How to ground "properly"?

[R.G.]

With apologies to Ray Charles: well it's groundin' time again...

Grounding is simple in theory, but it requires thinking in terms of currents, not voltages, the opposite to how most people learn electronics. Here is the essence of all grounding practice.
1. Every conductor is really a resistor of some value, however small. For extra credit, it's also an inductor.
2. Ohm's Law rules: a current through a resistor generates a voltage. For extra credit, an AC current through an inductor generates an AC voltage.
3. For all single-ended amplifiers (that is, one signal, one ground) wobbling the ground is the same as wobbling the signal wire: it all looks like signal to the amplifier input and ground voltage movements get amplified just like signal.
4. In modern electronics, "ground" means "a point of reference that you've decided to say is zero volts for all measurements and signal amplification purposes". It is ONE POINT. Any wire carrying "ground" to some part of the circuit also generates a remote voltage at the other end.
5. You simply can't make a "ground" voltage remote from your ground reference point be the same voltage as your One True Ground unless there is zero current going through it. All you can do is to make the inevitable voltage difference caused by the "ground" wire's impedance not cause noticeable interference.

The most reasonable way to deal with ground noise is to make sure that the currents that flow through each ground wire do not generate interfering "ground" voltage at their remote ends is to ensure that the currents flowing through each ground wire do not also carry "foreign" currents. This is the essence of the plethora of multi-color illustrations of grounding schemes - pre-cooked recipes for isolating the current-caused remote "ground" voltages by forcing currents to flow only in certain paths.

I classify "ground" into several categories, depending on the currents they carry. I think of them this way:
1. Shield ground: this is a ground that intercepts radiated fields and dissipates them, or moves the currents to where the implicit electrical loop can be satisfied. The more a shield ground totally encloses a circuit, the more effective it is. Shields dissipate RF waves and AC magnetic fields by converting the RF energy to heat in current loops in the shield conductor. They prevent electrostatic field interference by shunting the field voltage to the One True Ground, where it can't get back out. Shields carry any random voltage that comes in, and therefore there should be one and only one wire connecting the shield and the One True Ground in any box. There are some asterisks and caveats to do with high frequency RF, but for pedals, this is how it works. There is some subtlety involved about grounding input jacks, output jacks and so on.
2. Power grounds: Electricity runs in loops (... circuits...) and so power out of the power supply must return to the power supply. If you have one wire from the power source to the whole mess of stages in a circuit, all of the return current is mixed on that one wire, creating what I call a "sewer ground", a wire that carries all of the "used electricity" back to the power supply.
3. Rectifier grounds: This is a special case. The pulses of current from rectifiers into/out of a first power supply filter cap generate a 2x power line frequency voltage in all the two wires leading to/from that cap. If the rectifier return wire goes anywhere else than the negative terminal of that first cap, you WILL get power line ripple/hum that you can't get rid of by anything other than moving that wire.
4. Signal grounds: these carry the signal return currents. If you let any other currents (like power return currents) onto these wires, you will have interfering signals introduced into your signal. A test of practicality applies to this. If the currents are small enough, the introduced noise will be small, maybe small enough to not be audible. Pedals often qualify for the "small enough" exemption.
This could go on. You see the point: don't mix current sources on signal ground wires, because if you do you will introduce the currents times the ground wire impedance into your signal.

This line of reasoning is also behind the whole daisy-chain versus isolated power supply discussion. If you mix power currents on signal grounds you will get voltages that add to your signals. These may be small enough not to be noticeable, or they may be large enough to be a problem. I just got through with an analysis of daisy chain versus isolated power supplies for pedals, and found that in fact, what I've always thought about this is true: for most analog pedals, daisy chains cause no audible problem. It's when you put high power especially switching-digital type pedals on the same power supply that the actual ground wire paths make a difference. It's always fun when the practice does support the theory.

[POTL]

Greetings.
Considering more and more different diy projets, on different resources and schemes of different manufacturers I notice more and more interesting details.
Take the classic BOSS DM-2 as an example, the power section is divided into 2 parts, 9 volts feed all active elements in the sound path, a 22 Ohm resistor goes from 9 volts, it supplies voltage only to the delay section, roughly speaking, dividing the power into 2 independent parts.
This is not uncommon in BBD schemes.
Today I looked at the JHS Panther Cub board, and noticed that it is divided into 2 parts, they are connected by a 100 Ohm R15 resistor (located in the lower right corner of the board, next to the SOIC-16 chip), both of its contacts are connected to large polygons, judging by to all this is the division of land, by analogy with the BOSS described earlier.
Above, we can see the place for the tantalum SMD capacitor C3, it is not installed on the board, but had to be connected by analogy with a resistor, both contacts are connected to polygons, the case is located on the border.
Tell us about this stage of the separation of grounding, as I understand it is something in the spirit of analog and digital ground.
Does it make sense to divide the grounding into 3 parts
1) Analog
2) Digital
3) Audio jacks and power section.
Will such grounding be practical if you build analog pedals with digital control (Electric Druid Tap tempo as an example) or modulation with LFO?

[ElectricDruid]

Does it make sense to divide the grounding into 3 parts
1) Analog
2) Digital
3) Audio jacks and power section.

Please explain why the Audio jacks and power should be separate from the "analog" section? I'm not seeing what's to be gained from three parts here rather than two.

[POTL]

Does it make sense to divide the grounding into 3 parts
1) Analog
2) Digital
3) Audio jacks and power section.

Please explain why the Audio jacks and power should be separate from the "analog" section? I'm not seeing what's to be gained from three parts here rather than two.

sorry for the long answer, I thought this topic was ignored.
Yes, now I thought and realized that there is no reason to disconnect the power section from the audio signal.

