How To Stay Grounded When You Have Zero Potential

Ground is an interesting topic when it comes to engineering. Either it’s the reference level for a digital circuit (not necessarily at zero volts, either), or it’s the return path for current, or it’s the metal chassis, which shouldn’t be the return path for current or else something’s terribly broken. Erika Earl’s talk at this year’s Hackaday Superconference is all about ground.

The first type of ground to talk about is the ground in your outlets and walls. The AC safety ground is the third pin on your plug that should be attached to the chassis of your washer/dryer on one end, and somehow connected to the neutral wire somewhere near your breaker box. The theory of this being if a conductor touches the chassis of a lamp or appliance, all the current will go along that ground bus saving you from electrocution. It should also trip the circuit breaker.

But really we’re rarely dealing with mains power around here. When it comes to electronic design, we’re mostly dealing with analog grounds and digital grounds in circuits. Sometimes these are the same, sometimes they’re not, but they’re both usually referenced to 0 Volts, Add in some considerations for EMC, and ground loops, and you have an astonishing amount of knowledge wrapped up in having zero potential.

If you want to know about what ground actually is, this isn’t a talk to miss. Erika has tons of experience chasing down grounds as an audio engineer, and her career highlights including the director of hardware engineering at Slate Digital and the Senior Technical Engineer at LA’s legendary Village Recording Studios. There’s a lot of experience here, and if you want to where to find your ground, Erika is the person to ask.

100 thoughts on “How To Stay Grounded When You Have Zero Potential

    1. Ah the Wadsworth constant. Rarely is it ever wrong. I think there’s a browser extension that automatically starts all YouTube videos a third of the way through. The number of times you need to rewind is very low.

      Even though our data tech is getting immeasurably better, we’re just filling it up with bloat these days. Back in the dinosaur era when all info on the web was text-based, I’d bet we sent about the same amount of useful data without all the untold exabytes of video we have to host now to talk about or learn anything online. It’s kind of silly.

  1. “The theory of this being if a conductor touches the chassis of a lamp or appliance, all the current will go along that ground bus saving you from electrocution. It should also trip the circuit breaker.”
    What?! Is that how it’s supposed to work in the USA?

    Usually this does NOT trip a circuit breaker (MCB) BUT a GFCI/RCD[1] as long as the fault current is above a semi-safe threshold (often ~30mA unless industrial environment & other exceptions).


      1. To state the obvious service panel mounted GFCI/ circuit breaker where the first available. The family expert on everything helped out my mom by remodeling her bath room, he went on rant because there was no GFI. Mom ask me about it I told it was protected, and the unit was in the breaker panel, yurns out the expert never knew such a thing existed. Such is life.

        1. Although an isolated outlet, like in a bathroom shaver socket, won’t trip a GFCI in the panel, since it’s isolated by design by a transformer. But that’s deliberate, since there’s so much water and earthed piping around a bathroom. And the point of the isolation is so you’d need to touch both contacts to get zapped.

          In the UK an isolated socket is the only type allowed in a bathroom. A single one, rated < 100W (often quite a bit less), generally by the mirror and mirror light. They're the US type socket, for shavers only, and maybe electric toothbrushes, rather than our superior 3-pin UK type. I think it's a good idea, how much stuff do you need to plug in, in the bathroom?

          1. Ah, you don’t have daughters with hair dryers.

            It used to bother me to see a double insulated (no earth) hair dryer beside a hand basin with taps that are earthed through the plumbing.We don’t have bathroom isolators here and a hair dryer is about 1200 Watts. Any water could render the double insulation of the hair dryer useless.

            Also, it’s not true (in different situations) that you need to “touch both contacts” to get zapped.

            I saw someone receive a serious shock when trying to remove a broken incandescent bulb from a fixture when the circuit was live. They were standing on a wooden ladder. They put their left hand against a mirror for balance. The mirror was attached to a walk-in fridge which was earthed so the reflective and conductive surface on the other side of the glass was earthed. Due to their hand being flat on the mirror glass there was was sufficient surface area and proximity for their hand in combination with the reflective surface to form a capacitor able to pass sufficient AC current to shock them. They had to be defibbed.

