Power Line Noise

Electrical noise on the power line is something I've seen in several threads here. I've seen many suggestions for ways to clean up theis noise, many of which also have a bad side. The purpose of this thread is to try to put them all together in one place.

1. It's been suggested that MOVs can be used to clean up some of the noise. The noise removed by an MOV is primarilly high voltage spikes. It does this by shunting the spike to nuetral. By removing these spikes, an MOV should also increase the life of the attached electronics. I know of no negative impact of using MOVs.

2. I've seen lot's of people suggest power filters. A good power filter is probably the best noise remover you can find. The problem is that a GOOD power filter is sometimes difficult to find. Most of the power filters I've seen are 3 stage mirrored PI (inductor on both power lines) filters. Often, these are not adequate. A good power filter should have at least 5 stages, in a mirrored PI configuration. It should also be rated well in excess of the maximum expected current draw, so as to minimize the voltage lost due to resistance. A good example of this is item number EMI-21 from 'http://www.allelectronics.com/'. One important thing to remember is that EACH audio component needs it's own filter. This will keep noise generated by a component from getting into another component.

3. It's also been suggested that an isolation transformer be used. This should remove much of the high frequency noise. It needs to be protected by both MOVs (on both the incomming and the outgoing lines) and fuses. If the transformer has an electrostatic shield, the noise rejection should improve. The negative impacts of using an isolation transformer are weight (they're heavy), cost (they're not cheap) and increased impedance on the power line. The impedance should only be an issue for large power amplifiers, and only if the amp's current draw is a large portion of the rated power of the transformer.

4. I've seen a few people suggest that aftermarket power cords can decrease noise. If you're going from an 18 guage cord to a 14 guage cord, or adding shielding, I might believe that. My suggestion (for what it's worth) would be to use the heaviest power cord that you can find. Anything past 14 guage is usually hard to find.

I use items 1, 2 and 4 (14 guage shielded IEC power cords) in my system. I think that this helps to kill most of the noise that comes from the power line.
 
Hi Thoth,

MOV's will help reduce voltage spikes, but I don't see how they could have any other positive effect on line noise.

Actually there are two good reasons to not use MOV's. First, they wear out, an MOV is degraded every time it absorbs a voltage spike and will eventually fail, usually catastrophically. Second, MOV's pollute the power for every other piece of equipment on the same power circuit. They take a voltage spike that is on one line and turn it into a common mode voltage surge on both the line and neutral, this is actually harder for other equipment to ignore. In addition the way they are usually used, they dump noise to ground. Most audio engineers won't have anything to do with MOV based line protectors for that reason alone.

The best powerline protection product I know of is called SurgeX. This product uses a big honking inductor and caps to filter noise and spikes from the incoming power line. They aren't cheap, but they do work and don't wear out. I have seen these used to protect $100K mixing consoles.

That was a good point you made about each piece of equipment needing it's own filter. This is very true especially for power amps. Both linear and switching power supplies can feed noise back into the power line (the type of noise generated in each case differs, but that doesn't really matter in this case) and you don't want that noise getting into your preamp.

I have a hard time imagining that anything you do to a line cord is going to make much of a difference. Even if the few feet of cable between the wall outlet and your equipment is perfect, what about all the wiring inside your walls? Using bigger gauge power cords won't make any real difference unless you are using a long extension cord. Your line cord is still in series with the household wiring which last time I checked is not audiophile grade. Even if the line cord impedance is below 0.01 ohm, the wiring in your house typically has several ohms of impedance (I once measured the source impedance of the AC power at my lab bench in my 45 year old house, it was over 8 ohms).

Phil Ouellette
 
haldor said:
MOV's will help reduce voltage spikes, but I don't see how they could have any other positive effect on line noise.
There are places where the PRIMARY noise on the line is high voltage spikes. This would usually be near lots of high power electric motors (an industrial area).
Actually there are two good reasons to not use MOV's. First, they wear out, an MOV is degraded every time it absorbs a voltage spike and will eventually fail, usually catastrophically. Second, MOV's pollute the power for every other piece of equipment on the same power circuit. They take a voltage spike that is on one line and turn it into a common mode voltage surge on both the line and neutral, this is actually harder for other equipment to ignore. In addition the way they are usually used, they dump noise to ground. Most audio engineers won't have anything to do with MOV based line protectors for that reason alone.
Good point about the failure mode. As to the pollution, I guess it depends on how they're connected. I've seen them connected across the power leads. I've also seen them connected in a 'triangle', between the power leads, and ground. I think that the second method is the one that throws the noise (spikes) into ground.

