those topologies work in series, it is by accident.

Also of great concern would be what happens during a fault condition, or when the circuit voltage momentarily surges above nominal voltage, such as inductive kick-back, or secondary surges induced by lightning.

Has anybody thought about doing any testing. Like putting 5 of them on a board in series with 600 volts and see what happens. Measure the current, measure the voltage drop on each bulb. Raise the end of the board up two or three feet and drop it. Put 600 volts across one bulb and see how bad it blows up or does it just go out. Test both cheap and better made bulbs and mix the types up on the board. Of course if you ask an electrical engineer or a lawyer, you know what the answer will be.

It might be a little bit of Chicken Little on my part but I wouldn’t use them without adding a second wire to run them in parallel and adding a regulated converter. My theory is that the first LED bulb in the string would burnout. If it goes open, no big deal. If they all go short, then you have a potential fire hazard with the wire going straight to ground. Shorting or a arcing over is probably very highly unlikely but can you be sure? Incandescents always go open and would kill the entire string like a fuse.

I think another problem may be that the connections to 600+ and ground are essentially floating connections. Incandescent bulbs are pretty forgiving in that environment.

If you do an experiment, use a breaker to guard against any potentially spectacular failures. It may be fun to watch or be totally anticlimactic.

About these various concerns about the electronics in LED bulbs: where are the electronics in this bulb?

https://www.greenline.dk/osram-vintage-1906-led-standard-paere-e27-25w-6133

May have answered my question, this author disassembled a filament bulb and made a diagram of the circuit.

http://blog.spitzenpfeil.org/wordpress/2015/04/06/spontaneous-led-bulb-teardown-filamented-led-light-bulb/

that lamp should NEVER be used on DC let alone 600 volts. it might try to glow but would be followed by at least one or perhaps a series of bangs.

Seems to me that you could have these lamps custom made just like any other lamp, and then have the electronics modified to support operation in series on DC.

You could also have the lamps made without any electronics at all, and then modify the power at the source.

On high voltage Direct Current, incandescents can arc when they fail. That's why there are special Street Railway bulbs, either with no gas (vacuum) or Arc Resist if over 100 Watts. 30 Volt Street Railway bulbs can be filled with gas if they are designed to short when they fail and arc.