Star Ground Confusion

[mcgruff]

I've been reading up about star grounding and I think I get the idea behind it ie bits of the circuit which send big currents to ground can create voltages across the ground wire resistance which may create hum in other parts of the circuit which are connected to the same ground wire, eg a preamp stage.

So, in a star ground, different parts of the circuit each have their own connection to the main ground point (the earth pin on the IEC socket). In practice, individual components in different parts of the circuit will be gathered into nodes which connect to the main ground point. A safety (chassis) ground also connects to the main ground point, and this is the only ground connection to the chassis.

What's bugging me is that the cable connecting the IEC socket to the mains socket appears to be a shared ground wire. Doesn't this defeat the purpose? My inadequate theoretical understanding makes me think there should be multiple IEC sockets, all connected to ground (but only one with live mains voltage), to provide multiple independent paths to ground for a star ground system.

Obviously I haven't understood something about the way this works and would be grateful if someone could help me out. I have to design a good ground system for a project I'm working on.

[eschertron]

My basic (very basic, thank you!) understanding of star technique includes what "zero" means and where it should be located. From what I've read, the absolute zero voltage point should be whatever the lead coming off the negative reservoir cap is attached to. From that, the HV CT (or bridge ground) and each of the substars are referenced. The PE pin on the line voltage connector connects to the chassis. That's it. The chassis is referenced to the star ground, but it's a shield and so is of secondary importance to the signal grounds. There are very many threads about this subject. Refer to anything written by RG Keen.

What do you mean "the IEC socket to the mains socket" are you referring to the wall jack? A little confused

edit: after reflection, I think you may be talking about grounding a system of unconnected components, like the electrical distribution in a venue?

[dai h.]

(I'm not an expert but this is based on my understanding/reading) :


the safety ground isn't necessary for the amp to work. It's there primarily for safety (to provide a lower impedance path for fault currents to go back through instead of through any human's body (who may be touching the ground of the amp--usually the guitar bridge via the cable shield) to the floor or conductive path (worst case is the current going through your heart, your stoppping, you don't recover spontaneously, and no one is there to help(!)). Secondarily, the safety ground can help reduce noise by making a good (sort of) "sewer drain" path for leakage currents from the AC coupling capacitively to the chassis from the power transformer (if the grounds are wired correctly, they stay there when the unit is energized and keep flowing around and around and don't get into the audio (or enough to be a problem). I think it can also matter how far away the safety ground to actual ground/earth is and how good (low impedance) it is.

[guitician]

If your talking home wiring, the base of the star, for the third prong, would be the ground buss bar in the breaker box, where all the romex cables, that have that bare copper ground wire, end up tied to. In an amp, that same idea holds true. Run all your grounds to a single place. That limits the build up of any non-zero resistance that can lead to ground loop noise.

[sgelectric]

Here's some reference material:

Grounding

SG

[mcgruff]

Thanks for the replies. I think I'm beginning to understand that there are two different grounds: a common ground reference voltage for the circuit and an "earth ground" which comes from the mains socket earth pin.

So, if I understand this correctly:

(1) A good star ground gives different parts of the circuit their own connection to the common circuit ground so they don't have to share a wire with a noisy part of the circuit. (Thanks for the Aiken link. I've already read that a couple of times and it's slowly starting to sink in).

(2) After that, the common ground is connected to earth.

I guess it's better to connect the common ground to the IEC socket earth pin rather than the chassis?

[Chuck H]

I've read it both ways. ie: locate the chassis ground near the AC input with the "green lead" grounded very near that point. And also that you should locate the chassis ground very near the input jack.?.

I guess the reasoning in placing the earth ground near the AC input would be to keep any high current paths short and away from the input.?.

And I guess that the reasoning for placing the earth near the input would be that if signal paths have the shortest leads there is less trouble with noise from longer leads and their subsequent resistance.?.

There's an awful lot of ambiguity about it as far as I can tell.

I, personally, locate the earth ground near the AC input. That's the point where my other ground points terminate. I use one ground near the input for all preamp grounds (including associated filters). I use another for the PI and presence as well as any reverb or trem grounds (same for dedicated filters). And I use one for the power amp, CT's and main filters. All are isolated busses. This has worked well enough for me that I have stopped trying to interpret the logic that different tech's site for grounding schemes.