          2. Rob, in the case of isolating transformers, you do need to touch both “contacts” to be zapped, that’s the point of them. To isolate the mains voltage from earth for safety. But it was worthwhile reading your Heath-Robinson tale of the silvering on a mirror capacitatively coupling enough current, even at 50Hz (right?), to fibrillate some poor soul.

            I’m astounded by the concept of “walk-in fridge”. Like, domestically? Or just a commercial one?

          3. @[Greenaum]

            I do understand what your saying and I do understand what an isolating transformer. I have extensive experience.

            Your statement assumes that there can be only one fault.

            So I will explain, based on one assumption.

            Assumption: Electrical faults occur.

            Given that faults occur the next question is – Is the probability of a fault occurring in any way dependent or influenced by any previous and concurrent fault.

            With exceptionally few exceptions the answer is – No.

            If there were a testing procedure in place then probability of a second or more concurrent faults would be reduced but also if that were the case then the previous fault(s) should not be concurrent as it should have been detected by the testing procedure and rectified.

            The next question is what is probability of singular or multiple concurrent fault conditions by number of faults.

            The graph of this is close to an inverse exponent staring at x = 1 and y = sigma (1 fault) and ending at x = (max faults) and y = sigma (max faults).

            The next question is – what is the probably of harm, injury or death by number of concurrent faults.

            This graph is a gradual exponent with the probability of harm, injury or death increasing with the number of concurrent faults.

            So my point is this –

            If you assess safety on the basis singular faults of various types then you are making an assessment of potentially far more complex fault conditions based solely on those simple fault conditions that are less likely to cause harm, injury or death.

            All of the above is useful to know but it is a mere semantic if it is not integrated with the probability multiple human errors.

            Harm injury or death results most often from multiple faults and human error is the reason that the previous and concurrent faults were not rectified.

            An example – A RCD will trip at 25mA to 30mA, if it’s not faulty – go test the thing periodically.

    1. It works both ways…around wet areas you’ll find GFCI protection on newer construction, but not everything is that new and only brought up to current code if work requiring permits is to be done (and then a few bills might get you out of it).

      1. Interesting…Given how inexpensive the units are now, what sort of fool would spend a few bills to get out of it? I have heard many complain how much of a PIA they are, I suspect there where other issues with their work. I have had them in my well pit for 30 years, and have yet to gown down there to reset a tripped GFI.

        1. You should be test tripping it by the manufacturers specification, usually between 6 and 12 months. At 30 years, I would be inclined to say that I have had good service from that device and replace it.

        2. I’ve seen GFCI trip several times. Most common cause is due to a surge protector being hooked up. The TVS leaks at the peak of the waveform shunting just enough current to ground to trip the GFCI.

    2. The idea is that if you short a hot wire to ground, the low resistance path through the chassis will draw more current than the circuit is rated for and trip the breaker. Yes, this doesn’t always happen, but often this is the normal fault.

      I have had this happen a number of times. Once when installing an electric range outlet and one of screws holding the outlet to the box penetrated the insulation on one of the hot wires. I turned the breaker on and BOOM!. The result was electric arc gouging for a fraction of a second before the breaker popped. The screw was shortened by about 10mm. (3/8 in)

      1. Yup. The same principle works even with simple wire fuses. Hot to earth means unlimited current flows and pops the breaker or blows the fuse. At least assuming the fuse is the weak link in the chain. And it’s still not enormously safe. An ELCB / RCD is much more satisfactory, but back in the old days the earthed chassis was an obvious and simple safety feature.

        I’m amazed and suspicious that 3-pin separately earthed sockets are quite a recent thing in some countries. British standard has been the best in the world since some time around WW2. Safer in every way, except if you step on one!

    3. From the wikipedia article “A residual-current device (RCD), or residual-current circuit breaker (RCCB), is a device that instantly breaks an electric circuit…”

      Sounds like a gfci is a circuit breaker.

      1. No.

        “Circuit breaker” is always taken to mean “overcurrent limiting device”, basically an electromagnetically & thermally operated fuse. Draw too much current and it will open. Basically it’s there to prevent cabling fires caused by overloads.

