I've been meaning to put this post together for a while and I just finished off doing some final testing this evening to conclude things.
One of the places I help out is a trolley museum in Baltimore. Sometime between 2019 and late 2021 one side of the interior lights inside of a 1948 Philadelphia streetcar went out and nobody was able to get them working again by swapping bulbs, changing fuses etc. I went down and visited a number of friends in November of 2021 three days after the border finally reopened and one of my friends and I visited the trolley museum. They'd asked us to dig into the dead lights since nobody was getting anywhere with it.
Here's a quick rundown on how the two main interior lighting circuits work: They each are energized off the 600 V traction power through a fuse and a switch at the front of the car and are composed of a series string of 32 V, 1.6 A street railway bulbs that start behind the destination sign, run down one on each side of the ceiling to the back, where they're grounded through a resistor sized appropriately to have it all add up to the nominal 600 volts. To avoid the series string of incandescent light bulbs pitfall of if one goes out, they all go out, the shell of each socket is sprung so that if a bulb isn't present, it connects to the tab at the bottom and the circuit has continuity and the remaining bulbs stay on. The bulbs themselves have a self shorting device built into the base so that if the filament goes open, the bulb short circuits so the rest of them continue to be lit and it's obvious which one needs to be replaced. This fault was interesting right from the start because given all this, it's unusual to have one whole string out unless the fuse has blown or someone has installed a regular bulb somewhere instead of the self shorting type and it's burned out and gone open and taken out the whole string.
We had all the tools and a good amount of the portable test equipment in my truck and we started digging into this, and it got messy quickly. The first thing was a disagreement in approach. My preference was to use a multimeter and take out a bulb about halfway down, switch the power on and see if 600 volts was present. If yes, the fault would be closer to the back. If not, we'd have to move towards the front to find the break. My friend wanted to ohm the circuit with the power off. That would be safer, and we did have the working circuit on the other side of the car to compare to, but we were getting wild readings. Low values that were plausible mixed with high resistance but not open circuit that made no sense - until June of 2023.
7.6 ohms is reasonable.
U1272A wedged into a standee window for hands-free use.
Clamp on the power feed to the working set of lights and DMM on the fuse holder in order to establish the parameters of the working side.
1.6 A at 578 V is pretty close to what we'd expect for a string of 1.6 A bulbs that's put together to operate on a nominal 600 V supply.
U1115A remote logging display with their readouts on the dashboard also good to tell at a glance if the dead side is energized or not by moving the DMM over to the fuse holder for that one. The OL was from the U1272A that we were using to ohm the circuit out.
One of the things we found in the 2021 session was that one bulb holder and bulb in the destination sign at the very top of the 600 V circuit had poor contact and arced which chewed up the base of the bulb. We were able to burnish the tab at the bottom of the socket and salvage that but the bulb's base was damaged pretty badly and had gone open.
Then, when while we were trying to get to the bottom of why it still wasn't working and trying to figure out why we were seeing such strange resistance measurements in places, we discovered this in the compartment under the very rear seat:
Sometime while the streetcar was still in Philadelphia, the resistors used to make up the balance of the voltage divider so the whole thing adds up to 600 V had been replaced by additional lightbulbs wired in series. Worse, the set of three alligator clipped to the fuse holder being used to hold these popped off. That would cause an open or potentially a very bad, intermittent ground path through the chassis of the car. Luckily this is at the ground end, not the 600 V end of the string so no arc welding took place.
We ran out of time and I didn't get back for another visit until June this year. My friend and I resolved to get this sorted out and we went through it this time applying the 600 V and looking for voltage to disappear in order to find the break in the circuit. We got the test equipment and tools out of my truck and started working on it from the top of the circuit in the destination sign, beginning with the burned up bulb and socket pair we'd identified in 2021. Then we started dropping open lampshades and pulling bulbs and checking for 600V:
All of a sudden, with one of the bulbs out, the string came on minus a couple of burned out bulbs. Turn it back off, put the bulb back in, we got 3 V further down and no lights. This turned into a major headscratcher. 3V? Nothing? Take the bulb out and the functioning parts of the string came back up. At this point, I was standing in front of a light that worked and I had an idea: this was an example of a known good socket and a known good bulb that we could take the cover off this one and started cycling all the bulbs that weren't lit through it one by one and sorted them into good and bad bulbs. Then, take one of the known good bulbs and go through all the remaining sockets and identify any more damaged sockets.