YMMV

[Tooboob]

Current comes out of the PT secondary, flows through the bridge, standby switch, then hits some big caps, heads over to the preamp tubes through more resistors with more caps, and also goes through the OT primary and flows through the tubes (the lower preamp tube current bypasses the output through the preamp tubes), eventually reaching ground, and heads back to the PT center tap, right? Not really.

There's certainly a DC path there, and low-frequency AC can make it around those loops too, but most of the caps with high voltage and capacitance ratings in the design are the primary sources and sinks for audio frequencies and above. The preamp and PI tubes have dropping resistors followed by caps on their supplies. The caps should share a ground with the associated tubes. The preamp tubes are generally fairly low input swing and high gain. Noise on their input is bad. They should probably get their own star so that the stages agree about what ground is. B+ has lots of capacitance on it, and lots of line-related current through the caps. For audio frequencies, they are a big current load on the PT and also the supply for the OT. So do the caps go near the PT or OT? Fortunately, the OT tends to cancel B+ noise on a push-pull amp. You can put the PT and OT at one end with one rotated 90 degrees with respect to the other. and have the caps near both, stringing the PI and preamp tubes out to the input jack on the other end, and need an off-center handle due to the imbalance, or you can put the PT and OT on opposite ends and shield the B+ line to keep the B+ loop from radiating. What goes out as B+ mostly comes back from the power tube cathodes, so the power stage ground can be wrapped with B+ inside the shield. Running the lines on the tube side of the chassis (twisted) through grommets works too, but it seems kind of distasteful.

Much of this is a source of gentlemen's disagreements, and I didn't even go near input jack or PI grounding. The big points I'm making are that supply filter capacitors and the supply loads form loops you want to minimize, twisting wire in big loops makes them smaller, interconnecting circuitry to high-gain inputs should share local ground with the inputs, you don't want to ground a sensitive circuit to a high current circuit's return on the way back to the high current circuit's supply, you should minimize current through the chassis, and you should generally strive to make ground connections low-impedance. All these are good electronics practice. They are common elements of all successful grounding religions (and there are several).

There shouldn't be any appreciable current to the IEC socket earth pin. That's why the amp would still work with a 2-prong adapter, though it's a very bad idea for a tube amp. Terminate the IEC socket earth pin to the chassis using a single wire with crimped connections, star washers, etc. If the connection fails, you may not notice until there's a fault resulting in a live chassis. The whole amp basically floats without the earth ground connection. The earth ground connection references the whole amp to earth ground, by a single wire, with a reliable high-current connection for safety in the event of a ground fault, allowing the fuse, a breaker, or a ground fault interrupter to blow.

[jazbo8]

I guess it's better to connect the common ground to the IEC socket earth pin rather than the chassis?

Absolutely not! Please read the Aiken article (linked above) and Merlin's paper on grounding. The IEC earth terminal must be connected to the chassis for safety reason, it should NOT have any other connection on it.

"These appliances must have their chassis connected to electrical earth (US: ground) by a separate earth conductor (coloured green/yellow in most countries, green in the US, Canada and Japan). The earth connection is achieved with a 3-conductor mains cable, typically ending with 3-prong AC connector which plugs into a corresponding AC outlet. The basic requirement is that no single failure can result in dangerous voltage becoming exposed so that it might cause an electric shock and that if a fault occurs the supply will be removed automatically (this is sometimes referred to as ADS = Automatic Disconnection of Supply)." From Wikipedia - Class I Appliance

[Jazz P Bass]

To clear the air on IEC earth ground:
A properly wired 3 prong socket has the Black wire 'Hot'. 110Vac/ 220Vac (depending on country)
It provides the working current for the equipment.

The White wire is the 'return' for the circuit.
The current passes from Black, through the device & on to the White wire.

The Green wire is the safety wire.
It attaches to the same buss bar as the White wire at the mains breaker panel.
The safety part is the fact that the green wire is a separate path back to the panel.
If the devices chassis (which the Green wire is attached to) becomes live/ hot (through a fault in the device) the current will pass down the Green wire & trip the breaker that the device is hooked up to.

[R.G.]

Let me reinforce: grounding is all about where the "ground" currents flow, and what that current flow does to cause voltages across the conductors it flows in. It is complicated because you have to really understand what currents flow where to do this well. Simple grounding schemes are approximations that tend to be useful to cover many unforeseen things, and take the place of really understanding what currents are let flow through the circuits.