        A really good short to ground will happily pull 50A+ and pop a breaker nice and quickly. However if you stick a finger into a live appliance or take a bath with your hairdryer, you’re only going to pull about 100mA while you’re dying and the breaker won’t open. You are, after all, probably not on fire; the breaker can’t tell the difference between a human load and a lightbulb.

        A GFCI measures the difference between Live and Neutral currents, and opens if that difference exceeds about a couple of mA. The idea is that if the L & N currents differ, there is some current going somewhere else, and that should never ever happen. If there is current going elsewhere, it could be going through a human, so best to stop it.

        GFCI’s, by measuring current difference, are great because they can be set at a threshold lower than that required to kill someone, regardless of how much current the load uses. A GFCI won’t save you from getting directly between L & N though, so don’t do that – not that it’s easy to do so accidentally.

          1. No, they are fundamentally different.

            An ELCB measures the current/voltage through the PE (protective earth) going through it at the “drain” -> only stray / fault currents going this way are detected.

            A GFCI / RCD detects ALL stray / fault currents by measuring any power differentials at the “source”.

            They apparently were all named ELCB somewhere sometime ago but that was dangerously confusing and the IEC “decided” to separate the terms/devices

        1. So… a circuit breaker is an overcurrent limiting device, while a GFCI is a device that limits a current to be under a certain threshold? Right. Good distinction. ((Dis-)claimer: I am aware of the differences of the two device classes – i just realized the difficulty of putting it into a few simple words and was amused; no criticism intended)

          1. A breaker limits total current. A GFCI makes sure that the current going in/out on hot equals the current coming back in/out on neutral.

            Or… a GFCI makes sure that no current leaves the circuit’s loop (i.e. through you to ground).

            That’s the best I can do with a couple sentences.

      2. In a way it is, in a way its not. A circuit breaker is basically just a coil with a mechanical trigger plus a bimetal. Once the Current is high enough, the trigger flips the breaker open. (Or if you add too much load a bimetal will trigger the circuit breaker)
        So a circuit breaker can turn off shorts and over current.

        A RCD on the other hand is more of a monitoring device.
        The main part of it is a coil with the mains wires (like L,N or L1, L2, L3 and N) around. If everything is fine the overall current in all wires is 0, because the current flowing in from one wire is the negative of the current flowing out of the other.
        If you short to ground, the currents (L/N or L1/L2/L3/N) don’t add up anymore and the RDC will trigger and turn off the circuit (break it).

        RCDs and circuit breakers are used for different types of faults.

        A breaker is only used for cables and simple loads, to protect everything from shorts or overload.
        If you connect a chassis to ground, you want to create a low impedance to ground in case of any kind of fault (usually insulation) to create a short.
        In this case the short circuit triggers the breaker and power is turned off.
        This should prevent any damage on your equipment and any kind of damage to you by touching a bare metal chassis of any faulty tool or equipment.
        If however the current is not enough to trigger the breaker (corroded contacts or faulty insulation), your case is still connected to a few hundred volts and in reach of your hands.
        (A regular B16 breaker needs at least 20 to 22 amps to turn off! This will take some time)

        Here is where the RCD comes handy. A fault current flowing through ground (green/yellow wire or just any kind of GND connection) will trip the device above 30mA of fault current (remember the 20A from the breaker?).
        This will prevent an excessive amount of exposed voltage on your chassis, because the voltage will be shut off immediately (more or less).
        Since the RCD only needs those 30mA it is faster and more reliable (and can be used to prevent fires! Insulation faults with circuit breakers can generate a lot of heat since you need 20A to turn off mains. An RCD only needs 30mA).

        Since RCDs are only monitoring currents flowing in and out of a circuit, it wont detect any kind of short between the mains cables on its output.
        A working RCD will never, ever detect any kind of short between mains.
        That is a huge difference between a circuit breaker and an RCD! (you should have both installed, if you don’t you should get one or probably fire your electrician)

    4. This is the confusion I was expecting with these devices.

      ELCBs are 50 years old and they were used for a long time before residual safety switches where even thought of.

      Almost every device mentioned in comments here are variations of GFCIs.

