H. P. Friedrichs (AC7ZL) Homepage
Noise In Shack: Switching Power Supplies
Noise In The Shack
One of the greatest challenges to successful radio communications is noise. Nature hosts countless radio transmitters, including the sun, our atmosphere, and celestial objects, which produce signals that compete with, and often overwhelm faint man-made signals that carry the intelligence we are interested in. Hams call this QRN.
To this, we add the burden of noise signals that are the result of the unintentional radiation of RF from electrical and electronic devices. A faulty insulator on a power pole can produce a buzz that blocks large segments of the HF ham bands. SCR or TRIAC-based lamp dimmers can do the same. Computer equipment can generate "birdies" or "hash" that drives a receiver's AGC, reducing sensitivity. A water pump or a power tool switching on and off can produce annoying pops and crackles. One of the greatest threats that amateur radio has ever faced comes in the form of noise generated by so-called BPL systems. The idea of BPL is to provide broadband Internet service by supplying data through existing power lines. The problem is that the idea ignores the basic rules of physics, with the result that data-carrying power lines can blanket large regions with intense electrical noise. The noise is so bad that it can render large segments of the radio spectrum completely unusable.
An increasingly common source of man-made radio noise is the switching power supply. Switching power supplies, or switchers, appear in a vast array of consumer electronic devices, and are appearing more and more. They are now found even in the ubiquitous "wall-wart" configuration.
How does a switcher work, how does it compare to the classical linear power supply, and why does it generate so much noise? The answer to these questions begins with an analogy.
Regulated Power Supplies - An Analogy
Lets suppose you have a large tub or "kiddie pool." The pool has a small hole or drain in its bottom, where water is free to escape. To fill the pool, and hopefully maintain a fixed level in it, you use an outdoor faucet and a garden hose. You can spin the faucet handle to control the flow of water into the pool--- a little, a lot, or any rate in between--- in order to maintain a fixed level in the pool.
This system is a hydraulic representation of the classic linear power supply. The garden hose represents the supply's energy source (usually rectified power from a transformer). The pool represents the innate storage capacity of the supply (usually in the form of a large filter capacitor). The drain in the pool represents the energy that the power supply is delivering to the load, and the pressure of water delivered to the load (voltage) is related to the water level in the pool, which is why it is important that you maintain a closely controlled level.
To apply the analogy to a switching supply, you keep the pool, the drain, and the requirement to maintain a constant level in the pool. However, instead of using a garden hose to fill the pool, you arrange for a friend to deliver one-gallon buckets at brief, regular intervals. When the pool drains below the desired level, you call upon the friend to deliver buckets, which you subsequently dump into the pool. The moment that the level of the pool rises above the target level, you cancel your order for the next bucket. When the pool level falls below the target again, you resume delivery.
The primary difference between the linear and switching supply, then, is the manner in which water (energy) is added to the "pool." In the first case, water is added in a continuous fashion, throttled up and down depending upon the level of the pool. In the second case, water is added to the pool in a discrete manner, as fixed, uniform, and repetitive quantities of water. Control of the level of water in the pool is achieved by scheduling or canceling the delivery of future water buckets.
Given the logistics in delivering multiple buckets of water, or for that matter, discrete pulses of electricity, you might wonder why anyone would go the trouble of building power supplies this way. There are two reasons I can think of.
The first has to do with the way transistors work. In most linear power supplies, a transistor is used as the "faucet" by which to control the flow of electricity from the source to load. Ultimately, all of the current delivered to the load must first pass through the transistor (in fact, transistors used in this fashion are referred to as "pass transistors.) You might think of the pass transistor, in loose terms, as a rheostat or programmable resistor. The pass transistor, like a rheostat, meters the energy passing through it by dissipating unwanted energy as heat.
Heat produced by the transistor is really wasted energy, because it represents energy that was not used for any useful purpose. Thus, linear power supplies tend not to be very energy efficient. Since semiconductors like transistors suffer from exposure to heat, it is vitally important that the pass transistor be anchored to an appropriately large heat sink. The need for a heat sink adds to the cost, size, and weight of a linear power supply.