Star grounding is the only grounding scheme that can be proven to remove ground-current-generated voltage interference ahead of time, for any circuit. It does this by the brute force method of keeping every current in its own ground return wire, and as a side effect produces vast complexity of getting that many ground wires in.

Other grounding schemes, such as modified star, buss, chassis grounding, whatever, work fine in some circumstances, not others. They work to the extent that phase cancellation of currents and low ground wire resistance keep the voltages generated at sensitive ground points - like input references - low. They work best at low currents (obviously) and worst with high currents. They work differently with different physical layouts and different circuits (i.e. different currents).

Safety/Green Wire:
This wire should carry almost no current unless there is a fault in the power AC wiring or transformer primary, and then it's intended to conduct that fault current back to the electrical supply while keeping the chassis voltage less than a dangerous voltage. The only currents it carries in normal operation is the leakage currents from the AC wiring and parts, and this should be dramatically smaller than the AC power currents. Modern safety practices require this to be connected to the metal chassis at one point, that point being used for nothing else. It can be thought of as a safety rope for a rock climber: in normal practice, it does nothing, neither helping nor hindering the climbing, just there in case there's an accident. It's a minor issue, requiring just not getting fouled up in it while climbing.

Power supply ground:
There are one or two of these. In the one-ground scenario, the negative (for tube amps) terminal of the first filter cap and the mass of copper directly connected to it is the ground. The negative side of the rectifiers or CT of the power transformer must connect here and nowhere else to keep rectifier pulse hum down. The DC returns of all the circuits must eventually come here. These "sewer grounds" return the "used electricity" to the DC supply for recycling. Each of these wires has a voltage on the other end equal to the sum of the currents going into that wire times the wire's resistance (and inductor/capacitor side effects, ignorable at audio in most cases.)

You can use a two-position "ground". This involves having the first power filter cap receive currents from the rectifiers, and then taking its + and - leads to a second power filter cap through wires. The second filter cap - lead is then the reference AC and DC ground for the chassis. This minimizes hum by keeping the rectifier pulses in the first power filter cap, and moving the reference ground voltage over to the second filter cap, which does not then handle the high pulse currents of the first filter cap. This is improved by having some resistance between the two caps; with short wires, there is no effective filtering, and they look like a single lump of capacitor. Even sub-one-ohm resistors in both wires help. Note that this is a hum reduction strategy, not a help with interference/feedback from the circuit itself.

Signal reference ground:
Some of the ground return wires carry either no or negligible current. The input jack ground wires are an example. There's so little DC in these wires that they can generally be thought of as carrying zero. That's the only way a wire can have the same voltage at each end, when it carries zero current. There is a complex chain of reasoning about whether to connect these jacks to the chassis or to isolate them and connect them to the actual input circuit element ground back at the tube socket. The complexity is whether there are other inputs and/or outputs connected to the chassis, whether the chassis carries any current or not, and how much these currents are.

Local grounds:
In an isolated section of circuitry, "ground" currents may tend to cancel. This is the case for a triode section or two triode sections successively amplifying the same signal. And since a three-terminal active device gets its input reference signal from two of the three terminals, there is always some local ground current offset. It makes sense in many cases to use the tendency of signal currents to cancel locally by tying the ground ends of such a circuit together into a "local star". And since we can separate the effects of DC and AC currents with capacitors so easily, it makes sense to have a local decoupling cap to shunt the AC currents around in a local current loop to keep that AC current off the ground return wire(s). This matters most with high frequency capable amplifiers and high frequencies, and is why all digital chips and most analog chips tell you to decouple the power supply voltages with good high frequency caps as near the pins as possible. This keeps the high frequency AC in a tight local loop, minimizes the loop size, and supplies the fast voltage/current demands locally. This works the same in tube circuits, even though the "integrated circuit" is one or two triode sections. A decoupling cap from the local plate supply wire to the ground side of the cathode resistor(s) conducts the AC around in a loop locally and keeps much of this current off the DC ground return wire. All good designers use local decoupling until it's proven by Muntzing that it's not needed.