      The term ELCB once erroneously included residual current devices.

      Over time and in different locations RCD has meant – firstly: Reselable Circuit Device (often a over current circuit breaker but any other form of resetable device) and then later: Residual Current Device (without an over current breaker).

      Later Residual Circuit Devices were more commonly know as Residual Current Circuit Breaker and that is causing confusion because a “breaker” IS an over current device (resetable fuse). The reason for this is that modern RCCBs actually include a over current breaker as well as a Residual Circuit Device.

      The distinction between a RCCD and an ELCB(R) is this.

      An ELCB is triggered by the earth wire connected to the ELCB (and not directly connected to earth) is raised by a voltage, usually about 50Volts.

      A Residual Current Device or RCCD is triggered by a difference in current between Active (Live) and Neutral, usually 15mA to 30mA.

      In the case of a RCCD, it is ALSO triggered by over-current to the load like a resetable fuse.

      1. Uh now I’m completely confused on the terminology. According to Wikipedia a RCCD/RCD with over-current protection is a RCBO (EU) / GFCI (US) but farther down in the comments I got the impression that GFCI is the umbrella term?!

      2. One more term for you: “Combos”. These are combined RCD/Circuit breakers that both protect the cables from overcurrent and protect people from mA fault currents. They are just about universal in any new installation in Australia as they take up less room and are more convinient to install or retrofit in switchboards than separate RCD’s and MCB’s. RCDs have been mandatory on outlets and light circuits for years here and with the newly released AS3000 wiring rules, manditory on just about every other LV circuit too.

    5. At work we have IT system (Insulated Terra) that works fine even when one phase is grounded but you need to find that fault before other phase will get grounded as this fill close the loop and eventually trip something.

      When I started working an old electrician told me the way they used to find ground fault on a barge he was forking for around 40 years. They used not grounded phase, connect thick conductor to it and throw it on the floor (for “safety reason” it is important to throw it behind you because of arc that could hurt your eyes) and monitor what trips (or explodes). He was full of similar wisdom and tend to completely ignore any good practice. I was really surprised he made it until he retired considering how often he was working on high energy equipment with his careless approach and total lack of understanding of electricity.

    6. It all depends where and which conductor is shorted. If it is a direct short from hot to the chassis or through a low resistance path, it will trip the breaker. If the neutral is shorted to the chassis, you could have the neutral AND/OR ground conductor both carrying current depending the their relative resistance to current flow. This is where a GFCI comes into play and would trip.

      1. Easiest explanation is that a standard circuit breaker is simply limiting the current moving through the hot conductor. The GFCI is looking for the difference in current on the hot wire vs current on the neutral wire. The difference indicates current flowing through an unintended path.

        1. GFCI are not universally used because loads that are inductive (like motors) can false trip a GFCI. Look at it this way, when you are building a magnetic field you have current flow on the hot side that does not appear on the neutral side until the field has stabilized. Switched motor loads are most likely to fake out a GFCI but any device that is inductive can do this.

          1. This is not true. Inductive loads have nothing to do with tripping GFCIs. As long as all of the current from the hot wire returns through the neutral wire, you won’t trip a GFCI. The problem comes when there is significant capacitance to ground. This can be a problem with big motors because a) there’s a whole lot of wire with very thin insulation right next to a grounded chunk of metal, and b) sometimes the installation requires a physical capacitor to be connected from neutral to ground. In either case, some of the current intentionally returns through ground, and THAT is what trips GFCIs.

    1. You laugh, but glitches flying around on a vehicle can give you problems. I have a focus that occasionally resets the ambient air temp indicator on the passenger side mirror to absolute zero (or I’m told in the forums). Which shuts off the A/C until you reboot. I looked it up on the wiring diagram …. “grounded” somewhere in the car. A properly grounded sensor wouldn’t pick up glitches.

      1. Well, lower limit of what the sensor reads (reset to all zeros, whatever that was). If it’s below a certain (unknown) temperature outside, the A/C shouldn’t be on, and so the confuser turns it off.

        The sensor is outside on the mirror, reading the ambient air, not the cabin air. There’s another which reports ambient air temp to the dash display. I’m sure there’s a reason.

