grounding multiple choice

[yunger]

GROUNDING.pdf

i hate to admit it but the more i read about proper grounding, the more confused i get.
i am a visual kind of person so a made a pdf with a simple schematic of a se amp i built with different grounding options. which one is closer to the right way? or are they all wrong and why?
also, the ot power wire is connected to the b+ node. would it be better to be on the c5 node?
thanks for looking

[MWJB]

I like Option C the best, but I would give the preamp cathodes (R4 & R7) their own wires to the input jack ground as they are out of phase with each other (grounded end of the resistors may "float" a little above chassis ground).

I tend to work out how many preamp ground wires I'll need (one each for preamp cathodes, preamp filter cap, buss wire to pots & pot mounted tone caps) and run them from the input jack ground/switch terminal out to each component (a cathode resistor and its own bypass cap are fine to share a wire) then dress & trim when wiring up components.

You could hook up the B+ OT wire to C5 as long as you grounded C5 to the power amp ground. This would lower voltage to the OT & power tube plate, via R11.

R9 could ground to power amp (probably doesn't need to, I'm just being pedantic).

Triode power tube?

However "right" you get the schem, the physical placement of parts & wire routing will have the bigger impact.

[loudthud]

None of the above.

[yunger]

wow! great answers. thanks!
none of those drawings are how i actually wired my amp, however, i was completely wrong in my understanding of grounding. for some reason i thought the dirty grounds were the chassis point and the delicate signal grounds were the last in line...don't know why. it definately helps somebody like me to visualize it.

btw, lt, nice photo editing skills.

Triode power tube?

the schematic is a dumbed down version of my "neighbor friendly amp". it actually uses parallel 12au7s for se power tube.
i wasn't really my intention to troubleshoot my amp, more trying to understand the practice. it does hum from the power tube though. not sure if it's 60hz or 120hz yet since my oscope is still boxed up since i moved. maybe fixing the grounding properly will help that.

[Gingertube]

Option A and B vary only in the place where the zero volt "buss" is conected to chassis.
Option A is my preferred Option as it keeps the large circulating currents of the power supply (noise sources) at the opposite end of the 0V buss to the sensitive input circuitry.
Option C is just plain WRONG (sorry MWJB)

I don't really see any difference between Option A and the Loudthud method - he just arranges it as a few local star "grounds" which are then "bussed" together in the CORRECT order and the 0V to chassis connection made at the right place. When translating the diagram to REAL wiring his diagram is probably easier to follow and less likely for you to make a mistake.

Remember also that the 0V to chassis connection does not have to be a plain piece of wire, in fact it is more usual to see a network of a 10 Ohm resistor with back to back diodes and a capacitor across that resistor.

Cheers,
Ian
Cheers,
Ian

[yunger]

now i remember why i was confused about grounding methods.
option A is closer to how i actually wired mine, only i didn't ground to chassis anywhere near the input jack.

I don't really see any difference between Option A and the Loudthud method - he just arranges it as a few local star "grounds" which are then "bussed" together in the CORRECT order and the 0V to chassis connection made at the right place.

loudthud's method is actually closer to option B in that the first stage is closer to the chassis than the power stage, only he corrected the order of the parts inbetween.

[Gingertube]

Take 2 with more care now that I've had my morning coffee and brain is in gear.

Option A and B vary only in the place where the zero volt "buss" is conected to chassis.
Option B is my preferred Option as it keeps the large circulating currents of the power supply (noise sources) at the opposite end of the 0V buss to the sensitive input circuitry.
Option C is just plain WRONG (sorry MWJB)

I don't really see any difference between Option B and the Loudthud method - he just arranges it as a few local star "grounds" which are then "bussed" together in the CORRECT order and the 0V to chassis connection made at the right place. When translating the diagram to REAL wiring his diagram is probably easier to follow and less likely for you to make a mistake.

Remember also that the 0V to chassis connection does not have to be a plain piece of wire, in fact it is more usual to see a network of a 10 Ohm resistor with back to back diodes and a capacitor across that resistor.

A good paper on grounding which shows that 0V to chassis network, intended for HiFi guys but worth a read:
Earthing (Grounding) Your Hi-Fi - Tricks and Techniques

Cheers,
Ian

[yunger]

ah, confusion gone. i like concensus. and i have something to read too. thanks ian.

[MWJB]

Take 2 with more care now that I've had my morning coffee and brain is in gear.