The way I've got them connected is across the power leads (4 in parallel), mixed with 4 0.02uF 1KV caps. The power cord is 15 ft. long, 10 guage (plugged into a dryer outlet; 240V/two-phase/30A), with 2 0.005uF 3KV caps in the plug. The caps should pull out the small spikes, leaving the big spikes to be absorbed by the MOVs. I feel that this mix gives the best of both worlds.
The best powerline protection product I know of is called SurgeX. This product uses a big honking inductor and caps to filter noise and spikes from the incoming power line. They aren't cheap, but they do work and don't wear out. I have seen these used to protect $100K mixing consoles.
This appears to be an inteligent solid state spike filter (a smart MOV), combined with a hefty capacitor/inductor filter. I've never dealt with them, but there is the potential for a VERY good filter here. From what I can see, they've achieved that.

I have a hard time imagining that anything you do to a line cord is going to make much of a difference. Even if the few feet of cable between the wall outlet and your equipment is perfect, what about all the wiring inside your walls? Using bigger gauge power cords won't make any real difference unless you are using a long extension cord. Your line cord is still in series with the household wiring which last time I checked is not audiophile grade. Even if the line cord impedance is below 0.01 ohm, the wiring in your house typically has several ohms of impedance (I once measured the source impedance of the AC power at my lab bench in my 45 year old house, it was over 8 ohms).
I have high power amps (250WPC stereo) that came with an 18 guage (16 guage? I don't recall exactly) IEC power cord. When that was replaced by a 14 guage shielded power cord, the background noise dropped. This was probably more due to the shielding than the wire. On the other hand, I'm sure that the heavier wire cut distortion at high power levels.

[Edited by thoth on 09-04-2001 at 11:27 PM]
 
Thoth said:
I have high power amps (250WPC stereo) that came with an 18 guage (16 guage? I don't recall exactly) IEC power cord. When that was replaced by a 14 guage shielded power cord, the background noise dropped. This was probably more due to the shielding than the wire. On the other hand, I'm sure that the heavier wire cut distortion at high power levels.

Sounds like you are picking up noise from other equipment in your system. Interesting, I never have used shielded power cords before. Did you characterise the noise, was it inductively coupled line noise or was it RF? Shielding will attenuate RF, but the only thing that helps inductive coupling is either distance, a tightly twisted pair of wires, steel conduit or some combination of the same.

I wonder if you could get better effect by using a tightly twisted pair (line and neutral twisted at least 3 turns per inch) with a seperate ground wire? It would be interesting to try making some custom power cords like this and see how they compare to your shielded cables.
 
Some of the mega-buck power cables are essentially heavy gauge single strand (i.e. house wiring), which is why they're so stiff. I've never played with them, but they're reputed to be pretty good.
I haven't seen any white papers on this, but my guess is that they're using, say, 14 ga. single strand as a sort of anti-skin effect wire--less surface area for high frequencies to crawl along. This would tie in nicely with Thoth's comment about things being quieter, as it would tend to suppress high frequency garbage, simply by virtue of its larger gauge.
So, in a way, you could say that the wiring in your walls is audiophile grade, after all...

Grey

P.S.: Some people claim that ordinary house wire is the ultimate subwoofer cable. Since all you're asking of the wire is low frequencies, there might be something to that, too. I've done some limited experiments along those lines, but never finished. I'll get back to it one of these days.
 
Skin effect means that the as the frequency of the current flow goes up the current flow concentrates nearer to the outside of the conductor. The reason why you often see copper tubing used in military high frequency RF equipment is not because a solid conductor won't pass RF just as well, it's that the core of the conductor is not required to carry the current and you can save a lot of weight (very important in airplanes) by using hollow tubes instead of solid conductors.

If what you suggest is actually happening then the noise filtering would be improved by using the smallest diameter solid conductor possible (since the smaller the diamater the less skin region there is, therefore the less high frequency passed). Counter to the usual thinking.