  2. There is very poor international standardization of these acronyms. Even worse there is poor understanding of how these work.

    Also there are different “earth” systems in use in various places across the globe.

    RCD in some places is Resettable Circuit Device (industrial), in other places it’s a Residual Current Device (domestic/residential).

    A differential current breaker has been called a Residual Current Device (RCD domestic), Ground Fault Interrupter (GFI commercial), Ground Fault Circuit Interrupter (GFCI industrial) however none of these devices have a “earth” or “ground” connection. Instead they measure the difference between active and neutral currents.

    On the other hand the older Earth Leakage Circuit Breaker (ELCB commercial) or Earth Leakage Circuit Breaker Resettable (ELCBR industrial) have the earth go “through” the device so that the specific earth current can be measured. This is because in commercial or industrial environments there are multiple phases which makes differential measurements more prone to problems.

    To explain my terms –
    Domestic/Residential – homes shops offices where there is a single or split phase using neutral return or Multiple Earth Neutral distribution (MEN) schemes.
    Commercial – laundromats construction or fabrication heavy machinery that use three phase delta distribution with Multiple Earth Neutral (MEN) distribution schemes.
    Industrial – generation plants power distribution stations which use a mix of star and delta multi-phase distribution schemes.

    What I often point out about current imbalance devices (your common garden variety “safety breaker”) is they there are three way a human can connect to this supply system, one way has no effect on the human or safety breaker (N-E), one way has no effect on the human but trips the breaker (A-E), and the other way kills the human and has no effect on the breaker (A-N) so don’t overestimate the “safety” these device provide.

    1. “… Earth Leakage Circuit Breaker …. have the earth go “through” the device so that the specific earth current can be measured. This is because in commercial or industrial environments there are multiple phases which makes differential measurements more prone to problems.”

      Uh what? They can ONLY measure the “earth” current going through them, NOT the current going through you and directly to earth (through anything else that is earthed but not connected to the ELCB). -> I’d assume a GCFI reacts in more fault conditions than an ELCB (under the same outer conditions).
      The only practical way (I can see) to measure ALL fault currents in a power distribution system (with one earthed/grounded phase) is a GFCI and the physical concept behind them is always the same regardless of how many phases you use. How do multiple phases “make differential measurements more prone to problems”?

      I’m missing on terms explanation – whats “A”? I’ve only seen L=Line=P=Phase=(H=Hot) to N=Neutral and PE=ProtectiveEarth so far.

      Actually the fault condition (P-N) often triggers the GFCI as well because you’re usually not floating isolated in the air while touching P+N -> some current finds another way to earth/PE and you have the necessary imbalance to trigger it. It’s still the most dangerous fault.

      1. GFCI is an umbrella term for most of these devices. There are 2 sub categories – imbalance detection and earth leakage detection.

        I have only seen ELCBs in environments where all staff have at least a minimum professional education such as technician or engineer. There is no opportunity for current to flow to a different earth and the ELCBs are for different sections and the earth from each section comes back to the ELCB just as does the phase and neutral.

        Differential currents are easy to measure in single phase configuration. However in star and delta configurations the current drawn can be significantly out of phase to the voltage.

        This also happens to some degree with single phase and is why inductive loads like a fridge compressor will nuisance trip an imbalance “RCD”. It is worse with star 3 phase and even worse again with delta 3 phase.

        ELCBs don’t suffer this problem and they have some advantage in some environments but you certainly would not use them in a domestic environment for the reasons you described.

        The terms I was using for my country are –
        single phase
        A – Active
        N – Neutral (Multiple Earth Neutral)
        E – Earth
        3 phase commercial
        L1 – Phase 1
        L2 – Phase 2
        L3 – Phase 3
        N – Neutral (return)
        E – Earth
        3 phase industrial
        S1 – Phase 1 star Wye or Y
        S2 – Phase 2 star
        S3 – Phase 3 star
        N – Neutral return
        D1 – Phase 1 delta
        D2 – Phase 2 delta
        D3 – Phase 3 delta
        E – Earth

        1. Good eplainations, so where would the latest requirement fall in, arc fault circuit breakers? momentary detection of high amp draw followed by less current draw but somehow able to detect difference between incandescent load or starters.?

