Electrical feedback

Feedback or back feed?

Try Googling the terms.

Sounds like maybe a loose or open neutral in the circuit.

...downstream of the last 3-phase transformer in the supply and upstream of the installation under test.

GA

When I was working in AVAC, I've ran into several lose neutral connections in multiple high rise buildings within aluminum wire feeding sub panels. Biggest complaint was light bulbs burning out to fast.

Because of the inductive noise present on the circuit wiring conductors.

The tester lights used today are very sensitive and will indicate power is present at very low voltage/noise levels.

Lots of good answers so far. It helps to know your test equipment. The really sensitive testers can start glowing with as little as 10 VAC if you have a capacitive path to ground in the loop (such as your hand). The current in this case is below the threshold of sensation.

Most residential power feeds have a neutral tap on a transformer secondary tied to earth ground. If that neutral line connection to the center tap has become an open circuit, then voltage regulation is lost and what you have is an unbalanced voltage divider. It is not uncommon for some lights to get brighter while others are dim when that happens. It is also not uncommon for other devices to burn out with too little or too much voltage.

Use a 100W incandescent light bulb for a tester. I'll bet it won't be dimly lit.

Elementary lectures on electric discharges, waves and impulses, and other transients (1911)

https://archive.org/details/elementarylectur00steirich

This is generally a result of mutual inductance where the alternating voltage in one conductor can induce an EMF in a neighbouring parallel conductor.

This effect is commonly seen where an active and a switch wire form a pair running from a fluorescent light to the switch. The induced EMF in the switch wire can cause the fluorescent to glow dimly even when the switch is open.

Two common ways to tackle this problem are to separate the two wires, or to swap the switch wire and neutral at the luminaire so that the switch wire goes to the ballast rather than to one end of the tube.

This is one reason why I insisted on keeping my old Simpson meter when the new electronic Fluke models came out. The energy present is not really feedback. It is inductive or capacitive coupling. This can happen across the contacts of switches or with wire laying next to each other for even moderate distances. By comparing the values of an old mechanical analog meter that takes a lot of power to run to a modern electronic meter you can get 2 points on a curve and can distinguish between capacitive and inductive coupling. A test lamp will show some energy but you usually have no idea how much. It could be enough to kill you or you might be able to touch it and feel nothing. You have to know the characteristics of the test device. Working high and medium voltage, when "tic testers" first came out they would fire anywhere in a substation and indicate energy, but they were still useful and prevented a lot of deaths. Later they tuned the sensitivity down and they were even more useful.

Are there 'illuminated' switches in the circuit? Switches that have a little lamp inside so you can see the switch when it is off (and the room is dark) will naturally have some 'leakage' current due to the need to draw power for the bulb. This will often be enough to keep LED lights attached to the switch dimly lit when 'off.'

Building wiring can be very tricky, you should use this rule of thumb:

If you're not sure what you're doing or what is going on, hire a professional.

If you are completely confident that you know what's going on, you're probably overlooking something because you don't even know it exists(1), and you should hire a professional.

Notes:

1) More often then not, the problem that will surprise/confuse/kill you is due to something the person before you did when THEY thought they had all the answers.

At one fire that I responded to as a volunteer firefighter, we had already disconnected power from the house to avoid further fire in a mobile home. Upon checking house circuits, we found a night light which was still glowing dimly. We verified no power at the outlet, but the light continued to glow. All power was opened then by completely disconnecting the cable at the service entrance to the house from the pole (there was a disconnecting means switch which had already been opened). The light continued to glow. When it was removed from the outlet, the light extinguished. When it was placed in any other outlet on any different circuit, the light would glow. We verified that no other back feed circuit was in place. The only logical conclusion we could draw from this is that the light which had an LED was sensitive enough to be picking up enough noise from the wiring to glow.

Many of the new LED bulbs will likewise have a dim glow after being turned off. It is undetectible in the daytime unless the room is dark.

One other item to consider... the wiring in the house can act as an antenna. It is possible that it is picking up radiated signals from some other source such as a local TV or radio station.