A large diamater wire is going to have virtually the same impedance at RF as it does at DC (ignoring the effects of the wires inductance of the wire which isn't effected by the diameter of the wire). So in my opinion this is not what's going on.

Don't get me started about the audibility of wire. I am blessed with the inability to hear much difference between cables as long as they are appropriate for the purpose and well constructed. This means I have that much more money to blow on electronics. ;^)

Phil
 
Phil,
You are correct that skin effect concentrates current on the surface as frequency increases, but your conclusions might need rethinking.
As you decrease the diameter of the wire strand, the wire diameter eventually reaches the point where the high frequencies can travel throughout the diameter of the strand. This leads to something approaching 100% utilization of the wire diameter for both low and high frequencies.
Conversely, a larger diameter wire carries high frequencies only on the surface, but low frequencies can permeate the entire wire cross section, leading to high frequencies only being able to use a comparatively small percentage of the cross sectional area compared to low frequencies, which are able to use the entire area. The result is a decrease in carrying efficiency at higher frequencies. Hence, a large wire doesn't have the same Z at RF as it does at DC, which is exactly why they are able to use tubes instead of wires.
In other words, the usual thinking is correct.

Grey
 
Pete,
Actually, if you approach it from a purely numbers viewpoint, skin effect isn't really a serious factor at audio frequencies. But, like some other theoretical vs. what-you-hear debates, it does seem to matter--at least on good systems. For low to mid-fi systems, I'd say it's a waste of time to worry about it.
Now, from Thoth's side of the table (mine too, I guess), I imagine the question devolves to this: What frequencies are riding in on top of the AC? RF? Supersonic? Audio range? Subsonic?
And: How do these effect the perceived 'silence between the notes'? What mechanisms are involved?
N.B.: If we start talking spikes, then Fourier analysis will give us a spray of frequencies from DC to light--the only difference being that the problem is of short (let's say some fraction of a second) duration. This isn't really a separate problem unless the pure magnitude of the voltage overwhelms the ratings of some device downstream from the power supply. Elseways, it's simply business as usual, trying to keep non-60Hz AC out of the power supply.

Grey

I'm thinking out loud, here...
Supposing something rides in on the ground wire, not the hot. All assumptions are that ground is ground, and that ground remains stable. (After all, ground is where the "quiet" comes from, right?) But if ground *isn't* stable, it's going to raise hell with all assumptions that you can 'shunt the incoming trash to ground.' Ergo, you need some way to "ground" ground. I've got a half-formed idea about isolation transformers, but I'm going to sit on it until I've had time to think it through. Note that, at least here in the US, the implication is that anything picked up on the ground wire enters between the breaker box and the power supply, i.e. local interference such as RF, hash coming back out of digital equipment, etc. Shielded AC wiring might be in order, after all.
A friend of mine discovered that the outlet he was running his stereo off of wasn't grounded properly (old house with old wiring). He solved the problem by ramming a ground spike into the ground right underneath his window, then ran a purely local ground wire to that. I'm beginning to think he may have been onto something.
 
Are we believing the marketing hype?

G'day Grey,

I guess my point is, is skin effect simply a convenient reason, essentially invented in this application, to explain something we don't understand? Skin effect exists, I believe we can all agree with that, however so do many other effects. Every day the earth's magnetic field shifts slightly, nevertheless I feel quite confident that this effect can be discounted as affecting the operation of my system. Nevertheless the effect is a known scientific fact, just as is skin effect. There are virtually unlimited effects on the different components within the system, why the emphasis on skin effect?

It seems ironic that audiophiles as a whole latch on to simple things with dubious objective merit simply because they are easy to understand, yet ignore factors very much know to have a marked affect on the quality of the end product. I have easily lost count how many threads I’ve read about cables, and yet have barely heard mention (with the exception of “off-the-shelf” op-amps) of the affect of different semiconductors, other than saying “errr, match them the best you can”. Indeed, and correct me if my source is wrong, I believe it was Nelson Pass himself who makes this very same point. Why don’t we see “after market transistors”, “upgrade MOSFETS”, etc. on the market. Well perhaps it’s the cynic within me, but it seems a lot easier to market a couple of wires that can easily be plugged in and are universal, than an upgrade that possibly requires extensive modification to specific models. As if to prove this point, look at the vacuum tube arena; they are easy to replace by anyone, and “upgrade” versions are marketed at high prices.

