Avoiding ground loops

[zpyder]

When putting a circuit in a metal box and using the box itself for ground, does grounding all of the jacks where they are screwed in (ie, all of the connections are not made at "star" ground, but instead they are all made at different spots in the box) create ground loops?

Secondly, does using the metal chassis of a metal box for ground effect its shielding properties at all?

thanks & cheers,
zpyder

[Paul Marossy]

When putting a circuit in a metal box and using the box itself for ground, does grounding all of the jacks where they are screwed in (ie, all of the connections are not made at "star" ground, but instead they are all made at different spots in the box) create ground loops?

I have never had this happen (a ground loop) in a stompbox. In a tube amp, things change.

Secondly, does using the metal chassis of a metal box for ground effect its shielding properties at all?

No. In my experience, if you don't connect the shielding to ground, it's as good as not having any in the first place.

[rockhorst]

No. In my experience, if you don't connect the shielding to ground, it's as good as not having any in the first place.

As far as I know, shielding and grounding are separate issues. You stompbox is a Faraday cage, no EM-waves can penetrate a conductor since the electrons will screen the electric field the EM waves would cause.

Do you have to ground to the box?

[MetalGod]

I don't think any of the currents are big enough with stompboxes for it to be a problem (could be wrong).

[Paul Marossy]

As far as I know, shielding and grounding are separate issues.

If that's the case, then why do you ground one end of a shielded cable? I thought the idea was to shunt RFI/EMI to ground...

[R.G.]

When putting a circuit in a metal box and using the box itself for ground, does grounding all of the jacks where they are screwed in (ie, all of the connections are not made at "star" ground, but instead they are all made at different spots in the box) create ground loops?

Yes, but.

You often get away with this because the voltages and currents are so small. Sometimes you do NOT get away with it, and you get RF pickup, hum and/or oscillation.

Secondly, does using the metal chassis of a metal box for ground effect its shielding properties at all?

No. The metal chassis must be grounded to do a good job of hum shielding.

As far as I know, shielding and grounding are separate issues.

In the low frequency and audio world, this can be true.

You stompbox is a Faraday cage, no EM-waves can penetrate a conductor since the electrons will screen the electric field the EM waves would cause.

It is, but not grounding it can also lead to pickup of low frequency stuff. Ground it.

Do you have to ground to the box?

Yes, unless you like to hear hum and RF pickup.

I don't think any of the currents are big enough with stompboxes for it to be a problem (could be wrong).

This is generally correct, except for the ultra high gain distortion pedals. Unfortunately, that's what everyone wants.

I write up another view on grounding every week or two. It would be good to read any past articles with me as an author and "ground" in it.

Grounding:
Ground used to be literally the dirt under our feet. It has come to mean any place we designate as a place with 0V as a reference. In power equipment like amps, the signal ground is connected to the power line safety ground and ultimately to the entire planet through a ground rod in the electrical system. There are safety ramifications of that. But in a battery powered unit, "ground" is wherever we say it is, and everything else is referenced to that.

Most importantly, your signal is referenced to that. And whatever your signal is is what gets amplified and/or processed by the effect. All this works fine as long as all the places in your circuit that you say is "ground" is really at the same exact voltage.

But there are no perfect conductors at room temperature (yet). Metal chasses and copper wires have resistance, and so any two points with current flowing between them simply cannot be at the same voltage.

"Ground" is used three ways: as a reference voltage for signal and measurement, as a shield against interference, and as a sewer to convey all the used charge carriers back to the power source after the circuit is done with them.  As you might imagine, contaminating your pristine signal ground with voltages caused by currents in the sewer or currents caused by EMF shielding can be a problem.

The best solution is to separate pristine signal and measurement grounds from shielding and sewer grounds. If two points are electrically connected by only one wire, there cannot be current flow between them, and so the voltage drop across that wire must be zero. This is the point of star grounding. If your signal ground touches the shields and sewers at only one point, it cannot be contaminated by rising and falling with the currents on them. So your signal path stays clean of any ground interference.

If you connect signal ground to sewer ground at a distance from the sewer's power supply, the sewer-induced voltages get picked up just like signal and amplified by the circuit which was expecting signal. Likewise with shields. The way Faraday cages work is that the conductive material develops eddy currents that cancel the externally applied EMF internal to the cage. Across the shield, there may be induced currents. This effect gets more intense as the conductivity of the shield gets worse. Again, copper and steel are resistors.  So there are induced shield currents and voltages at different places across the shield in response the impinging field.

Can you hear sewer-induced and shield-induced ground voltages? In some cases yes, in some cases no. The more your effect circuit has a low gain (i.e. unity for some phasers and delays) then the less likely that the sewer and shield voltages can be heard. The higher the gain of your circuit and especially the higher its input impedance (think MOS and JFET input megadeath plasma torch distortions) the more chance you have for these ground signals to be heard.

So you need to ask yourself - do you feel lucky today?

If so, just hook up the grounds any old way. Generally you're gonna be OK.

If not, connect signal ground to the chassis at one and only one place, counting where the jacks touch it. And take your power connect to the power source separately from your input signal ground.

If that's the case, then why do you ground one end of a shielded cable? I thought the idea was to shunt RFI/EMI to ground...

You connect a shield in one place so that it does the job of eating any impinging EMF fields by dissipating them as heat in eddy current loops, not by shoveling them into two different ground places that it can then pry into different voltages because both the shield and the ground wires connecting the two shield connections have resistance to the flow of current, and hence a voltage difference in response to the impinging EMF.

[Paul Marossy]

You connect a shield in one place so that it does the job of eating any impinging EMF fields by dissipating them as heat in eddy current loops, not by shoveling them into two different ground places that it can then pry into different voltages because both the shield and the ground wires connecting the two shield connections have resistance to the flow of current, and hence a voltage difference in response to the impinging EMF.