Output Grounds:
Special care must be taken with the high currents in outputs, especially speaker outputs. The best practice is to isolate the speaker jack from the chassis, and return its "ground"/common to the output transformer. All the speaker currents then flow from and to the transformer winding. If these currents share a part of the chassis metal with input jack grounds, you can have interference by the speaker currents wobbling the input jack grounds around.

For amps with feedback, best practice is to return the speaker currents to the output transformer, then use a separate resistor or conductor from the OT common to the local ground of the PI, which is where the feedback is intended to go.

This is only the broadest brush on grounding. As I said, the right way to do it is to know all the currents and all the grounding resistances, and take them all into account. This is a huge task for most people. The well-worn solutions known from vintage amps are usually copied, and often work OK if not perfectly, because there was a LOT of effort in those old amps to get grounding acceptable if not great. And in many instances, copying works well enough.

Mother Nature will always tell you whether you guessed right or not.

[trobbins]

With respect to the speaker output for valve amp - which originates from an isolated OT winding - it is considered better for safety to connect it to the 'ground' of the amp, otherwise the speaker wiring is a floating voltage source, and if the voltage is high enough (as in some amps) then it would be a hazardous floating supply. Of course if feedback is taken from that winding then that provides a de-facto grounding path.

From a simple perspective, any point of that speaker winding can be taken to 'ground', such as one end or other of winding, and typically if the speaker connector is polarised then the 'negative' polarity is used. Even an unused tapping can be used, but if the tap is 'outside' of the used portion of winding (eg. a 16 ohm tap when only the 0-4 ohm portion of winding is used) then that makes the speaker signal voltage more hazardous.

There will be capacitive currents flowing in the 'ground' speaker side wire. Those currents would include a return path to OT core (ie. chassis), and a return path to OT primary and hence to output stage cathode ground point. So if the speaker side ground is taken to the 'star ground point', then those capacitive currents flow in the chassis to star ground link, and the output stage cathode end to star ground link - both of the those links may be just the same star ground junction depending on configuration.

[R.G.]

Of course if feedback is taken from that winding then that provides a de-facto grounding path.

It makes a great deal of sense to force this reference to be where you want it - that is, at the local ground of the phase inverter "lump". This ground connection carries the return current for any currents fed through the feedback resistor/cap network into the PI. Since the PI is itself connected to the DC power ground point, this grounds the output well enough and forces the output "common" lead of the OT to be as close to safety ground as are the other accessible metal points - like input jack bushings.

From a simple perspective, any point of that speaker winding can be taken to 'ground', such as one end or other of winding,

But only if there is no feedback to the PI. The PI demands the correct feedback phase. If you have PI feedback, you're stuck with "grounding" the place on the OT winding that gets the phase right. Otherwise, yes, you can ground any ONE point of a floating transformer winding.

There will be capacitive currents flowing in the 'ground' speaker side wire. Those currents would include a return path to OT core (ie. chassis), and a return path to OT primary and hence to output stage cathode ground point. So if the speaker side ground is taken to the 'star ground point', then those capacitive currents flow in the chassis to star ground link, and the output stage cathode end to star ground link - both of the those links may be just the same star ground junction depending on configuration.

Yep. It then gets important to worry about what currents flow through the chassis that you can't control - like capacitive currents from the windings in the OT, and magnetically induced currents from wire loops near/around the chassis, and so on. That makes it important to not use part of the chassis as a signal reference ground unless you know by testing that the current flows in the chassis from whatever source don't elevate the input signal reference grounds and induce feedback or noise. It's also good to remember that the reverb return is a high gain input.

[mcgruff]

Absolutely not! ... The IEC earth terminal must be connected to the chassis for safety reason, it should NOT have any other connection on it.

OK got it (I think). Thanks. All shields including the chassis connect to the common circuit ground but the chassis is the only thing connected to mains earth (the safety ground).

Lots of good info posted here which I'm reading very carefully.

[Chuck H]

The thing is... It all terminates at the chassis!!! Lot's of talk about the chassis being the only "earth" and grounds being separated.?. But where and how then do grounds terminate with a 0V reference if not the chassis???

This is the utterly ambiguous part of grounding IMHE. So... Should the input be nearest the earth ground or should the high current power supply be nearest the earth ground??? I've seen PCB designs where the power supply is nearest and the preamp is grounded via a buss with a small value capacitor from the input side to the chassis. I'm not sure what this achieves, but I'll hope for an explanation and education on it.