PCB Ground Loops and How to Prevent Them

[@petersage5157]

Making another post a few months later cus I wanna. Another option in cases where ground loops might otherwise be unavoidable and the receiving equipment has an unbalanced input is to use a ground compensated output. It's not particularly well known, but it's reasonably simple, requiring only a spare op amp in a differential driver topology at the sending unit, but you need a balanced cable to pull it off. Noise picked up by the cable is referred back to the sending unit through the cold line and compensated out, hence the name.

[@axelBr1]

The drawing of the two items of equipment with each power point having a separate earth location, is probably incorrect as the earth connector of the socket will be wired back to a common earth bus bar at the distribution box. In the UK we also have ring mains rather that the hub and spoke arrangement in the US and other countries, so the earth connection for us between neighbouring sockets should be very low. Not touched on, was that the ground line of a power supply is "connected" via the transformer to the neutral wire and not the earth. The neutral is generally grounded somewhere in the power system, but the use of a transformer means the voltages in the equipment are isolated from the mains voltage.

[@petersage5157]

"Ground lift" switches are fairly common on pro audio equipment for this reason. Another _possible_ option for breaking a ground loop is to use a cheater plug at one end. *Do not do this* if there is any part of the equipment connected to the chassis for system ground! There are many well-documented cases of singers grabbing a microphone and getting a spicy mains-referenced surprise from this. I'd love to see a video on RF grounds vs system ground. There are a few (mostly poorly corroborated) anecdotes of guitarists picking up local ATC chatter on their unbalanced shielded cable, and the commonly recommended solution is 10nF from the *isolated* input jack to shunt RF interference to the grounded chassis.

[@sc0or]

When a user disconnects a signal cable, the device with such cheater plug becomes not safe to use. As you might guess, to use a cheater signal cable could be a better solution. However, both devices must properly connect a signal ground to earth/case. Very often Chinese boxes have just a soldered signal ground to a shield

[@petersage5157]

@@sc0or Yes, that's why I used italics and bold for emphasis. Can be done, but very much not recommended because because high risk of electrocution. The microphone shock scene in the movie _Almost Famous_ is a composite of several true stories. Doug Ford mentioned such incidents in one of the videos he did with Dave on EEVblog. In your typical bar-with-a-stage performance venue, the stage equipment and the mixing console will be on separate breaker circuits, and the hundreds of milliohms resistance in the ground leads will be enough to induce hum. I presume the ground is typically lifted at the mixer end so the shield will essentially be a driven guard (here we are again) at the instrument's common potential all the way to the console input jack.

[@scotshabalam2432]

Please call our US plugs by their technical name: "Surprised Amazed Man".

[@magnuspihl6974]

Motherboards for PC's will have each plated mounting hole to the chassis. I assume they are all connected to ground plane. Isn't this risk for ground loop problems?

[@gendronw]

Chassis is like an infinite electron well. If you are not passing large amount of current in it, it should it be at the same electrical potential everywhere. By "tapping" the PCB plated mounting hole to the chassis, you are kind of trying to have also the "ground" or common net on your PCB to be at the same potential everywhere as your chassis. Everything is the same net, same potential, so no currrent loop.

[@Zachariah-Peterson]

It's only a ground loop if you develop a potential offset along the conductor path. No potential offset = no ground loop. The term "ground loop" refers to the current driven through different conductors that make up ground, not a physical loop of conductor.

[@derekkonigsberg2047]

This is why I'm still confused on the case of mounting holes, system ground, and a metal chassis connected to protective earth. Many PC motherboards connect all the chassis mounting holes to system ground. This video says to NOT connect chassis mounting holes to system ground. Yet other advice says to perhaps connect one hole, and connect the rest through capacitors. Still can't seem to get a clear answer as to what the correct approach actually is.

[@Zachariah-Peterson]

@@derekkonigsberg2047 Maybe I can clarify. I'm not saying that you should always only have 1 hole connected to chassis and the rest un-connected. I'm saying two things. First, if you happened to have a ground loop and you knew it was via a connection between two mounting holes, disconnecting a mounting hole is one way to break it. There are other reasons not to connect system GND to chassis via all mounting holes which relate to a user's possible exposure to return currents, but that is an unrelated issue. Second, I am saying that even if you did connect all the mounting holes to system GND and chassis, that does not mean you automatically get a ground loop. Remember, the term "ground loop" refers to the current driven through a ground net by a potential offset, but a potential offset in your ground doesn't magically appear just because you created a loop-shaped region of conductor. Either the conductor is imperfect (such as Earth), or the loop of conductor received stray magnetic fields, or possibly the loop of conductor was formed via parasitic capacitance.

[@rb8049]

The chassis should connect to the ground shield running 360degrees around each connector. This is where the connections should be made. Just that. This forms a faraday cage.

[@Andrew-dp5kf]

I always isolate any comms like RS485 so there’s no issue with grounding

[@AlbertRei3424]

how do you isolate them?

[@magnuspihl6974]

How? You mean with isolated RS485 transceiver or just providing only A and B at the connector?

[@Zachariah-Peterson]

@@AlbertRei3424 I would advise reading the link from Howard Johnson in the video description.

[@ehsanbahrani8936]

Thanks a lot ❤

[@rflick6695]

Aren't you only supposed to ground one end of a shielded wire?

[@frankstollar8492]

yeah how to correctly connect shielding on USB port which can be host or device would be really helpful. I found so many different suggestions and don't know which one is the correct to follow.

[@OMNI_INFINITY]

*Thanks for posting that. Seems weird that connecting all the grounds together at **03:24** isn't the same as connecting grounds together in a circuit. Why is that? How is it acting like a battery instead of a wire?*

[@varunjain496]

It’s because the conductor (here earth) between the two earth connections is not ideal. It has some resistance and if the current through that path is high enough it will cause a significant voltage drop or a potential difference between those two points.

[@Zachariah-Peterson]

The conductor at 3:24 is essentially the earth connection and it could be high resistance, this is due to the length of the wiring. Ideally the potential difference across this connection should be 0 V, but because the resistance is non-zero there will be a potential drop across that connection. The loop can also easily receive stray magnetic fields due to its physically large size. In a PCB, the copper connections are very short, and for a small width trace the resistance could be much lower. The trace loop area is also much smaller.

[@OMNI_INFINITY]

@@Zachariah-Peterson Thanks for explaining👍🏻

[@ehsanbahrani8936]

Thanks a lot

[@AlbertRei3424]

So these currents flowing in ground loops are essentially common mode currents?

[@Zachariah-Peterson]

They can be observed as common mode currents yes.

[@ЕленаПицина-ю8э]

In which video did you tell us about connecting ground plane to chassis ground? Thank you

[@Zachariah-Peterson]

It is the linked video in the description, that video discusses multiple aspects of grounding, including some uses of chassis ground.

[@nashaut7635]

Thank You!

[@big_whopper]

Or use balanced signaling, like all pro audio!

[@Zachariah-Peterson]

A lot easier in pro audio than it is in, say, industrial or automotive where there are some digital signals that have to be routed over cables. When the interfaces are standardized you can't just change it to balanced (differential).

[@nikolaydmitriev5939]

Thank U!