Option A and B vary only in the place where the zero volt "buss" is conected to chassis.
Option B is my preferred Option as it keeps the large circulating currents of the power supply (noise sources) at the opposite end of the 0V buss to the sensitive input circuitry.
Option C is just plain WRONG (sorry MWJB)

I don't really see any difference between Option B and the Loudthud method - he just arranges it as a few local star "grounds" which are then "bussed" together in the CORRECT order and the 0V to chassis connection made at the right place. When translating the diagram to REAL wiring his diagram is probably easier to follow and less likely for you to make a mistake.

Remember also that the 0V to chassis connection does not have to be a plain piece of wire, in fact it is more usual to see a network of a 10 Ohm resistor with back to back diodes and a capacitor across that resistor.

A good paper on grounding which shows that 0V to chassis network, intended for HiFi guys but worth a read:
Earthing (Grounding) Your Hi-Fi - Tricks and Techniques

Cheers,
Ian

Option C is "wrong", explain how? It's essentially 2 star grounds (taking account of my additional comments), one for the preamp, one for power amp, a well proven & widely adopted scheme. Works for me and other builders, day in, day out.

It is not more usual to see 10ohm resistors & diodes to 0v, that is pure fantasy.

[Steve Conner]

In a low-gain, single ended amp, you'll probably get away with any of the above. I prefer B because it allows you to ground the input jack directly to the chassis, which helps keep RF out of the chassis.

C is more or less equivalent to B. No current flows in the power amp ground, and only a tiny current flows in the preamp ground, so the voltage drop in C's extra ground wires is negligible. And both ground wires are inside a grounded chassis, so you can hopefully ignore induced voltage in them, which would otherwise make B preferable to C. (the induced voltage in these two ground wires appears in series with the grid of the power tube, so you want to minimise it, by minimising the loop area formed by the wiring)

[exclamationmark]

It is not more usual to see 10ohm resistors & diodes to 0v, that is pure fantasy.

Depends on who built it - 90% of commercial peavey/mesa/whathaveyou high gain amplifiers (and even newer fenders) will probably employ this method as a ground lift switch, as recording direct from the pre-amp out is a fairly common occurrence (and ground loops will ensue). If you're talking about handbuilt amplifiers, then I guess not too many have effects loops/emulated outputs, or care about being used in this way.

Aside from that, I don't see anything wrong with option C either. It looks to me like the purest form of a star grounding system out of all the options.

[yunger]

However "right" you get the schem, the physical placement of parts & wire routing will have the bigger impact.

In a low-gain, single ended amp, you'll probably get away with any of the above.

i am glad i understand the theory a little better now but i think the above quotes have been my experience so far. i guess iv'e been lucky.

i'm pretty sure what noises i hear in the amp i used as example are from dc ripple. don't think any grounding scheme will help that.

[eschertron]

Good discussion!
Since I'm new to the design process, I'd like to ask why none of the schematics have the power supply ground (C6) going to the chassis at the line ground point. My understanding is that shunting the power supply currents to earth as far away from the preamp ground point is desirable. Where have I missed the point? Thanks!

[Steve Conner]

The power supply currents circulate inside the right-hand half of the schematic. They don't go down the ground wire.

[R.G.]

To add a bit to Steve's absolutely correct comment:
Grounding seems mysterious because almost all circuit work is voltage-centric. We learn to think in voltages for a number of reasons, not least that voltages can be measured without opening up a conductor and sticking in a meter in series.

But grounding has to be thought of in terms of what current is flowing in what conductor or wire. And it also directly contradicts one of our cherished (but slightly wrong) ideas - wires are NOT electrical shorts. They are resistors. Low-value resistors to be sure, but resistors all the same.

And because they are resistors, they do cause voltages to exist between their ends because a current flows down it.

If you happen to do something that seems OK because you think wires have zero resistance, but where the small voltages matter, you get odd and difficult-to-solve grounding problems.

For instance: if you happen to use the same wire to carry speaker return currents, which are large, and the input ground from the amplifier input jack, then you are adding to the input signal a voltage proportional to the speaker current and the wire resistance. While this is a small voltage, there is a lot of gain after it, and this can and has caused oscillations as well as odd tone consequences if the phase shifts in the amplifier make it emphasize some frequencies and not others.

Another example that's frequently a problem: the wire from the high voltage winding center tap carries big currents, and carries them in sharp pulses once each time the AC power line peaks. This wire ought to go directly to the negative terminal of the first filter cap, and nowhere else. If the signal grounding of the amplifier shares any portion of the wire going between the HV center tap and the first filter cap negative, then those current pulses show up as a voltage across that section of wire, and they're big enough to jiggle the "ground" for the rest of the amp up and down at twice the power line frequency. So you get a buzzy hum that's impossible to cure unless you figure out or stumble into changing that wiring.

