Mains/earth ground

[Atomicrox]

Today we finally had an electrician change my house's power meter (the one used by the electric company to bill us) to a newer model.

This included installing a mains ground. Until a few years ago pretty much every house here had an ungrounded 2 wire mains system: neutral and a 110V phase.

The electrician tied the ground "rod" (actually a L-shaped galvanized stake) to the neutral wire from the distribution system. He also tied the ground rod to a wire that goes to the "third prong" on electric outlets.

I don't know much about grounding, but that seemed strange. What's the point of having three-prong outlets if the ground and the neutral are tied together at the ground rod?

I asked him and he said that's what local electrical regulations require.

Is that the correct way to do this?

 

[USAbro]

That's how I've always seen it done, right before the main hits the breaker panel. Although come to think about it, it does seem a little screwy. Maybe it has something to do with the direction the power flows.

 

[Sigurthr]

That is indeed how it is supposed to be.

The neutral bus is tied to earth ground to prevent a floating neutral/chassis situation; many appliances have their chassis tied to the neutral prong, allowing neutral to float in reference to ground can result in a shock/electrocution hazard (user at earth potential, neutral at above/below earth potential).

The third "ground" prong is tied directly to ground to provide a return path in a fault situation where either the hot or neutral is compromised. Additionally, many appliances have their chassis tied to ground, and rely on this for faraday shielding and RF ground-plane and ground reflection. This wiring is usually not meant for high current, typically half the wire gauge of the Neutral or Hot lines. It is simply meant for fault conditions and to prevent floating potentials.

The location of the ground connections is also important. There should be a high current ground connection to a local ground rod at the distribution transformer. This is to shunt lightning strikes and line faults to earth prior to reaching your home's panel. There should be a direct bond between the electrical panel's ground bus and a home ground rod to shunt lightning strikes to earth before going through your house's wiring. Neutral is tied to the panel's ground bus, and by extension to the ground rod for the above reasons.

Depending on the soil conductivity at the location of the house's ground rod, the length of the rod, the diameter and type of cabling connecting it to the panel, and several other factors, this connection may actually be a rather high impedance to true earth. This can cause transient floating situations in the house's wiring. This scenario also precludes use of the ground prong in household wiring for any RF applications, such as Amateur Radio and Tesla Coil work. Additional ground rods can be added to greatly improve the earthing, simply by spacing rods out evenly and either connecting each additional rod directly to the main rod, or by connecting every rod to every other rod (depending on how many you add this tends to be less efficient). If your house is large or if the location you plan on doing RF work in is far from the electrical panel, you may find the household wiring presents too large of an impedance to earth as well. A second "bank" of ground rods directly exterior to the RF work station can be used to drastically lower the impedance seen by the Rf equipment.

Of particular interest to TC work; you should RF-couple the hot, neutral, and ground line to each other using Class-Y and Class-X rated capacitors. This prevents any RF voltage differential from developing across any two of the three conductors, protecting your equipment, as well as reducing the perceived impedance to 1/3.

 

[Atomicrox]

I'm glad it's correct. I never really thought of the ground wire as a failsafe for a compromised hot or neutral, only as a way to keep chassis at ground potential and redirect lightning/surges to earth. Still seems a bit strange to me, though. It's a lot of extra wires and outlet prongs for something which should ideally never be used

Do you know of a simple way to measure the ground (not the soil, the whole "system") quality?

TC work is part of the reason I asked. oneTesla uses mains ground as RF ground, which I've always read was a bad thing. The wire used between the rod and the panel is very thick, but the wire between the panel and outlets is too thin. I'm quite sure it'll be a problem...

 

[Sigurthr]

Do you know of a simple way to measure the ground (not the soil, the whole "system") quality?

TC work is part of the reason I asked. oneTesla uses mains ground as RF ground, which I've always read was a bad thing. The wire used between the rod and the panel is very thick, but the wire between the panel and outlets is too thin. I'm quite sure it'll be a problem...

Yeah it can be measured but there's no real easy way. Same way you measure any impedance, really. Take a voltage source, a current shunt, and a sensitive meter, and measure the voltage drop across the shunt. In this case you need a low impedance RF source and to use the rms voltage drop, so either a high quality dmm rated for true rms at RF frequencies or an oscope. A working SSTC can be used but it's difficult since the secondary has a high resistive impedance. You'd have to model or calculate the expected impedance and compare that to the measured over all impedance to get a ballpark of the unearthing impedance. Not all current flows through the secondary base, but most does. You don't need exact impedance to determine unearthing quality anyway, just a rough order of magnitude estimate.

I had to run a custom ground line to a ground rod array from my office. I measure around 30 ohms at 1Mhz, which is pretty darn good for the length of cable used. At normally lower TC frequencies it's even lower impedance.

A good inexpensive ground cable is RF rated amateur radio coax. Just bond the core and the shield at both ends.

 

[Atomicrox]

Not really worth the trouble of measuring, then. Probably easier to just add more rods.

Do you figure this ground is good enough for the oneTesla? They're using the third prong in all their designs.

 

[Sigurthr]

The truth? No, it isn't. The "fudged" truth; you can get away with it. Don't expect even near 100% output or expect it to not have a VERY high chance of frying household electronics though. You present a high impedance to the secondary base and what you wind up with is those RF currents capacitive and inductively coupling to the mains wiring.