          1. I’m not the best person to ask about modern requirements as I left this field about 30 years ago.

            I worked in commercial and industrial (power generation and distribution) up to about 127kV so I saw a lot of things that most people never see.

            Domestic and commercial have leaped ahead with technology in these devices where industrial still uses a lot of old technology like explosive extinguishment that will never be used in other fields for obvious reasons.

            I have looked at modern low voltage (<500V) arc breakers and they're fascinating with their energy absorption techniques and use of chemicals/elements.

            When I did work in that field, safety standards were poor and personal safety was left up to the individual. The reason I am here to talk about it is a combination of RTFM, a respect for safety and unfortunately common sense. I say unfortunately as this one aspect is the reason that there are several others that are not here to talk about it.

            One example would be comparing dropping a shifter on a MV bus bar (about 12kV – 18kV) and dropping a stick of gelignite on the same bus bar. The gelignite is not likely to explode whereas the shifter certainly would. The exploding shifter would also release 4 times the explosive force than a stick of gelignite can.

            Ironically I have received two electric shocks, both were from 240V (domestic) both were in a residential environment. One was while having a shower due to faulty wiring and the other was from using a faulty electrical appliance. Both were also in the days before RCDs.

            My old safety spiel to newbies was –

            Your working with something that is invisible to the human eye, can travel straight hard solid materials, travels at close to the speed of light, can leap through the air and kill you instantly.

        2. ECLBs are used for medical or server environment where you don’t want you power to be cut off instantly.
          I believe some medical environments even use isolated power sources for their equipment.

          In these cases a RCD simlpy wouldn’t work.

          1. Medical environments use a whole raft of devices.

            In ops you will find isolators. In gas you will find ELCBs and RCDs. Medical grade RCDs trip at 15mA were the home variety trip at 25mA or 30mA. Non-patient public areas often have 25mA or 30mA RCDs.

            Medical grade equipment will have opto couples in it.

            They often use air gapped transformers. You can’t switch off critical equipment for a storm.

        3. Thanks – learned something new (naming schemes for phases).
          GFCI is an umbrella term? ok then RCDs/RCCDs and ELCBs are all GFCIs but the first ones detect differential currents at the “source” and ELCBs measure fault currents/voltages at the “drain”. Is that a better/proper distinction?

          This is not the first time I’ve read about RCDs tripping falsely because of inductive loads but I’ve never experienced it nor heard of it. Maybe all domestic devices with inductive loads around here have good enough PFCs?

          And I don’t get this false triggering of RCDs with inductive loads anyway. They don’t care about the voltage, only the currents through all phases (and N) matter and the sum of those should be zero no matter the load.

          Is the phaseshift (2/3*pi) in combination with inductive loads the culprit?

          1. You can’t economically correct the phase derived power factor of an inductive load. Old fluorescent tube light fittings had a capacitor in them (that is entirely unnecessary) in an attempt to correct the average phase difference for a household because the older spinning disk power meters were effected by power factor but don’t tell anyone that lol.

            Although a residual current device is a “current” sensor, they don’t actually work that way.

            They have the active and neutral would in opposite electron directions (the same physical direction) around a ferrite or ferric former. A secondary is would onto the same former for the trigger circuit.

            Now, firstly to over simplify and then explain why – The current in the secondary circuit is a function of the secondary voltage and the secondary load. The secondary load is also inductive.

            The voltage in the secondary winding is a function of the primary voltage and the turns ratio. So it is not actually a current sensor.

            However, the secondary load (the trip coil) needs a minimum Power to operate. Since the secondary voltage if fixed by the turns ratio. This would mean there is a minimum power required in the primary where the voltage is also fixed so there would be a minimum primary current requirement.

            If only it were that simple.

            The secondary power is also effected by the permeability of the core – think of like the internal resistance of a battery.

            The secondary voltage is also proportional to the primary voltage and not primary load current which are significantly out of phase with inductive loads like fridges.

            Added to that is that the trip circuit load is an inductor.