There are two main schools of grounding - separated-wires grounding and ground plane wiring. The idea that you can eliminate grounding interactions from circuits by separating what current flows on what wire ultimately leads to star grounding.

Star grounding works (at audio, at least, as noted below) and can be proven to work both theoretically and in practice, but as Bruce mentions when he says it doesn't work, this is nearly always traceable to the person not implementing it right. And implementing it right is a tough mental exercise because you have to know what currents flow in each ground wire, and most people simply don't understand the circuit well enough to do this right. That's not a pan about the people doing it, it's a statement of fact and is directly related to the way we all think about circuits, which is in terms of voltage, not current.

The other school, which reduces to ground plane wiring, is the idea that big, expansive areas of thick conductors in the form of sheets of metal is better. And so it is, provably, at RF and in any other situation where the inductance of the ground conductor matters much. All conductors have some inductance. Round wires have more inductance per unit of length than the same area of conductor hammered out into a flat sheet, because the distance that the always-there magnetic field that a flowing current causes is shortest around a round conductor, and biggest in a wide, flat sheet of the same cross-sectional area. And this is frequency sensitive. At DC, all that counts is resistance and conductor area. As frequency increases from DC, inductance matters more and more.

One instance of this is that the lead length of a capacitor make the capacitance itself not matter above a certain frequency. When the inductive impedance of the leads on the cap gets bigger than the impedance of the actual capacitance, then it's no longer a capacitor. At all frequencies above that, it's a DC-blocking inductor. Worse, it can resonate and cause really odd problems. Generally this is only an RF concern but it illustrates the point.

Big sheets of metal or big round conductors do help in that they have more cross sectional area in most cases, so they have lower resistance to cause current-generated voltages. That helps, and may be enough for some cases. They have another hidden advantage in certain special cases - sometimes currents from one section of a circuit tend to cancel currents from another. If your special mechanical setup helps currents injected into ground to cancel other currents, you get lower ground-induced noise, at least for the frequencies where cancellation works. If tinkering with where you connect ground to a buss changes ground-induced noise, you are doing Easter-egg noise cancellation.

It is very much easier mentally to put in big areas and chunks of conductors than to understand the circuits in enough detail to know the currents flowing and provide for them in the grounding net. And running a ground wire to every single place that nominally attaches to circuit ground is a huge number of wires. In practice, nearly everyone who attempts star grounding does a local collection of grounds per circuit section, then runs a smaller set of wires to the actual star ground point, either intentionally or accidentally. But it's still a lot of work, doing something that's hard to understand, so no wonder that it gets done slightly out of kilter.

I mentally distinguish between different kinds of "ground" based on what currents flow through them. There is "reference ground" which are low- or no-current grounds that tell an input where 0.00000V really is. This is, for instance, what you'd like the grid leak resistor on a triode to be connected to.

There is "signal ground", which is the return for signal currents between different circuit sections or different circuit-boxes. For instance, there is an AC current from the plate of one triode to the grid-leak resistor of the next triode in the amplifer. A current flows in this path, and must return to the circuit that sent it by means of a signal ground path.

There is "shield ground", which is what keeps currents generated by incoming RF waves from getting into your circuits.

Ideally, shield ground should carry only the radio-induced currents, and should connect to the audio signal ground in only one place. Often, it's best if this place is either right at the input jack (if there is only one) or if the input jack is "grounded" to the chassis with a low-inductance ceramic cap right at the input jack. This keeps the RF currents from traveling down the input jack ground wire into the amp and out on the shielding where it won't get into audio.

There is "safety ground", which is used to keep AC power line failures and faults from killing you. The same conductive shielding metal cage around your circuits that keeps RF out can be used to keep you from getting electrocuted if it's tied to the incoming AC power line safety ground line. What this does is blow the AC power line fuse or breaker if there's a accidental connection between the AC power line voltage and your chassis, or between AC power line voltage and your signal ground. This is the reason you have to connect signal ground to safety ground. Presumably you don't *really* want to die for your music.

Finally, there is what I think of as "sewer ground". This is the return of used electricity back to the power supply. All of our circuits have to return the power supply current they use back to the DC-maker. These currents are larger than either reference grounds or signal ground currents. We nearly always want to keep the used electricity, which currents may have little or nothing we want on our zero-volt reference lines or our signal return lines, out of the paths that carry the grounds for reference or signal. When these currents get mixed, single ended circuits have no way to ignore them and they get mixed into the signal and amplified again.

It's far easier to make a great, big hunk of metal be "ground" and then just connect everything up to it and hope for the best.

Sometimes it even works OK-ish enough.