Confusing Info on Grounding Layouts...

[EETStudent]

As I progress through my first "original" build (by that I mean not following an old Fener layout and trying to build an amp based on my own experience), Ive come across some confusing information regarding the "best" grounding scheme. Surely there are multiple ways to successfully achieve a quite grounding scheme. Ive probably read the article posted on Aiken Amp's site a hundred times, and I was able to get a good grasp on it. When I implemented it, I was getting just a SLIGHT buzz(grounding). I know I implemented it correctly according to his instructions, but I would like to perfect it. I went with the 2-star ground scheme, where the first star point is located at the first filter cap node. It is joined with the PT center taps, output tube grounds, output transformer ground, and then a wire run back to a PT bolt. The second star point is located at the second filter cap, and it joined by the remaining two filter caps, and then individual wires from local preamp grounds are run back to their respective filter cap. Meaning, the input jack ground(isolated), and V1A cathode are tied together and run to their associated filter cap. And so on...

I am confused now after reading a little about Trinity Amps's recommended grounding layout. He also points his users towards the Aiken Amps article for more information. He recommends a single star point where all local grounds are tied together, and individual wires are run back to the single star point(at the first filter cap negative), and a single wire is run to the chassis. BUT, in some of his layouts, he shows a scheme where ALL preamp grounds are strung together and grounded at the chassis near the isolated input jack, and all power supply grounds(including all filter caps) are tied together and grounding at a PT bolt. Can anybody offer further clarification?

[R.G.]

I can offer a clarification, but you probably won't like it. And the post will be followed by posts from people who disagree with it for various reasons.

The real, true, only secrets to grounding are:
1. Know what currents flow through the ground resistors (since all wires ARE resistors) and keep those from interacting in a bad way.
2. If you can't do (1), you can use cancelling/offsetting currents to lower the current-induced voltages and cancel things. You can only do this if (a) you really, really know what currents flow through what conductors or (b) you easter-egg around and get lucky.

Star grounding in the pure form is the only prescriptive, known-to-work-ahead-of-time variant of (1). It solves the issue of interacting ground-current-caused offset voltages by means of forcing the currents to flow in separate conductors where they simply cannot cause unwanted interactions.

Star grounding at audio is possible, but only barely, because of the large number of wires. Practicality and sloth make humans seek ways to not have to do this. The most motivated humans will come up with rationalizations about why star grounding doesn't or won't work, in their view. This makes them not doing it an OK thing, and casts doubts on people who do. What's not to like about that?

The most common star-derived way to do grounding is to gather things into local "stars", then star ground the local stars. This works pretty well, and saves a lot of wiring. It works pretty well because most local stars have resulting ground currents which are simple, not serving to feed back voltages in a phase that causes oscillation, and not having enough gain served by the local star to cause oscillation if it happens to be phase-ugly. Local stars tend to not include hum-inducing parts as well, so they get good(ish) marks on that.

Star grounding is not desirable at RF. At RF, field effects conspire to make ground plane techniques best, as the return current in the ground will follow the sent current above it in the plane, and the grounds will tend to NOT interact if the rest of the layout is reasonable.

So the clarification is: grounding is not prescriptive, unless you do star grounding. In the case that you don't, you have to be either very good about knowing what current wants to/has to flow from all points noted as "ground" back to the one-and-only place in the circuit that is real, true, honest-to-deity, zero volts, or you have to be lucky. There exist an infinite number of grounding layouts that are equally good to star grounding. But they are not known ahead of time, and are layout and circuit-dependent. (That means, you have to make them up to fit the circuit and physical layout.) This probably accounts for the variations in recommended grounding schemes.

[defaced]

Have you identified the frequency of the offending noise by scoping, measuring, or some other means?
Is the chassis open, or is the component side shielded some way? If it's not shielded, you may be fighting EMF while thinking it's grounding.
Are you plugged into anything else besides your guitar, amp, and speaker?
Does it present this noise with nothing plugged into the input?
Does the noise change with the control settings?
Is the earth ground on your outlet good?
Have you tried chassis referencing the other end of the circuit (i.e. the input) instead of the first filter cap?

If this leads you to a dead end, can you post clips and pictures of your build?

Grounding is a great rabbit hole because it's often taught poorly and somewhat difficult to understand, so you can waste alot of time looking down it if you don't consider other more practical options first.

[loudthud]

R.G. gave a pretty good outline of what I've found works best. In TUT 3, KOC goes into it in depth with illustrations to make it clear. Link: TUT3 - The Ultimate Tone Vol. 3 . It may seem expensive, but it's one of the best investments I've ever made.

[R.G.]

Sorry. I feel better now with that off my chest.

As I progress through my first "original" build (by that I mean not following an old Fener layout and trying to build an amp based on my own experience), Ive come across some confusing information regarding the "best" grounding scheme. Surely there are multiple ways to successfully achieve a quite grounding scheme.

Yes. There are an infinite number.

In all of this, you need to remember that there are only two ways to get exactly zero volts between two points. One is to connect them with a superconductor, the almost mythical connection that has no resistance to electrical flow, the other is to make the current between them zero. With real, non-superconductor wires, you can either have zero current flow between two points, or live with the voltages not being the same.