            The maximum power transference from primary to secondary will occur when the primary voltage and load current are in phase. However you want the device to trip at the same differential current when the voltage and current in the load are out of phase.

            The only real solution to this problem if you are designing the RCD on a budget is to increase the sensitivity of the trip circuit to out of phase conditions and hence nuisance tripping from inductive loads especially with the spikes caused when motors start – ie the inrush current. The motor can start at any point in the voltage phase and is completely unpredictable so there is no real way to compensate for this in a budget RCD.

      1. I once worked in a building full of communications equipment that was made by Siemens. Everything was written in German which wasn’t a big issue as circuit diagrams use an international language.

        All the equipment modules had acronyms that meant nothing in English but we all knew what they did so it didn’t matter what they were called.

        Some times a bunch of work mates would go to a public place for lunch and be talking about their days work using all these acronyms. Occasionally someone would a question about the acronyms like – “What’s a SNRZ ?” only to be even more dumbfounded when no-one could answer their question because we had no English language explanation. We would all just sit there with blank expression on our faces while we try to find an answer.

  3. “the reference level for a digital circuit (not necessarily at zero volts, either),”
    A reference level is the level from which all voltages are measured, so by definition, the reference level IS zero volts.

    1. Not sure if you are purposefully being obtuse or genuinely confused, but the point that the reference might not be the same as the ‘0v’ that you care about, ie ground or the 0v terminal of the supply rails. Could easily be a fixed offset from your scopes ground or even a moving reference in some systems.

        1. Correct, the potential of any ground including the earth itself is not at the same potential everywhere. When I was working in the communications field we once had a problem with a noisy communications circuit between a satellite communications terminal and another center a mile away. In order to try and isolate the problem we pulled the cable grounds at one of the sites and found 325V between the cable ground and the earth. Both sites had extensive underground grounding grids. The issue was that the island was a volcanic rock island (a poor conductor) and one of the sites was much closer to the ocean than the other site (nearer the ocean, better ground). The permanent fix was to move to optical fiber to eliminate the ground loop noise. It would have required a huge conductor to pull both sites to the same ground potential and that itself would have presented problem in that the electrical ground and communications ground could not be at the same potential. This makes things both dangerous and noisy electrically.

  4. This was very informative and straightforward. I appreciated it. And I would have made more of an effort to attend Supercon if I knew there was going to be so much practical information.

  5. The whole point of local ground (not protective earth) is that net is defined to be zero volts. Remember voltage is always a relative measurement between two points – you can pick any node you want and define it to be ground, which means zero volts.

    1. Uh, no. Ground is ground. A piece of equipment’s ground point can be connected to ground (earth), and/or to other equipment’s grounds, without adverse effects (excepting loops). Ground is very often the chosen reference point, of course. Not necessarily the same thing as protective earth.

      You can pick any node you want and define it to be the REFERENCE point and arbitrarily call it 0 volts. That doesn’t necessarily make it “ground”.

      1. I once organized a LAN-party in larger domestic building and supplied some tables with power from distant parts of the building (a few dozen meters perhaps in at least two directions).
        While connecting an ethernet switch of one of those tables to the main switch I got zapped.
        The building was not properly grounded or not all “grounds” were properly cross connected and the shielding of the ethernet switches were not at the same “ground” potential.

        I switched to unshielded cables and it worked like a charm – no more electrocutions.

          1. It depends. If the devices are both tied to earth or another ground grid, you do not want a second ground path through the comm cable. Different resistances to earth will cause current flow through the interconnected ground (this is called a ground loop). If only the head end equipment is earthed then you want the shield to pull the far end equipment to the same ground (picture a PoE device with no earth connection at the device itself). The important thing is one and only one point where the ground network attaches to earth or you need an extensive grounding grid to pull multiple points to earth with very low resistance (telephone central offices often have huge grids under the entire site so that you can make connections to earth ground all over the place and be assured they are all at the same potential). In general shielded Ethernet causes more problems than it solves and should be avoided in most cases.