In the case of the switching supply, the pass transistor does not act as a rheostat, but as a switch, meaning it is either fully on, or fully off. When the transistor is on, it delivers a "bucket" of energy to the load, when it is off, it does not. When a transistor is fully on, very little of the energy passing through it is wasted as heat. When a transistor is fully off, negligible current flows,so again, very little energy is wasted. Since the pass transistor in a switching supply only resides in one of these two states, switching supply circuits are highly efficient. High efficiency means that little waste heat is produced. Thus, heat sinks can be reduced in size or even omitted, making the supply smaller and cheaper to produce.
The inefficiencies of the linear supply become unacceptable when a linear supply must regulate a relatively low voltage output given a relatively high voltage input. In this case, excessive amounts of heat must be dissipated by the pass transistor, so much in fact, that it renders the design impractical. To get around this, linear supplies are fitted with a transformer to convert and reduce the input voltage to a level that is somewhat closer to the desired output voltage.
This brings to me the second reason why switching supplies are attractive. Transformers, of the type used in linear supplies, are expensive and heavy. They contain lots of copper wire, and steel plates that compose the transformer's core. The more current that the power supply must supply load, the larger the transformer must be. This is because transformer capacity is related to the cross sectional area of the core and the gauge of the wire used to wind the transformer. Transformer size is also dictated by its operating frequency. As a rule of thumb, the lower the frequency, the larger the transformer's core (and thus the transformer) must be. Supply mains in the United States oscillate at 60 Hz. This is a low frequency, requiring comparatively large transformers.
Switching supplies, on the other hand, do not necessarily need input transformers. The high efficiency of transistor switches means that a low-voltage regulated output can be sourced directly from a high-voltage input. If safety considerations demand the use of a transformer to provide isolation, switching supplies win again. The rate at which "buckets" of energy are deposited into the "pool" can be on the order of 10,000 -100,000 times per second or even faster. At these high frequencies, a transformer's core can be made much smaller, and even fabricated out of materials other than steel. Once again, the switcher can be smaller, lighter, and cheaper.
So Why The Noise?
A clever fellow by the name of Jean Baptiste Joseph Fourier determined through mathematical means that any complex waveform can be described as the sum of a series of sine and cosine waves.
For example, if you were to set up a sine wave oscillator to produce a 1 kHz signal, and you displayed that signal on an oscilloscope, you'd see the smooth, graceful, curved sine wave that one would expect.
However, if you added more oscillators to the system, one at 3 kHz, one at 5 kHz, one at 7 kHz, and so forth, and combined their outputs, the composite wave you would have created would look very different. In fact, if you added enough additional oscillators, tuned to odd multiples of the fundamental frequency of 1 kHz, and you combined them in proper proportion, the oscilloscope display might surprise you. In short, the screen would display a square wave!
Obviously, this concept is reversible. If a use a signal generator to produce a 1 kHz square wave, Fourier tells us that, implicit in that signal, is an infinite series of sine waves whose frequencies represent odd multiples of the fundamental frequency. These additional waves are called harmonics.
Now consider an energy source that is connected to switch that is opening and closing at regular intervals, let's say 1 kHz. If you plot the output of the switch, what does it look like? A square wave! However, implicit in that square wave is a sequence of sine waves, whose frequencies lie at odd multiples of 1 kHz. The 1 kHz switch produces harmonics at 3 kHz, 5 kHz, 7 kHz, and so on.
This property is not unique to square waves. As I said before, any complex waveform contains numerous sine waves. What is worth noting is that waveforms that exhibit rapid rise or fall times, or "sharp edges," are loaded with strong sine waves oscillating at many times the frequency of the fundamental.
The connection between this concept and our discussion of switching supplies should now be evident. By its very nature, the switching supply produces the kind of waveforms that are loaded with harmonics. In the case of a switching supply operating at 200 kHz, the 3rd, 5th, and 7th harmonics lie right in the middle of the AM broadcast band. These harmonics and higher ones, leaking out of the power supply through its input or output terminals, or simply radiating from the wiring on the supply's circuit board, will wreak havoc with any nearby receiver.
Some manufacturers go to great pains to suppress this type of electrical interference. They may apply aggressive filters to the input and output sides of the supply, and shield the electronics with a metallic box. Unfortunately, all too often, they do not. In a future installment, I will describe a few examples of switching supply interference that I've observed in my shack, and what I could (or could not) do to resolve the problems.
Document Revision 1, xx/xx/2005
Document Revision 2, 10/14/2005