What do you think about the LFOs and control parts in the Envelope and Octavers.
I often see how the power is separated using a resistor and an additional capacitor (as if separating the network and at the same time creating additional filtering), does it make sense to put a 10-100 ohm resistor between the audio ground and the ground of the control part of the circuit?

[ElectricDruid]

does it make sense to put a 10-100 ohm resistor between the audio ground and the ground of the control part of the circuit?

In my view, no, absolutely not. You want Ground to be all the same, everywhere. That's the ideal. Voltage drops between one bit of "ground" and another bit of "ground" are where the problems start. The point of star grounding is to avoid cumulative currents flowing through long daisy-chained wires and making voltage drops due to the resistance of the ground path. The bigger the currents, the bigger the volt drop. Star ground minimises those drops and gives each area its own direct path back to the central ground. So *adding* a deliberate resistor in the ground path seems like a very bad idea to me.

Remember that I'm totally uneducated in electronic engineering though, so if people with more background want to tell me I'm completely wrong, Im happy to learn something new.

[R.G.]

The problem with ground as a concept is that we no longer use a rod driven into the planet itself as a "ground" reference. We use instead the idea that "ground" is equivalent to "any place we want to use as a point defined as zero volts".

This makes grounding problems mysterious; so does the idea subtly propagated by schematics that anywhere you put a ground symbol is really 0.000000... volts.

It can't be. Wires, no matter how big, are low value resistors (with apologies to the folks working on superconductors; we can't get superconducting wire these days), so the only way to have truly zero volts across any conductor we can actually use is to have the current through it be zero, as well.

That's the idea behind having ONE wire to your external chassis. Monodes ( points with exactly one wire connecting them) cannot have any current flowing to and from them. All induced currents have to be resolved within the monode itself, so it simply can not cause any voltage on its connecting wire. This is why we ground external metal shells. Any currents caused by impinging RF, or magnetic fields is dissipated as heat by causing eddy currents within the shell itself, and can't change the voltage anywhere else. Nor the current, although worrying about current flow seems to be a lost art.

So it simply cannot be that "ground" is the same voltage everywhere if "ground" carries currents other than zero. Wires are really resistors, so any current going through them makes a voltage, the ultimate triumph of Georg Ohm carried down to today's world.

That means that anywhere you think is "ground" is really not ground, unless it carries no current at all back to the One True Ground you have defined as your local measurement point. Or, more importantly, no current at all back to the place that your INPUT signal thinks is zero volts.

This all gets swept under the rug in most cases (f'rinstance, I still use "ground" as a concept in my circuit work) where the total amount of current flow times ground wire resistance is low enough that the erroneous signal it introduces is "low enough" for what you're doing. Making metal-head-banging-MegaBlitz-Belchfire distortion pedals makes any input ground offset crucial.

[Rob Strand]

In my view, no, absolutely not. You want Ground to be all the same, everywhere. That's the ideal. Voltage drops between one bit of "ground" and another bit of "ground" are where the problems start.
The point of star grounding is to avoid cumulative currents flowing through long daisy-chained wires and making voltage drops due to the resistance of the ground path. The bigger the currents, the bigger the volt drop. Star ground minimises those drops and gives each area its own direct path back to the central ground. So *adding* a deliberate resistor in the ground path seems like a very bad idea to me.

What you have said is true.   However, now and then you find some audio equipment which has hum from the day you bought it.   At first you don't notice it then later on it annoys you then you set out  to fix it.    This mostly happens with equipment with non-isolated input sockets.   Back in the day, isolated input sockets weren't available.    Non-isolated sockets makes it hard to adhere to a star connection.

In a few of these cases I've cut the the ground and added a 10 ohm resistor across the bridge.   The thing to realize is the hum problem is due to multiple grounds so the 10 ohms resistor is to stop "bad' ground currents going down one of the ground wires and let the circuit on the other side of the 10 ohm resistor use the other existing path.   It's hard work to find the best place to split the ground.  Often the 'other' existing path might not be that clean either.   The point here is you are making the most of a bad situation.

These days you would add isolated sockets.

The hum can be from the power supply, from ground loops when connecting one device to other devices (say a preamp to a power amp), or from eddy currents in the chassis.

In some equipment you might see the circuits like this, see the 10 ohm resistor in the bottom left,
http://solaris.no/electronics/power_amp/ETI300W/eti300W_curves_1280.gif

[POTL]

In my view, no, absolutely not. You want Ground to be all the same, everywhere. That's the ideal. Voltage drops between one bit of "ground" and another bit of "ground" are where the problems start. The point of star grounding is to avoid cumulative currents flowing through long daisy-chained wires and making voltage drops due to the resistance of the ground path. The bigger the currents, the bigger the volt drop. Star ground minimises those drops and gives each area its own direct path back to the central ground. So *adding* a deliberate resistor in the ground path seems like a very bad idea to me.

Remember that I'm totally uneducated in electronic engineering though, so if people with more background want to tell me I'm completely wrong, Im happy to learn something new.

In fact, I see no problem in changing the ground voltage.
The main ground is connected to the audio jacks and power connector as well as the audio circuit.
The voltage will only change in the control circuit, LFO / Envelope detector etc.
The control circuit will be isolated from other elements and for active components the voltage range between power and ground will simply be reduced.
Other parts of the circuit, as well as other effects in the board, will not contact the isolated ground, neither through the audio jacks, nor through the power jack.
But this is a theory, perhaps in practice it does not make sense.I try to learn from the example of large and experienced manufacturers.
And among the new generation of manufacturers (EQD, Catalinbread, Walrus etc.) JHS are now the most experienced and their products are the most interesting and complex.