Notice I’m not saying that cables don’t make a difference, I used to be in that camp until I set up an experiment with speaker cables and was proved wrong. However I will confess that I have been unable to hear any difference between different interconnect cables I have tried.

So are we taking the marketing hype to be fact?

Just stirring the pot in the hope somebody may like to come up with some answers. Have taken cover and await the flak.

Cheers,

Pete
 
An alternative view

In my experience, spike suppression using MOVs (line to neutral) is most beneficial and is a must for all equipment. I usually use them before the distribution block so that the whole system is protected at a single point. The points raised about failure are relevant, but these devices are so cheap that the can be replaced on an annual basis, or more frequently if your mains supply is particularly bad.

I do not believe that line input filters should be fitted to audio equipment, they should be limited to that for which they were designed, pcs etc. For a line filter to be effective it must have a significant impedance and increasing the impedance of the supply does nothing for the sound quality of the system. Also, having reduced the rf noise at the input, we then proceed to add some back in due to the rectification process. The place to filter out all rf noise is on the supply rails after the rectifier. This will reduce both ac borne and rectifier generated noise.

So far as noise on the earth is concerned, the best approach is to use, wherever possible, the computer environment arrangements of 'clean' and 'dirty' earths.

The suggestion of using an additional earth spike is, IMO, a complete waste of time. If I remember rightly, this was suggested some time ago in an article by Ben Duncan and it was claimed that the earth spike dramatically improved the bass definition. The resistance to earth of a single spike is going to be many times higher than the normal mains earth, particularly if the mains earth is bonded to incoming gas and water pipes etc, as it should be (though this is of less relevance in modern properties with plastic gas and water supply pipes).

I have tested a very, very large number of earth spikes on lightning protection systems and to get below a 10ohm resistance usually required an array of multiple spikes each 8 or 12 feet long. The number and length of the spikes depends on the soil conditions, but my measurements were made at several hundred different locations over a wide geographic area and so can be considered representative.

To put the matter even more into context, telephone exchanges required a central earth bar with a resistance to earth of less than 1ohm, before the supplementary bonding of incoming services. To achieve this resistance, the majority of sites required a grid of between 50 and 100 spikes. This is not feasible in the average domestic environment.

As it is unlikely that a single spike will make any difference to the earth impedance, the best approach is to ensure that all incoming services are adequately bonded and that the earth wiring resistance is minimised by uprating the cable size, if necessary, and maintaining tight connections.

Geoff

[Edited by Geoff on 08-30-2001 at 07:25 AM]
 
GRollins said:
Phil,
Conversely, a larger diameter wire carries high frequencies only on the surface, but low frequencies can permeate the entire wire cross section, leading to high frequencies only being able to use a comparatively small percentage of the cross sectional area compared to low frequencies
Grey

The point you are missed is that the volume of a wire increases as the square of it's radius. For example, the volume of copper in a wire more than doubles as you increase the radius from 4 to 6 mm. So even if the current is limited to traveling in the outer region due to a skin effect, there is actually more copper in that outer region than there is in the core. Again this is only for extremely high frequency signals which are easily filtered out by even the most rudimentory RF filters (The power transformer itself is probably sufficient to filter these frequencies).

Phil
 
Re: Are we believing the marketing hype?

Pete Fleming said:
G'day Grey,
It seems ironic that audiophiles as a whole latch on to simple things with dubious objective merit simply because they are easy to understand, yet ignore factors very much know to have a marked affect on the quality of the end product.

Peter I agree totally with you. I have seen people spend ridiculous amounts of money on speaker cables which they then use to drive speakers with passive crossovers in them. Why worry about a few milliohms of impedance in a cable or connector when the crossover is going to add 1000's of times that much impedance to the system (and as a bonus add some nasty phase shifting around the crossover frequency).

