2 grounds, or not 2 grounds.?

[deadastronaut]

just put this here to stop the hijacking of the pictures thread!..ok.

so when breadboarding, do you guys use one blue - rail for the - 0v power? only..?
and the other blue - rail for signal ground only then?....

and always keep em seperate?..

[Manny]

I tend to keep them separate if I can. Although it's more force of habit than for any particular reason.
There don't seem to be that many projects that are affected by joining them. (your tiny trem is an obvious exception!)

[deadastronaut]

I tend to keep them separate if I can. Although it's more force of habit than for any particular reason.
There don't seem to be that many projects that are affected by joining them. (your tiny trem is an obvious exception!)

cheers manny thats what i thought with the trem,  but there is some discussion on the pics thread that i have moved here with some very
differing opinions on this subject.

[jefe]

I use the old "star grounding" method. All grounds go directly to the input jack. When breadboarding, I do pretty much the same thing. All grounds lead to the same place, don't they?

[deadastronaut]

well this is what i thought..but others have different ideas..

just trying to pin this myth/fact/mystery down once and for all..

[R.G.]

well this is what i thought..but others have different ideas..
just trying to pin this myth/fact/mystery down once and for all..

Pinning it down once and for all will probably prove impossible. This is because in general there is more than one way to interconnect all points which should be at 0V (notice I did not say "ground") in any circuit or system to get acceptable results. And humans being what they are, they will decide that whatever worked for them once is The Way, at least until it doesn't work next time.

I've typed this into this forum several times. I guess another time won't hurt.

Historically, it's called "ground" if your English derives from USA technical tradition or "earth" if you got your English from the UK tradition, because the physical planet itself was taken as the ultimate source of voltage reference. A metal rod driven into the soil and connected to a wire was  "ground" or "earth" connection. The planet itself could absorb any human-generated electrical current with no change in its characteristics. Later people realized that for most purposes only the voltage difference between any two points mattered electrically, and that any point in the circuit could logically be called "ground" for measurement purposes as long as you were consistent about it, at least for measurement and analysis purposes.

As more about electricity was discovered, and circuits using finer and smaller signals were used, people realized that their first approximations were not strictly true. In particular, a wire, however large, was not truly zero ohms. Every wire was a resistor, and generated a voltage equal to the current through it times its resistance. True, the resistance might be milli-ohms or even micro-ohms, but when you're using signals in the millivolt range, the resistance of the wires causes a perceptible error.

And that is the central issue with grounding. All wires are resistors - and inductors, if you're playing with fast edges or RF signals. So they generate a voltage in response to current through them. If you have a circuit where it is important to reference signals to a true zero volts reference, you cannot do that if the wire that carries that zero volt reference also carries current. The only way to get the voltage across a resistance (in this case, wire) is to have the current through it be zero.

What makes this confusing is that while we have to have reference grounds for signals, electricity only travels in full circuits (that's where the name comes from), so to run a circuit, you have to power it from some source of electricity, and the "used" electricity must return to that power source. In single-supply circuits, one of the power wires is also connected to the zero-volt reference, so it's "grounded". This is a problem because people then see the side of the power supply which is connected to zero volts as "ground". The wires back to the power supply carry the overlapped "used current" of all the circuit's stages. If you use any point on this wire as a reference for your delicate inputs, the inputs will then amplify whatever noise is on that wire because of the current generating a voltage in the resistance/inductance of the ground wire. That's why I refer to the wires returning current to the power supply as "sewer ground". They literally carry the current-sewage for the circuit. In general, it's a bad idea to put sewage into your inputs.

Star grounding gets around the issue of ground contamination by making every ground be on an isolated wire. It's impossible to contaminate a wire with ground currents if the ground currents flow on other wires. Star grounding is the only way to positively assure ahead of time that no grounding issues arise. It may not be the only possible way to get good grounding, just the only way that's a sure solution in general before you know the mechancial layout and wiring.

And that's another issue - grounding cannot be easily included in schematics. Schematics in fact make it worse by taking the convention that all points labeled "ground" are the same. They are not, and cannot be in the real world unless star grounding is used when the circuit is actually wired up. In reality, all those points on a schematic with a ground symbol have resistors between them.

In grounding, you have to know what current flows in which wires to do it well.

In effects, we get away with grounding that would be disastrous in bigger systems because the circuits and boxes are so small. This forces the wire lengths to be small, and generally reduces the resistances that ground currents flow through. Only in really bad cases of mal-connection of grounding, or very high gain pedals mis-grounded does it rise up to smite us. When that happens, people who haven't had to think about grounds before get mystified, and embark on their own journeys of learning.

You might like to read: http://www.edn.com/archives/1994/112394/24df1.htm

[deadastronaut]

@RG.
thankyou very much.
now that explains it very well indeed...thanks for that and the link,
i'm sure it will be read a million times by many diyers too...great info... rob.

star ground on pedals then...and bread boards....great!.

[Manny]

Thanks R.G.
Very insightful indeed 

[MoltenVoltage]

From what RG said, it sounds like you want to connect all the grounds at the input jack to the extent possible.