      2. No, ground isn’t ground. You often CANNOT connect a ground on a piece of equipment to earth without causing a problem. This is especially true with older tube electronics where chassis could be hundred of volts above earth. This is why there are isolation transformers. If you use an oscilloscope without isolation you will have a problem because the o-scope probe ground reference is often tied directly to chassis ground of the scope. When you connect that probe ground to the circuit under test, you could directly short hundreds of volts from the circuit under test to the earth terminal of the scope. You also see this configuration often in precision instruments where they are trying to avoid noise (there is usually some imbalance that causes some small current flow to/from earthed connections). It that configuration you will often see opto isolators separating the measuring circuit from the support and display circuits.

        1. Even the National Electrical Code has gone back and forth on the correct way to do things. For example, there are two ways to do a subpanel in a garage that the NEC has gone back and forth on. You could either break the ground/neutral connection at the subpanel and run the ground and neutral back to the main panel to ensure only one connection between ground and neutral (problem is that the neutral and ground might not be at the same potential in the subpanel. The other method is to independently ground the remote building and leave the neutral/ground jumper intact on both sides (problem is that there is now a ground loop between the remote building and the main). Both methods require a ground rod at each side. Fixing one problem creates another.

          Old school they would run ground and neutral back to the main panel and not ground the remote building (dangerous because there could be an unintended high resistance ground and there is not a direct ground for lightning protection).

  6. As someone who works on protective relays for transmission lines, I felt her presentation was lacking. Her presented understanding of ground reference seemed to end at the third prong and be summarized with that’s all an electrician would be concerned with. I’m sure she knows some things, but I felt her presentation was more concerned with instagram filtered photos and stories than actual substance.

    1. I very much agree, but on a broader scope. Having done similar audio-engineering, digital-design, product compliance, static mitigation, and industrial-equipment work involving “ground problems”, I can say that not only did the talk essentially fail to define how those are different from a GROUNDING perspective, but it also failed to give any decent summary of those “methods” that were often mentioned. There was no explanation of about any of the technical issues or solutions that anyone who has read Ott’s original or updated works would be very well versed in. “Reading More” wasn’t even part of the talk, as if all an engineer needs to know was somehow in that talk. In summary, there is essentially ZERO technical or even illustrative understanding of WHY or HOW multiple interconnected grounds in a system cause issues.

      It also appears that the presenter has no (good) training in devising, planning, writing, scripting, or giving presentations. Typically, good technical talks don’t include the audience being invited to call for the presenter’s dancing on stage, then-and-there.

      If I were grading this, I’d give it a B-, as lacking factual, work-situation relatable and actionable information, and being poorly organized and awkward in presentation. If this was their first, hopefully the next one will show similar indefinably exponential increases in quality.

  7. Elliot Williams says:
    A breaker limits total current. A GFCI makes sure that the current going in/out on hot equals the current coming back in/out on neutral.
    Or… a GFCI makes sure that no current leaves the circuit’s loop (i.e. through you to ground).
    Neither breakers or fuses “limit” current, they only interrupt the current. Peak fault current from a household mains receptacle conformant to NFPA70 can be on the order of 10E2 to 10E3 amps. If it is important for you to limit current, than you must provide some type of series impedance.

    1. Thank you! I am 37. 11 hardware products to market shipping in high volume to over 32 different countries, and was the chief tech at a commercial recording facility with 14 production rooms, four large format analog mixed signal consoles, and loads and loads of gear. I have some experience, about 15 years. Hopefully enough to present there is an important difference between “grounds” and trust the open and inclusive HaD community to backfill with collective experience and knowledge.

  8. Ground is something I deal with on a daily basis. We work on high current (15-20amps) machines and when we run into a bad circuit, it can wreck havoc with the digital circuits in these machines. When you run into common mode problems, it can be frustrating. Then, you have the customer have the outlet checked, and their “maintenance” person says the outlet is ok because they never checked it UNDER A LOAD condition. Which, is why I carry a suretest test device. Checks the voltage, neutral to ground, voltage loss at 10 or 20 amps.

  9. Grounding is a fundamental problem in audio systems… it’s unfortunate that the video and the comments don’t go that deep into them.

    Neil Muncy was the AES guru of ground and shielding problems in audio.

    A sample of the problem space:

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.