Fluorescent/LED lights and AC power signals

[Hawkwings]

I recall reading something on this forum about how either fluorescent or LED lights for residential purposes degrade the AC power signal that the power company sends through the lines. I can't find any references to it now however. Does anyone know what this phenomenon is, or am I just making things up?

[aerius]

http://bbs.stardestroyer.net/viewtopic. ... 2&t=144768

And there you go.

[StarSword]

Being an electrician by training (note, I don't have my license yet, and I'm having trouble finding work as a helper), I'll give it a shot.

The problem stems from the electronics. LEDs are a type of diode (which should be obvious when you break up the acronym). The very definition of a diode is a device that allows current to flow only in one direction. This means that during the 1/120 of a second that the current flows along the forward-bias of the diode (positive to negative direction on the diode), current flows normally. During the next 1/120 of a second, the current flows in reverse, what's called "reverse-bias" (negative to positive), and the diode does not allow current to pass, breaking the circuit.

The reason fluorescent light electronic ballasts, computer power supplies, and other such AC-to-DC converters (called "rectifiers") degrade the AC signal is fairly straightforward. During the forward-flow half-cycle, current passes through a diode and into the item, but also charges a capacitor, a device which resists changes in current. As the current falls to zero and begins the backward-flow half-cycle, the capacitor supplies a current that maintains the flow through the component. This is called "half-wave rectification," because you're only taking half the AC waveform into your device.

Since you're drawing the current irregularly (as opposed to steady, regular flow), it screws with the waveform. You can see the difference if you take apart a rectifier and look at it with an oscilloscope.

There are ways to minimize the irregularities. One is to use a larger capacitor. Another is full-wave rectification, which, typically via a second set of diodes, allows you to pull the backward-flow half-cycle as well as the forward.


But you can't eliminate the irregularities entirely, and they're transferred throughout the grid. One result is lowered lifetime on items like motors, which depend on a regular power supply. Also, I've personally seen heat-blackened insulation in an older breaker box at Durham Tech, which can be traced to all the fluorescent lights and computers in the building.

[Darth Wong]

I love your use of diagrams to illustrate the point. Most people don't bother with that sort of thing.

[aerius]

A few things to note. An LED is a diode but the current has to be rectified before it gets there, they can't handle a reverse bias beyond 5-10V without going up in smoke. You need to rectify the current and regulate the voltage before it gets to the LED, in other words it needs a low voltage DC power supply.

Also, pretty much everything uses full-wave bridge rectifiers these days, they come in a nice small package and barely cost any more than a single diode. With regards to capacitors, this is where it gets fun and counter-intuitive. A capacitor resists changes in voltage, and beyond the minimum required to keep voltage ripple reasonable a larger capacitor will actually create more noise. To reduce the noise you actually want an inductor between the rectifier & capacitor to limit the peak currents and lengthen the amount of time it takes for the capacitor to charge from the incoming current. The longer explanation can be found here.

[StarSword]

I love your use of diagrams to illustrate the point. Most people don't bother with that sort of thing.

I actually wanted to get some better diagrams than the ones from Wikipedia, but since I graduated from Durham Tech I can't access the Blackboard site of my Industrial Electronics class anymore. Those diagrams are actually incomplete: they left off the capacitor entirely. If I remember correctly, it's supposed to be wired in parallel with the load.

A capacitor resists changes in voltage...

Dammit. I always get capacitors and inductors mixed up.

[LaCroix]

Actually, a reservoir capacitor to smooth out ripple is a standard feature in any AC-DC converter nowadays, so you get a very flat "sawtooth" wave, depending on size of the capacitor , with a slow rise in voltage at the start, also depending of size (big C evens out better, but slows the start up more).

[Starglider]

Since you're drawing the current irregularly (as opposed to steady, regular flow), it screws with the waveform.

As aerius pointed out, half-wave rectification is virtually extinct. The main problem is the low power factor of switched-mode power supplies, and to a lesser extent directly connected semiconductors. Power factor is a measurement of how close the current draw waveform for the device is to the voltage waveform from the supply. Low power factor is bad as it decreases efficiency and potentially unbalances the local grid. A simple resistive load, e.g. an incandescant light or electric heater, draws current almost exactly proportional to voltage applied and has a power factor of one. Inductive loads such as motors tend to lag the supply, while capacitive loads lead, but their power factor can usually be corrected with a simple capacitor or inductor. Switched mode PSUs, including most electronic devices and electronic ballests in fluorescent lights, play hell with power factor because they draw power for tiny fractions of a second at arbitrary points in the cycle; without suppression, they also inject a lot of noise. Complicated regulation circuitry is required to fix this; fortunately electronics are cheap these days. I'm sure there's tons of reference material online for this if you google 'power factor correction'.

P.S. I think the electrical eng. modules I took were harder than any of my compsci stuff. I could never remember the integrals for the trig functions.

[aerius]

A simple resistive load, e.g. an incandescant light or electric heater, draws current almost exactly proportional to voltage applied and has a power factor of one. Inductive loads such as motors tend to lag the supply, while capacitive loads lead, but their power factor can usually be corrected with a simple capacitor or inductor. Switched mode PSUs, including most electronic devices and electronic ballests in fluorescent lights, play hell with power factor because they draw power for tiny fractions of a second at arbitrary points in the cycle; without suppression, they also inject a lot of noise. Complicated regulation circuitry is required to fix this; fortunately electronics are cheap these days.

Also, old style linear power supplies generally don't kick out noise in the MHz range, you'll get harmonics out to maybe a few tens or maybe 100 kHz which can generally be taken care of with a relatively simple LC filter. A few caps between the phases and to ground combined with a couple inductors & resistors and the noise is pretty much gone. Switched mode PSUs will not only dick up the power because they're all C input devices, they also radiate noise that's well into the radio frequency range, the noise spectra on those things goes out into the 10's of MHz, which turns every freakin' wire it's connected to into an antenna which makes filtering a royal pain in the ass. You basically need an RF shielded box and RF suppression techniques to get the noise out of the damn things.

We got a demonstration of this in one of my lab classes where the TA took a portable radio, tuned it in between stations and held it next to a SMPS unit. You could hear the noise radiating from the power supply.

P.S. I think the electrical eng. modules I took were harder than any of my compsci stuff. I could never remember the integrals for the trig functions.

Same thing here. My compsci stuff was tedious and frustrating at times because of all the little dumbass mistakes I made, but I never had any issues understanding what had to be done and how to do it. The EE courses I had to take left me scratching my head and going "WTF??" on more than one occasion. Strangely enough the EE course material has stuck with me far better than the compsci stuff, using a Windows or Mac computer is about where my computer knowledge ends these days.