I assumed it was preferable to connect them at the ground terminal of the voltage regulator rather than at the input jack so you don't introduce noise on the power input side of the regulator.  Am I way off?

---

In any event, it seems like tracing the shortest distance from the signal-carrying component to 0v will let you know where it might pick up noise.

Do any of the advanced PCB design programs take these factors into consideration?  Can you designate an "audio signal path" and have the program alert you when there are potential problems?  I've been using the limited ExpressPCB software.

[MikeH]

From what RG said, it sounds like you want to connect all the grounds at the input jack to the extent possible.

I assumed it was preferable to connect them at the ground terminal of the voltage regulator rather than at the input jack so you don't introduce noise on the power input side of the regulator.  Am I way off?

In pedals it doesn't really matter as long as they're all the same point.  If you're using the typical insulated DC jacks, then there won't be any resulting ground loops. 

[earthtonesaudio]

I breadboard using horrible practices, so that potential problems show up more easily.

[R.G.]

From what RG said, it sounds like you want to connect all the grounds at the input jack to the extent possible.

Depends. What are the currents flowing in every one of those wires?

Star grounding is relatively robust, in that if you do a pure star ground, there is only one current from one section of the circuit flowing in each wire, so it kinda doesn't matter. However, almost no one has the tenacity to put two dozen ground wires in a single effect, and maybe fifty in an amp. It gets out of hand. So everyone uses a modified star. Making the chassis connect to the signal ground only at the input jack is a good idea, for RF reasons. Or making the input ground isolated from chassis, with its own wire, but a ceramic cap from input jack ground to chassis right there, again for RF. Grounding practice is actually a continuum, with star grounding being what you want at DC (which actually includes the entire audio spectrum in this viewpoint) and slowly changing over to ground planes at RF, where the inductance of the conductor means more than the resistance. A capacitor with leads over a few millimeters long is an INDUCTOR at any significant RF frequency and never looks like a capacitor because of the self inductance of the leads.

Pedals are actually a kind of no-fault sandbox for grounding, except when the practice gets pushed to high input impedances and super high gains - which is where the gain-monster distortion pedals are these days. But for low-gain analog-only pedals, you can often get  away with almost anything because the frequencies and gains are so low.

I assumed it was preferable to connect them at the ground terminal of the voltage regulator rather than at the input jack so you don't introduce noise on the power input side of the regulator.  Am I way off?

Not way off. I actually prefer that myself because it forces no DC offsets. But starring at the input jack works OK.

There is an interesting practice that guarantees the maximum problems with power supply current, though. Imagine if you set it up so 100% of the "sewer ground" currents ran through the input jack wire. Oh - wait a minute! Using the input jack power switching trick does that! All of the power current goes through that wire. It's almost like ground problems are built in! 

I laugh every time I think of this one. There are ways around it that make for pedals with much more oscillation immunity, like simply making the output jack be the one that switches the power side of the sewer. Another is my favorite, using a PNP to switch power into the pedal and pulling its base down through 10K or so with a wire to the "ring" contact on a stereo input jack. I only know of one commercial pedal line that worries about things like this. 

Do any of the advanced PCB design programs take these factors into consideration?  Can you designate an "audio signal path" and have the program alert you when there are potential problems?  I've been using the limited ExpressPCB software.

Most of the higher end tools let you designate special warnings on a per-net basis. But they also caution that autorouting is not for analog signals and when there is a critical analog net, you call a human layout wizard. So far I don't know of an program that will help much with this one.

You need to (1) know what signal's on the trace and (2) what signal is on the traces it passes (3) the impedances of the signals involved and (4) the sensitivity of the circuit for those traces to crosstalk, then take that into account in routing the traces, and adjusting the spacing from other traces and any shielding that may be possible from planes and interspersed ground/power/low impedance lines that may be available. Unfortunately, that amounts to encapsulating all the design knowledge of the designer for every possible circuit into the layout program, so mostly they punt.  Sometimes, you just gotta call a human.

[Mugshot]

hey RG, what about this idea:



pcb ground is connected to PSU ground; in/out, signal cables and box are grounded together. only the first box gets to ground both its signal and pcb grounds. im assuming a lift switch could be used to implement this.

merit or demerit?

[jcgss77]

Unless I am mistaken, and if I am painting a correct picture from R.G.'s sewer anology, isn't your design basically feeding all the sewer from the amp into the stompboxes in that layout?  It is good to see people thinking.  We would never improve if no one ever thought outside the box.

Thanks R.G. for your endless knowledge in probably the most sensitive issue in our world.

[jkokura]

Not on topic...


Did anyone notice how awesome that guitar looks in paint! I am fully able to recognize a les paul, but it's such a rudimentary drawing. Well played!

Jacob

[Mugshot]

hahaha, that guitar's how les pauls should be built!  i just dished out a quick illustration for the guitar in MS Paint 

to ground or not to ground, that is the question 

[deadastronaut]

its more like an SG PAUL hybrid,

either way i wouldnt want to play it much!..