I have a confession to make. I'm an EE (electrical engineer) and my day job is designing signal processing equipment. A big part of what I have learned by doing this is to that it pays to look at systems in their totality and determine where improvements will give the biggest benefit before you start changing things. I often see a bad case of "can't see the forest for the trees's" from audio "tweakers". Once one part of the system has been improved it is a good idea to step back and reexamine the rest of the system instead of just mindlessly trying to keep improving the same thing over and over again. Fixation on one particular aspect of a system to the exlusion of the rest leads to very diminishing returns for your efforts.

It is human nature to feel that one is unique and has been given special knowledge that is only shared by a select few. Entire industry are built to cater (and profit) from this. A plausible explaination makes it easier to sell the latest $$$$ widget, so concepts are often borrowed from other disciplines that have no relationship to audio to explain the supposed benefit.

This is where concepts like skin effect in audio cables came from. In reality you have to get above 100 MHz before this effect becomes measurable and you only have to take it into account in your designs when you get are above a GHz. So obviously, skin effect must be the reason why those $2000 speaker cable do such a great job of "reducing the fractal component of the treble response thereby increasing the transparency and shimmering 3D imagery ..." I may want to copyrite that phrash, it could come in handy selling some widget I create in the future. ;^)

The fact is that electronics is a science not an art form. In reality, there isn't nearly as much mystery or arcane knowledge involved with designing audio gear as the average audiophile thinks there is. I can see it now, disclamers on the packaging "no amphibians where harmed in the making of this cable", I guess "eye of newt" is no longer the hot ingrediant. Both Douglas Self and Rod Elliot have sections about the excesses of the high end audio industry on their web pages, it makes for fun but infuriating reading.

http://www.dself.demon.co.uk/
http://www.sound.au.com/

Check it out with an open mind and see what you think.
 
So why do so many accept the hype?

Yes I agree. I find it intriguing that people would willingly pay 2000 British Pounds for a pair of interconnects (I’m presently looking at the advertisement), ignoring the fact that the signal has already travelled through many metres of PCB trace. As mentioned, I’m not suggesting the cables don’t make a difference, but why not instead upgrade the trace? When I was in the industry I would often solder wire onto the trace to repair a cracked PCB, so it’s certainly possible.

Having said the above, I must reiterate my stance that just because something cannot be measured doesn’t mean it doesn’t exist. Likewise, just because something has good specifications doesn’t mean it will be musically rewarding. Nevertheless I believe many manufacturers of audio accessory products really are treading a very fine line between fact and fiction. It seems we simply accept being lied to by manufacturers, passing it off as "marketing hype", however with the cost of some products becoming absurd is it time we begin asking for serious evidence to support some of the claims made? It seems a shame that some manufacturers are benefiting from such tactics, while others, possibly with genuine advances to provide, fail to remain profitable.

Cheers,

Pete
 
Pete,
Yep, the pricing and hype on wire (and a lot of other audio stuff for that matter) verges on criminal. But as I've vented on this sufficiently elsewhere, I won't take up space here with it.
Now, I happen to agree with something that you said in an earlier post, to wit: Is skin effect just a convenient label for some other audible phenomenon? Quite possibly. After all, the numbers indicate that skin effect is of negligable importance at audio frequencies. Could it be that thin gauge (single or multi-strand) wire has some other effect? Dunno. Can't think of one. That doesn't mean that it isn't there. But in the meantime, I'm content to call it skin effect until someone comes up with a plausible idea that has a different convenient label. Yes, I'm aware that this is potentially counter-productive since, if you think you've got it tagged, you won't go looking for something new. But I hope that I'm flexible enough that if someone pops up and says,"You know this thing you've been calling skin effect? It's actually caused by a different effect," that I'll be able to absorb the new idea on its own merits.
Phil,
If:
a) The wire has x cross section and carries both a low frequency and a high frequency.
b) The low frequency has available to it the entire cross section of the wire.
c) The high frequency has available to it only, arbritrarily, 30% of the area of the cross section due to skin effect (for the moment, we'll set aside question of whether skin effect is an audible effect).
Then:
d) The resistance seen by the low frequency is determined by the resistivity of the metal used as a conductor, the length of the wire, and the entire cross section of the wire, whereas the high frequency has, by definition, a higher resistance due to the fact that it has available only a smaller portion of the cross section, since resistance is inverse to conductor area. (Your statement that volume increases with the square of radius is quite correct--you condemn your own argument with that one point. DC has all the wire, whereas high frequencies have only part of it...so the larger the wire, the more pronounced the difference between DC [or low frequency] and high frequency conductivity.)
If you truly feel that your conclusions regarding skin effect and proper gauge of wire are correct, then rush to the patent office and lock it down in your name.
Geoff,
It'd be nice if MOVs had some form of indicator--perhaps changing color due to the heat dissipated?--that would serve to indicate when they were beyond their useful life. That would help take the guesswork out of when to replace them. I have no opinion, myself, on audibility of MOVs, but have noted that some people (I think Phil mentioned this above, and perhaps others) think that they have deleterious effects on sound quality. Unless and until I play with them, I'll refrain from taking either side of the debate.
Note that I'm not necessarily saying that a ground spike will make someone's system sound better. I just happened to remember my friend's system, and even then he only put in the spike because his house was old enough that he had ungrounded sockets. An electrician told him it'd be easier than running fresh wire all the way back to the breaker box. I don't recall whether there was any difference (bass, silence, or any other characteristic) in the sound. He was revamping his system at the time, and any sonic effect (good or ill) was swamped by all the other changes he was making.
The resistance to ground through a ground spike might be higher...granted...but if it's to a 'quiet' as opposed to a 'noisy' ground, it might still lead to an improvement. What do you think?