I know the idea is to not connect both ends of the shielding to ground - I was trying to make the point that on the basis of believing that the farady cage did not need to be grounded, then neither would one end of a shielded cable need to be connected to ground, it could just just wrap the signal carrying wire and magically wipe out EMI. Most of us know that isn't true.

What I did not know is that it ate "any impinging EMF fields by dissipating them as heat in eddy current loops". Very interesting! 

[R.G.]

Yeah, I was fascinated when Dr. Ramakumar pointed out to us that what happens when EMF waves hit a conductor is that they try to propagate right through. In doing so, they set up eddy currents in the conductor that make an opposing EMF field that cancels the impinging EMF inside the shell, to the degree that the EMF energy does not get eaten up by the I-squared-R heating of the conductor.

The degree to which conductive shields are effective depends on their conductivity. A superconductor will generate an equal and opposite field that perfectly cancells the impinging EMF field, in much the same way that it will generate an equal-and-opposite M-field and suspend a permanent magnet in mid-liquid-helium. Normal conductors are not perfect, and the effectiveness depends on the frequency (skin depth increases resistance) and material (raw conductivity).

[Paul Marossy]

Fascinating stuff! 

[zpyder]

R.G.-

As always your post was insightful and educating...

What I'm hearing is star grounding is the way to go to be sure, but in most situations I wouldn'even be able to tell the difference

If not, connect signal ground to the chassis at one and only one place, counting where the jacks touch it. And take your power connect to the power source separately from your input signal ground.

I don't quite understand the "take your power connect to the power source sperately..." part.  Are you saying to ground the sewer to a dfferent location than the star ground??

thanks
zpyder

[Toney]

What I believe RG is saying here is: separate your signal ground Ie ground on the circuit board and the negative power source.

To do this you cannot use the normal stereo jack trick on the input to fire up the circuit by connecting the negative to the ground via the jack. You will have to switch the power direct from your 3pdt. Then return it before your filtering on the board. Or use an on/off switch.
To have only one grounding point on the chassis, you will need at least one plastic type jack and connect the grounds with a wire from ring to ring.

Think thats right...

[R.G.]

You're correct, Toney.

The stereo jack trick is a poor choice for high gain distortion effects. However, because effects are small and uncomplicated (by many electronics standards at least) we get away with murder metaphorically in ground and power wiring, so we sometimes don't notice the problems.

The ideal grounding for an effect is to have the power lead which connects to the power source - battery or DC jack - go to the PCB. From the PCB entry point, one wire goes to the input jack ground. This may also connect to the shielded metal enclosure. Ideally, the output jack will be isolated from the shell, and have its own ground wire from the PCB.

That makes jack power switching a problem. I have some solutions, but I think the right thing to do is to write a definitive article on grounding and put it up on GEO.

As I mentioned, there is the ideal way to do things, and then what we can get away with most of the time. Almost every effect today just gets away with it.

[Primus]

I can promise you it would be appreciated. By the way, my Spyder is coming along great. I hope to finish it this weekend if parts arrive.

J

[David]

Posit that you have 2 or more circuit boards in one box - like a "Jimi in a Box".  They both (all) require the same power supply.  In this scenario, are ground return points "star-grounded" on each individual board, or are the individual boards star-grounded to a single point in the enclosure?  Suppose further that you're going to do the little 100R resistor and 1000uF cap filter trick.  Can you have one of those on a single daughterboard and distribute power to all circuit boards in the enclosure from there?

[zpyder]

100R & 1000uF ??  A hi/lo pass filter at 1 Hz?  Why?

[David]

100R & 1000uF ??  A hi/lo pass filter at 1 Hz?  Why?

AC hum, Grasshopper...

[scaesic]

You will have to switch the power direct from your 3pdt.

i don't think you can't do that, it takes a while for chips and the circuit to settle into the steady state, by switching the power to the circuit with the 3pdt you'll hear the circuit powering up. Not to mention only true-bypass will work.

What I believe RG is saying here is: separate your signal ground Ie ground on the circuit board and the negative power source.

I know this needs to be done when you use charge pumps to avoid the pulses generated being audible, but what's the theory behind using it in any old circuit?

[Gilles C]

You're correct, Toney.

The stereo jack trick is a poor choice for high gain distortion effects. However, because effects are small and uncomplicated (by many electronics standards at least) we get away with murder metaphorically in ground and power wiring, so we sometimes don't notice the problems.

The ideal grounding for an effect is to have the power lead which connects to the power source - battery or DC jack - go to the PCB. From the PCB entry point, one wire goes to the input jack ground. This may also connect to the shielded metal enclosure. Ideally, the output jack will be isolated from the shell, and have its own ground wire from the PCB.

That makes jack power switching a problem. I have some solutions, but I think the right thing to do is to write a definitive article on grounding and put it up on GEO.

As I mentioned, there is the ideal way to do things, and then what we can get away with most of the time. Almost every effect today just gets away with it.

I agree with you about the power switching.

My solution to that problem has always been to use this kind of jack instead of the standard stereo jack.



The switching is completely isolated from the ground.

And I must say that you answered a question I had about the grounding of the Output jack in my next circuits. 

But an article about the subject would be welcome.

Gilles

[Toney]

Hi Gilles,

I remember you mentioning those before.
I think they may well solve the problem.

Do you have any more info? Manufacturer part no?

Your in Europe, right?

Has anyone seen these in the States or Australia??

[Gilles C]

I'm in Canada. I can get them locally, as well as the Neutrik ones.

I can't tell the Manufacturer right now, but maybe someone can. I know they are easy to get, at least in U.S. and in Canada.

I'll check if I can find more tomorrow if nobody else answered. I have to go to sleep...

Gilles