High Frequency Shunt for In-wall Power Wiring

For the same reason others have mentioned, I never considered it practical to filter the entire supply load.

What I am looking for instead is to provide a low impedance path, from the building's wiring grid to ground, for frequencies higher than 50/60Hz. From the responses so far, am I to assume this is not technically possible? Yet, it is common practice within electronic devices, capacitive decoupling, amplifer LPF's, etc. I see no difference in the basic concept. It seems to be just a matter of scale and component selection.

To reiterate, this item is not intended to be a conventional ripple or surge filter on the mains grid. It is a shunt for a relatively low amplitude of HF noise riding thereupon. I believe I have already explained that, for this app, it is not desired to incorporate this capability into each and every connected device.

pnielsen said:

From the responses so far, am I to assume this is not technically possible? Yet, it is common practice within electronic devices, capacitive decoupling, amplifer LPF's, etc. I see no difference in the basic concept. It seems to be just a matter of scale and component selection.

Click to expand...

Common practice - but NOT how you seem to be imagining it.

To get any decent amount of rejection it has to go through components (as do all the examples you gave), as well as shunted to ground.

As I've already explained, you would need to do this on the incoming supply - hence big and expensive - otherwise individual filters on each device that needs filtering (which is why it would make MUCH more sense for it to be internal).

Expanding on what Nigel says above, simply providing a shunt path to ground merely results in a higher current drawn from the mains, if the mains wiring is assumed to have negligible resistance. To get attenuation you need to form a voltage divider, i.e. a series resistance/impedance as well as the shunt.

alec_t said:

Expanding on what Nigel says above, simply providing a shunt path to ground merely results in a higher current drawn from the mains, if the mains wiring is assumed to have negligible resistance. To get attenuation you need to form a voltage divider, i.e. a series resistance/impedance as well as the shunt.

Click to expand...

What about a different configuration? When I look at the schematic of a series resonant tank circuit, I do not see the load current passing through the filter. IOW it only needs to be rated for the relatively small quantity of HF current, not the 50/60Hz. An array of such, providing overlapping low Q's, might serve my intended purpose.

one thing you need to consider, is that the higher you go in frequency, the less that a house's ground system IS ground. if the electrical box is in the basement for instance, a single ground wire from the box to a second story outlet might be 20 or 30 feet long. using the 20ft length as an example, would be 1/8 wavelength at 6.15Mhz, and the ground wire could be picking up a significant amount of energy. if you have a company nearby that uses an induction heater at that frequency, your "ground" at that outlet could be several microvolts to several hundred microvolts above ground. your ground wire then would be re-radiating some of this signal. your "ground" is no longer an effective RF ground, since a shortwave receiver with a 1 microvot sensitivity would definitely pick this up as a signal well above the noise floor.... and this would be WITHOUT an antenna connected.

you can find more info here: http://radioworks.com/nbgnd.html


also, for what you were describing, the big iron answer would only be a first step. the reason a lot of equipment carries it's own line filters is to keep RF from leaking OUT of the equipment. this is especially true when the power supply is a switching supply. but the use of many filters, especially when the equipment has an analog supply, is to keep RF out, and as you add more equipment of this type, you get kind of a "distributed filtering" effect, which would be an efficient follow up to the big chunk of iron and copper in the basement. to really make it effective, you would probably need to make the exterior of the building into a faraday cage. your house would look awfully funny covered in chicken wire...

pnielsen said:

What about a different configuration? When I look at the schematic of a series resonant tank circuit, I do not see the load current passing through the filter. IOW it only needs to be rated for the relatively small quantity of HF current, not the 50/60Hz. An array of such, providing overlapping low Q's, might serve my intended purpose.

Click to expand...

Again, as you've provided almost zero information we can't really comment, but under certain specific circumstances, and with exact known single frequencies of interference, you could perhaps notch them out in that way.

Don't forget, low Q means low attenuation.

unclejed613 said:

one thing you need to consider, is that the higher you go in frequency, the less that a house's ground system IS ground. if the electrical box is in the basement for instance, a single ground wire from the box to a second story outlet might be 20 or 30 feet long. using the 20ft length as an example, would be 1/8 wavelength at 6.15Mhz, and the ground wire could be picking up a significant amount of energy. if you have a company nearby that uses an induction heater at that frequency, your "ground" at that outlet could be several microvolts to several hundred microvolts above ground. your ground wire then would be re-radiating some of this signal. your "ground" is no longer an effective RF ground, since a shortwave receiver with a 1 microvot sensitivity would definitely pick this up as a signal well above the noise floor.... and this would be WITHOUT an antenna connected.

Click to expand...

OK, but that is a singular resonance determined by a design fluke. The ground would still be effective for the vast majority of other impinging frequencies.

Since you seem to be knowledgable in this field, can you please explain why the solution described in my previous post is unworkable, and still requires "a big chunk of iron", as you put it? Or is this simply a case of never been done ... so we don't really know?

pnielsen said:

OK, but that is a singular resonance determined by a design fluke. The ground would still be effective for the vast majority of other impinging frequencies.

Since you seem to be knowledgable in this field, can you please explain why the solution described in my previous post is unworkable, and still requires "a big chunk of iron", as you put it? Or is this simply a case of never been done ... so we don't really know?

Click to expand...

so with the typical house wiring, you get multiple "design flukes" at many different frequencies.... it's not a "design fluke" it's a natural property of any length of wire. a 1" piece of wire resonates at 11Ghz and becomes a "significant portion" of a wavelength long at 1.1Ghz. "ground" is only "ground" if the wiring length is less than 1/10 wavelength long for most applications of the term. as you get longer than 1/10 wavelength, the wire begins to turn into an antenna, and develops partial standing waves. longer than 1 wavelength and you have full standing waves and partials. there is no way to avoid it.

This shows the important effect of different series resistance (impedance) values in the path from the supply to the 'filter'. I made no attempt to tune the LC circuit to any specific frequency; the principle is the same for any frequecncy.

For 0.1Ω assumed source impedance there is virtually no attenuation at the resonant frequency, whereas for 10Ω (much higher than a house wiring system should have) there is ~22dB attenuation. Clearly, as has been said above, you need to pass current through a high-AC-impedance series component (i.e. huge choke) at the house mains input to get any significant filtering.