Grey
 
GRollins said:
Pete,
he only put in the spike because his house was old enough that he had ungrounded sockets. An electrician told him it'd be easier than running fresh wire all the way back to the breaker box.Grey

Ouch! That is a serious code violation and could get someone killed. The reason for the ground in a wall outlet is not for noise control but to provide a redundant return path for fault currents so the circuit breaker will trip in case of a wiring failure inside the protected equipment.

The only safe way to use the older 2 prong wall outlet with modern metal cased equipment is to use a cheater plug that has the ground pigtail wire connected to the center screw of the outlet (which per NEC code is supposed to be connected back to the breaker panel ground bus). A seperate gound rod may have too much impedance between it and breaker panel to trip the breaker during a fault condition. If this happens the person who touches any metalic part of that gear is going to get shocked, potentially fatally. This is the reason why guitarists keep getting electrocuted during gigs. They love to use old tube amps that don't have proper safety grounds and when something goes wrong inside the amp they become the ground path.

Your friend got seriously bad advice.

If you really want to clean up your power wiring and are willing to pay a few bucks to an electrician, have him run a dedicated circuit (line, neutral and ground) from it's own circuit breaker directly to your audio equipment and don't plug anything else into that power outlet. I routinely specify this for data collection systems installed in factories (I was a field service engineer before I got into design) and it helped reduce ground noise which is a real problem when measuring small signal levels, thats also where I first learned about MOV suppressors polluting grounds.

Phil
 
Re: An alternative view

Geoff said:
I do not believe that line input filters should be fitted to audio equipment, they should be limited to that for which they were designed, pcs etc. For a line filter to be effective it must have a significant impedance and increasing the impedance of the supply does nothing for the sound quality of the system. Also, having reduced the rf noise at the input, we then proceed to add some back in due to the rectification process. The place to filter out all rf noise is on the supply rails after the rectifier. This will reduce both ac borne and rectifier generated noise.

Currently, all my equipment is commercial off-the-shelf (Parasound, NAD, Denon and Sony). None of it is modified. In this case, the only way to get rid of the power line noise is externally. The line filters I'm using have current ratings significantly in excess of the current draw of my amplifiers. This was intentional, for the purpose of minimizing the increase in impedance of the power line. Also, I placed significant amounts of capacitance (0.22uF) across the power line, as it should help to remove noise, without increasing the 60Hz impedance.

If I were building my own equipment, I'd certainly do my best to add adequate noise rejection to the power supply. This would include large capacitor banks, bypassing with film capacitors, and an inductor (PI filter). I agree that it is better that way.
 
Geoff said:
You have a valid point about commercial equipment. Internal modifications play havoc with the warranty.

The warranty isn't the issue. In my power amps (where external filters have the largest negative impact), there's not enough room to do much. Commercial equipment is made as small as possible, to save on cost.

On the other hand, DIY equipment doesn't usually worry about space. If you find you need a change, there's usually room to make it.