"I noticed an energy saving light bulb, fed from a 2-way switch system, was trying to light up very dimly every 3 seconds, over and over again, even though the switches were in the 'off' position."
You can imagine this writer's consternation when he noticed this flickering energy saving lamp, wondering if his handiwork of rewiring the house was at fault. This wonder grew when he measured the voltage at the lamp socket and read over 90VAC on his digital multimeter!
To make matters worse, the responses to a cry for help such as this usually attracts all sorts of answers including one which claimed it was a "borrowed Neutral" (whatever that is!?).
Before embarking on the proper answer, it has come to light on more than one occasion that many have no problem understanding capacitors especially when they can see them (such as in a fluorescent fitting). But, when it is an intangible item called 'capacitance', then haze creeps in!
In simple terms; When a wire is 'live', it sets up an electric field around it. Any item within that field will be subject to that field; And the closer to the source of the field, the more it is influenced.
The same with wiring. If a wire is running with a 'live' one, some of the field of the live wire will be making its way in to the dead one. Admittedly, the energy is exceedingly small by comparison, so anything such as a direct connection to Earth or via some reasonable load (even a small incandescent lamp) will comfortably negate the effects of the field being picked up by the dead wire.
However, should this wire be isolated or have a very high impedance load (such as the lamp mentioned here), then the chance of some of this field being reflected as a voltage on the wire is very high indeed.
Electrical cables have a mutual capacitance of about 100pf/metre upwards so it only takes 10m to get a whole 1nF causing (at 230VAC) 73µA current flow.
Adding to this problem is the "looping back" of the control wires (shown in the picture alongside). When in the correct switch position, the Live would be fed to the remote switch, and couple not just to the return wire, but to the secondary control wire too, this being connected to the return wire. Here up to 200pF/m could be realized.
An energy saving bulb is known in industry as a compact fluorescent lamp (CFL), and these are high-frequency oscillators fed from direct rectified mains (a typical circuit can be seen here). The available "leaked" current would slowly charge the smoothing cap until the oscillator could 'fire'. This would then discharge this cap till it could no longer keep oscillator running and thus the CFL burning - and repeat this process ad nauseam.
The problem may be temporary such as only occurring shortly after the lamp is turned off. This can be attributed to the lamp being warm and the gas igniting more easily. When cold, the energy from the oscillator is absorbed totally by the filaments and the gas has no time to ignite before the capacitor runs out of steam.
As for a cure, solder two 150kW resistors in series (to form 300kW) and solder this combination across the smoothing cap in the CFL. This will keep it from reaching the 'firing' voltage of the oscillator. Please do not be tempted to use a single 270k 1/4 watt resistor, they are not designed for 250V across them permanently (and the rectified voltage, as is, can reach 360V - and that is if the REC stays within 253VAC).
This is only one of the many types of problems that can be attributed to cable capacitance, especially when one of them is wiggling about at around 700V peak-to-peak! This is not a power quality related issue, just simply one associated with the way things are done.
The same influence can exist on multiple circuits where they share a common path - such as when all circuits are fed to another part of the premises through a common trunking.
Isolating a single circuit to test for insulation i.e. where the Phase and Neutral are disconnected from the source, can often yield voltages to Earth in the order of 30 to 100V (dependent purely on the capacitance between the affected circuit and those within the same trunking).
This can play havoc with insulation testers with 'inhibitors' that prevent testing on 'live' circuits. In many instances, the only methods of combating this is either power down all circuits within the trunking, or prevent the inhibitor from doing its work!
I have received a number of emails from folk who have found AC driven relays remaining closed when using a long control cable. Although not in itself a power quality problem, I've seen the most amazing answers on 'help' forums (even to the point of blaming harmonics!), so I though I would add it in here.
I must admit I do have a hard time believing the cable could have sufficient capacitance so as to hold an AC relay closed, there is also the possibility the cable may be waterlogged thus lowering the impedance even further.
One could put a resistor in parallel with the relay coil in the hope of increasing the loading on the control circuit and therefore reduce the effect of the cable capacitance. This is not a bad idea, but the resistor has a by-product of generating heat - a much undesired effect. This method also carries a danger in that should the resistor burn out the controlled circuit will again remain energised.
The first and simplest cure is to connect a capacitor in parallel with the relay coil. The control cable capacitance and coil capacitor will act as a capacitive voltage divider. If the 'loading' capacitor is large enough then the resultant voltage across the relay coil will be low when the control cable is open circuit. A few microfarads should be all that is required (a power factor correction capacitor from a small fluorescent lamp should do the trick).
The second and better cure is to move the control voltage to the control side of the cable i.e. feed a voltage down the cable to the relay, and not use the cable to remotely interrupt the voltage to the relay coil.
It must be remembered that AC is the problem here. There is one other danger in that should the cable become water-logged (thus lowering the resistance of the cable significantly) you land up back where you started being the controlled circuit remaining energised with the first cure, or possibly blowing fuses with the second.
The third and best cure is to change the control voltage to DC. It has two inherent safety features being the capacitance is no longer an effect to be concerned about, and should the cable become water-logged any contact points 'plate' and become a high resistance thus removing any danger of the controlled circuit remaining energised.
© 27.04.04 / 17.11.04