Here's a homework assignment for you. Estimate the wire gauge of the third prong wiring. Estimate the rough length of wire between the outlet you'll use and the ground rod installed. Find a calculator or formula for determining the inductive impedance of a wire. Plug in your wire gauge and length, use a frequency of 500KHz. Now you've got a ballpark impedance for that mains ground. If it's over 200-Ohms you're in pretty bad (I'm being excessively generous with the word "bad") shape. You want around 50-Ohms or less ideally, with 75-Ohms being the upper acceptable* limit.

*Impedance matching of the transmission line to the RF source (secondary base) is an important factor, and while a mismatch of too low impedance transmission line to current source can indeed cause a higher than desirable SWR, it in practice isn't much of an issue as other losses are dominant. What is a major factor in practice is impedance mismatches where the transmission line or earth termination points are at a higher impedance than the source. This is why earthless ground planes don't function well for Tesla Coils. To get an acceptable impedance match at TC frequencies the ground radials have to be extremely long to even approach 1/4 wavelength, and ground radials should be at least one half wavelength for even a modest match.

 

[Blarg King]

Sorta related-ish question (sorry for the hijack, but better than making a whole new thread)

I have a small portable generator, and the manual says for safe operation I need to ground it by connecting the metal frame to a ground. Would taking some heavy gauge wire and attaching one end of it to the metal frame and the other to say a copper rod/pipe that's been hammered into the ground work? Or am I completely misinterpreting this?

 

[Sigurthr]

Yep that'll work fine Blarg. 60Hz is practically DC as far as impedance goes. Use a five foot or longer rod.

Someone dig up the old image of a generator and a bucket of dirt. Lol!

 

[Blarg King]

Ah excellent. Im surprised that there isn't a cable for gounding pre attached to the frame with how much the manual says to ground the darn thing.

 

[USAbro]

Is someone building a Tesla coil? If so can you post images?

 

[Sigurthr]

Go see the GIANT thread on it in the Geek's corner. Don't want to derail this further.

 

[Atomicrox]

The truth? No, it isn't. The "fudged" truth; you can get away with it. Don't expect even near 100% output or expect it to not have a VERY high chance of frying household electronics though. You present a high impedance to the secondary base and what you wind up with is those RF currents capacitive and inductively coupling to the mains wiring.

Here's a homework assignment for you. Estimate the wire gauge of the third prong wiring. Estimate the rough length of wire between the outlet you'll use and the ground rod installed. Find a calculator or formula for determining the inductive impedance of a wire. Plug in your wire gauge and length, use a frequency of 500KHz. Now you've got a ballpark impedance for that mains ground. If it's over 200-Ohms you're in pretty bad (I'm being excessively generous with the word "bad") shape. You want around 50-Ohms or less ideally, with 75-Ohms being the upper acceptable* limit.

*Impedance matching of the transmission line to the RF source (secondary base) is an important factor, and while a mismatch of too low impedance transmission line to current source can indeed cause a higher than desirable SWR, it in practice isn't much of an issue as other losses are dominant. What is a major factor in practice is impedance mismatches where the transmission line or earth termination points are at a higher impedance than the source. This is why earthless ground planes don't function well for Tesla Coils. To get an acceptable impedance match at TC frequencies the ground radials have to be extremely long to even approach 1/4 wavelength, and ground radials should be at least one half wavelength for even a modest match.

Here we go. He used two wire gauges, 16mm^2 from the rod to above the panel and 2.5mm^2 from above the panel to the outlets.
Est. 12m from rod to panel, d=4.51mm
Est. 7m from rod to outlet, d=1.78mm

Wiki's formulae (second and third on Straight wire conductor at Inductor - Wikipedia, the free encyclopedia) aren't linear so I had to use the smaller diameter for everything.
With 19m @ 1.78mm I got 37.7uH, which gives 59.1 Ohm @ 250kHz (oneTesla's frequency according to them).

Not that bad, then. Did I do it right?

He did use multifiliar wires, though. Does it matter?

 

[jcranmer]

I never knew this was the way it was done. Learn something new all the time. :beer:

 

[Sigurthr]

Multistranded and single core are identical, even with skin depth calculations, it only matters if the strands are electrically isolated from each other (litz wire).

Your math looks solid.

So, ~60-Ohms to the fundamental at 250KHz, ~120-Ohms to the first harmonic. It's certainly usable. Remember that DRSSTCs are discontinuous, so you have a lot of high order harmonics from the interrupter, and they'll see a much higher impedance. Expect noise to be high, so be cautious of where it is operated and unplug and short the inputs of sensitive electronics the first run.

Compare your ~240-Ohms @ 1MHz to my TC ground's ~30-Ohms @ 1MHz to get a good relative gauge (pun intended). My ground system isn't even considered ideal by DRSSTC standards, so I bolster it by capacitively coupling all the mains wiring together with the Class-Y caps. I have it such that not only are the connectors all kept at equal potential to RF, but the caps act as RF shunts at the power outlet that my equipment is plugged in, both at the Tesla Coils, and at the sensitive equipment in the same room.

In short, I think it's certainly safe to try, as long as you take precautions like I mentioned above (unplugging and shorting the inputs on anything friable). You certainly shouldn't see any appreciable performance degredation, and that's a very good benchmark. The next benchmark up is zero interference to local equipment (my lab's example), and above that is negligible measurable interference and spurious emission to sensitive detectors (professional lab grade example).

 

[Atomicrox]

Sounds like a plan. Do you have any interesting links about those class-Y capacitors?