Modern power supply operation

[lcaillo]

The question of the value of power line filtering and regulation keeps coming up. There are lots of assumptions that are made about line noise and video, so I thought it useful to discuss how most power supplies work in order to give some context to the matter.

Nearly all modern video equipment uses a switching power supply. What this means is that the 60 Hz (USA) a.c. is rectified to d.c, then that d.c. is switched on and off at a frequency of between 50 kHz and 150 kHz and these pulses passed through a transformer (called a converter or switching transformer). The secondary pulses are rectified to d.c. again. Then the d.c. is filtered and distributed to the various circuits. Some of the lines are regulated further by additional circuits. The voltage on the secondary of the converter is regulated by varying either the frequency or the pulse width of the d.c. pulses through the primary of the converter. Some systems use feedback to tightly regulate, some do not.

So let's consider what happens to the a.c. and the noise that it may carry. Lets say we have a few volts of noise on the a.c. line. This would be a pretty high level. Most sytems typically have noise in the tens or hundreds of milivolts. he a.c. enters the set and goes through a line filter. This may remove as much as some power conditioners or may not. Assume it does not. It is then rectified by a traditional bridge rectifier in most systems. At that point it is d.c. and the noise may have survived if the diodes in the rectifier are slow. In practice much of the noise will. Then it is filtered by relatively large capacitors. Much of the noise will not make it past that point, but let's assume it does. Now we have about 160 volts d.c. with a couple of volts of noise riding on it (this is actually a VERY unlikey case but being conservative...). That d.c. is now switched on and off and passed through the windings on the primary of the conerter. At this point the waveform is about 120 kHz pulses that look like spikey distorted square waves. If the noise is in a frequency range that the converter is efficient at passing, it will become part of that series of pulses. On the secondary side it looks similar. Frequencies that fall outside of the resonance of the converter are essentially filtered. Now the problem is taking that inherently noisy pulse and rectifying it and cleaning it up. You have here what looks like really nasty noise that is hundreds of times the level of the noise that may have survived, and in the same frequency range. High speed diodes are used to rectify the signal back to d.c. and further filtering is done to remove the switching noise from the conversion process. Now the d.c. might be used, but it is often regulated down to lower voltages by an additional regulator which has additional filter capacitors and likely coils to remove any remaining noise.

Now, before I am accused of being biased against line conditioners, let me say that THEY DO WORK! I have measured it many times. They do clean up line noise. That is why I have one on my bench. I want reference levels that are clean. The point is that that same line noise that they clean up would not have made it to your video in MOST cases. I have measure this as well. I repair hundreds of units each year. On my bench I have an old line filter that is rather crude but effective. I have measured the noise with and without it, and in the shop with lots of other equipment, test equipment, and computers running, I often have line noise that is easily seen with a scope. I have tried measuring the output of the power supplies in many units to see a difference with and without the line filter and NEVER been able to find the line noise nor any remnant nor harmonic of it when the filter is not in place.

As for regulation, we commonly vary line voltage to test units, and most power supplies are well regulated between about 90 volts and 130 volts. They are likely to be less efficient at 90-100 volts and eventually run a little hot, but for the most part, modern switching supplies are very resilient and have very stable output over these ranges.

That said, if you believe that a line conditioner is needed, or just feel better having one, by all means, use one. If you understand how most systems work, however, it is clear that it is rare that the effect would be visible. I hold open the possibility, and am open to anyone to demonstrate a visible benefit. My experience is consistent with the analysis of how power supplies work above, however, and the reasonable conclusion that it is unlikely that line noise will be visible in most consumer components or displays, except when caused by ground interactions. This is a matter for another discussion, but it likely accounts for the reason that many people experience visible improvements when adding a line conditioner or even a good surge suppressor.

[Wayne A. Pflughaupt]

Great info as usual, Leonard.

That said, if you believe that a line conditioner is needed, or just feel better having one, by all means, use one. If you understand how most systems work, however, it is clear that it is rare that the effect would be visible.

What would we be seeing if it's “visible?” Would it look different with a CRT monitor vs. LCD vs. a projector, or would it be the same no matter what the display device?

Regards,
Wayne

[lcaillo]

Hmm, good question, but tough to answer. I have not actually seen the effects. I hold open the possibility in order to remain objective. Also, many people have reported seeing improvements. Any time lots of people report an experience I assume that there is some reason that is not just mass placebo effect. What I have heard reported are changes in contrast, background noise, brightness, color saturation, and ringing on edges. Since all of these can be associated with ground loops and ground noise, I assume that most of the effect comes from creating a common ground point in the line conditioner. I also assume so because I have heard of the same kinds of improvements when users connect a surge suppressor that protects signal lines, where again, the grounds get tied together. I have analyzed the matter from every angle that I can approach practically and done a fair amount of testing and my best guess is that many people simply end up cleaning up ground problems and this likely is the reason that they think that they get an improvement due to a line conditioner. There is also the possibility that there are "golden eyes" analogous to "golden ears" that see things that I do not. There is also the possiblility that some people are just deluded by expectation bias and believe what they want to believe about their perceptions. I just try to figure out the underlying effects that might be present in the technology.

We simply never run into noise issues in our systems, unless there are interconnections between components at a great distance and we get ground or induced noise. We nearly always solve these by placing circuits on the same breaker or at least the same bus, or by fixing ground or shielding issues. Occasionally, when there is not another practical solution, we use isolation transformers, but this is very very rare. We sometimes use more expensive line conditioners in elaborate systems, but most just use basic Panamax surge protectors and good grounding practice.