The superdud dead bulb that took the whole string out wherever you put it in was the cause of the strange high but not open ohms readings we were getting in 2021 and the 3V as measured with the high impedance DMM when we were checking for voltage loss this year. The self shorting mechanism didn't short the bulb out the way it should have but it didn't fail completely open either and was the cause of the high impedance that was enough to break the circuit but had enough continuity to give strange ohms results.
Then, we had another damaged socket partway down the interior of the car where none of the known good bulbs would light up even though the rest of the string was on. This implied the socket itself was remaining shorted at all times, even when a bulb was installed. I figured this should be visible and possibly repairable so we cut the power and disassembled that light fixture:
Notice one of the brass screws is missing? It had worked its way loose and was shorting out the socket bypassing the light bulb that was in it. It was hanging down when I removed the bulb and fell out completely when the two of us disassembled the fixture to begin repairing it. Luckily we were able to find it on the floor and it hadn't rolled away and disappeared somewhere.
The culprit screw is about to be reinstalled. After we got that socket fixed up and the fixture reassembled, we replaced the burnt out bulbs with spares, reassembled everything and tested:
Full interior lighting and both sides of the destination sign lit up for the first time in years. The handful of lights in the middle of the ceiling are all 32 V
regular bulbs wired in parallel along with the headlight and energized off the streetcar's low voltage power supply that's backed up by battery. You do
not want a self shorting bulb anywhere in that circuit because as soon as it burns out and shorts out, it would be a dead short across the low voltage supply and pop the fuse for the headlight circuit.
For what
should be a straight forward series string of incandescent bulbs, finding and fixing all of the faults that built up in this one was an adventure with:
- A couple of plain old burned out bulbs that needed to be replaced
- Burned up bulb/socket combo in the sign box
- Daisy chained string of bulbs that was alligator clipped in but fell off while being troubleshot replacement for resistors
- Shorted socket where no bulb would ever light up
- The superdud with the failed shorting device that would take out the entire string, unless installed in that shorted socket.
I took the superdud home and put it on the insulation tester just to validate that it was responsible for the bizarre measurements we were getting, which I finally this evening. For convenience, I was going to use a ceramic lamp socket to hold it and provide terminals for the test leads to clip on to and I measured the socket empty first to make sure it wasn't going to unduly influence measurements with the bulb in it:
I set the user defined voltages for the 50 and 500 V ranges to 32V and 600V respectively to simulate the in-circuit operating voltage this bulb should have across it and the 600 V open circuit voltage it had across it due to the malfunction. At 600 V, this socket can be pretty much considered an open circuit at >260 GOhms.
1.44 GOhm at 600 volts.
2.42 GOhm at 32 volts.
The 1.44 GOhms with the 600 volts open circuit voltage applied made the approximate 3 V we measured downstream from the superdud make a bit more sense given the value of the superdud, the rest of the equipment on the streetcar it was in series with, then the 10 MOhm meter to ground. The high impedance meter and very high impedance voltage source created an unintentional voltage divider circuit. Taking resistance measurements around the superdud in 2021 and going from low to very high impedance readings also makes sense now.
Luckily it was a dud light bulb responsible for this. The resistance measurements we were getting in 2021 had me concerned that there was a metal fatigue break somewhere in the cabling in the ceiling of the streetcar that developed over 70-odd years of vibration in service. Replacing a faulty bulb was much easier than replacing wiring.
Anyways, that's a good example of some of my portable test equipment kit being used in the field for totally unrelated to work volunteer stuff.
