Today I noticed something peculiar. In our garden, we have a spruce in which I have put a string of LED lights. The lights are powered via a switch mode AC/DC power supply that creates the low voltage required by the LEDs from the mains voltage. This power supply is connected to an extension cord and a timer inside the garage and I have set the timer such that the lights are on only while it is dark outside. However, the LEDs were mysteriously glowing dimly in the middle of the day when they were supposed to be off.

Hmm, that is odd I thought.

I did not think too hard about this at first, but a while later I realized what could cause this behavior. If the timer happens to be connected such that it breaks the neutral wire of the mains voltage instead of the live wire (phase), there could perhaps be enough capacitive leakage via the Y-capacitor going from neutral to protective earth inside the AC/DC supply such that the supply gets enough power to dimly light up the LEDs.

Y capacitors are present in virtually all switch mode AC/DC supplies. They are connected between phase and earth as well as between neutral and earth. Their purpose is to limit the amount of unintended radio frequency noise generated by the switch mode supply that is radiated back onto the mains network. There is usually also an X capacitor for the same purpose connected between phase and neutral.

In addition to the Y capacitor, there will also be leakage via the capacitance of the cable from neutral to earth as well as from neutral to the external world which currently consists of a quite wet lawn. But I would guess the leakage via the Y capacitor is much larger than the leakage via the cable capacitance.

The equivalent circuit for the case when the timer breaks the neutral line is shown below.

The dashed arrow shows the flow of current from the phase wire, via the AC/DC circuitry in the dotted box, the CY2 capacitor and back via protective earth. There will also be some current from from the neutral (disconnected) wire of the extension cord and the cable capacitance to the earth wire, CcableNE as well as a little current flowing in the capacitance from the cable neutral wire to the outside world (which as of this writing is a wet lawn). This capacitance is not shown in the figure above.

So, a hypothesis about what is happening is fine, but how could I test this? Well, that is not too hard. Just rotate the timer 180 degrees in the wall outlet so that it breaks the other wire, which according to the hypothesis would be the phase.

The rotation of the timer changes the equivalent circuit as shown above. Now the phase is stopped in the timer and no leakage can occur.

After this maneuver, the LEDs in the tree were completely off, so the hypothesis seems to be correct.

I would guess there are numerous similarly mysterious cases of LEDs powered via AC/DC converters connected to protective earth that light up dimly even when they are supposed to be off. The remedy is to make sure that the live wire is interrupted by the circuit breaker, rather than the neutral wire. In installations connected to wall outlets, this can be done by rotating the plug (if you are in country where plugs with protective earth can be rotated). In fixed installations, it is necessary to swap which wire is connected to the light switch, which probably requires the intervention of an electrician.

It is that time of year when one puts electric advent stars and candlesticks in the windows. I usually use timers to control them, but this year I ran into a problem. Some of my timers were of the type “EverFlourish EMT757A”, probably bought at Clas Ohlson:

Despite having a notoriously bad user interface (that forces me to google the user manual each time I want to use them), the timers, that had been laying in a drawer since last season, showed an entirely blank LCD display, even when connected to a mains outlet. I guessed that an internal battery just need to be charged, so I left them connected for a day to allow them to be charged up. The display however remained blank.

Hmm. I had encountered similar timers before that needed the internal battery to be replaced, so I decided to have a look inside to determine if this was the case again. It turned out that the screws used to keep the units together had tamper proof tri-wing heads. Fortunately, I happened to have bits that fit this kind of head, so this did not deter me.

I found that there was indeed a tiny chargeable NiMH battery inside the unit (the green component with white corrosion on the side, close to the center of the picture below). Despite the unwanted corrosion, I found that the battery was actually charged to 1.3 V or thereabout, which is fine for a NiMH battery. Maybe it did not need to be replaced after all?

But why then was the display dead?

I put the timer back together again and had another look at the front. Pressing any of the normal buttons did not result in any reaction, but there was also a tiny little recessed button marked RESET. I used a pen to press it and voilà! The display came to life!

Stupid me to not try this before I disassembled the unit.

I suppose the reason the timer behaves like this is that the processor controlling the display probably does not have a proper power on reset circuit. So when the supply voltage is ramped very slowly, as it is when it is powered from a battery that is trickle charged from a state of being completely dead, the processor does not jump into action properly, even after the supply voltage reaches its proper level.