Ive probably read the article posted on Aiken Amp's site a hundred times, and I was able to get a good grasp on it. When I implemented it, I was getting just a SLIGHT buzz(grounding). I know I implemented it correctly according to his instructions, but I would like to perfect it. I went with the 2-star ground scheme, where the first star point is located at the first filter cap node. It is joined with the PT center taps, output tube grounds, output transformer ground, and then a wire run back to a PT bolt. The second star point is located at the second filter cap, and it joined by the remaining two filter caps, and then individual wires from local preamp grounds are run back to their respective filter cap. Meaning, the input jack ground(isolated), and V1A cathode are tied together and run to their associated filter cap. And so on...

I am confused now after reading a little about Trinity Amps's recommended grounding layout. He also points his users towards the Aiken Amps article for more information. He recommends a single star point where all local grounds are tied together, and individual wires are run back to the single star point(at the first filter cap negative), and a single wire is run to the chassis. BUT, in some of his layouts, he shows a scheme where ALL preamp grounds are strung together and grounded at the chassis near the isolated input jack, and all power supply grounds(including all filter caps) are tied together and grounding at a PT bolt. Can anybody offer further clarification?

There are some highly specific requirements on the incoming power supply side of things. They are driven by the need to serve many masters with a grounding scheme. The rectifier/filter setup in an amp is a noisy thing, with BIG pulses of current running through it at twice-powerline frequency. There is a lot of harmonic content in the pulses, so what current travels in what wire matters.

For grounding, think in terms of current loops. There are two current loops that include the first filter cap. One is on the transformer/rectifier side, and includes the BIG current pulses through the rectifiers. The other is everthing on the other side. Big current pulses cause voltages even in small resistances. There *will* be pulse voltages on the wires leading to the + side of the first filter cap, and on the wires coming out of it and going to the transformer center tap. The way to prevent these from making noise on other wires is to force the currents to travel on specific wires. In particular, the wire coming from the rectifier(s) to the (+) side of the cap need to go only there, and the wire from the CT of the PT needs to go only there. The CT wire from the transformer should not go to the chassis, or to any other point except the (-) terminal of the first filter cap.

If that wire hooks other places, then some of the currents between the first filter cap (-) and the actual transformer CT cause voltages which are impressed on the points that hook to that wire. These are VERY low impedance voltages, and they WILL jerk that other voltage point around. If it's part of your signal wiring net, you WILL get a buzz from the pulses.

You get to define where the One True Ground is. It is usually simplest to call the (-) terminal of the first filter cap the One True Ground and measure every other voltage from it. From that One True Ground, other wires carry the ground point to other places.

The transformer CT and rectifiers simply cannot jerk the other grounds around at twice-line-frequency if you do this right, because whatever voltage they induce is only found on the wires to that loop. So if you reference an input to the (-) side of the first filter cap, it will be at the One True Ground. [Note to purists: yes, I know this is a simplification, but it's a useful one for the original poster.]

If all "ground" wires star out from that point, they cannot interact with one another. The output stage, for instance, uses a lot of current, and produces currents on the ground wire that are full-wave-rectified versions of the signal current in the OT primary. By forcing those currents to travel on their own wire, they cannot interact (i.e. feed back) with the earlier stages.

There are several kinds of ground. There is reference ground, a source of pure 0.0000000000... volts used to tell delicate inputs where ground is. If this wobbles around, you will get interference on those inputs to the extent that they do not have common mode rejection. The most common triode stages in guitar amps have 0db common mode rejection, so they need a clean ground reference. There is shield ground. This is the chassis, and it forces RF fields to travel around the circuits instead of causing RF reception in them. There is safety ground, which keeps you from getting electrocuted. And there is what I call "sewer ground", which is the return of all the "used electricity" currents from the amplifying stages.

You only get two choices. You can separate these on different wires, or live with how they interact in the wire resistance.

We need a shield ground to be around the whole circuit to keep RF pickup out. We know that we need that shield ground to also be safety ground, and be able to conduct perhaps 25A of current back to the AC power line safety ground wire to keep us from being electrocuted by a failure in the AC power line circuits. If we try to get cheap and connect other signal ground things to this ground, we may get current caused ground interactions. Signal ground should connect to the shield/safety ground shell in 1.0000000... places. Any more than that and we're counting on being lucky. Sometimes we get away with this, but if we also try to use the metal chassis/shield for speaker return currents - well, we better be lucky, cause that can cause intractible oscillation. If we run some of the rectifier pulse currents through the chassis, well, the sensitive inputs just may be jacked around with an audio buzzzzzzzz.

There are three choices on input grounding. These are (1) hook them to the chassis out at the jack and hope for the best, (2) hook exactly one of them to the chassis out at the input jack, insulate the rest of them, and do NOT connect any other signal grounds to the chassis, and (3) insulate inputs from the chassis and connect them instead to the other place we've chosen for "ground" inside the grounding net. There are many more mixtures than this, but (1) is what mostly gets done, (2 and (3) can be shown to work well. Other than that, you're counting on luck.

But I'm wandering. There should be exactly one connection of power and signal ground to the chassis. My preference is to make this be from the first filter cap (-) to a place of my choosing that's not a PT bolt. The PT bolt should not be used for connecting the AC power line safety ground to the chassis for safety reasons. Safety specs say to make the AC power line safety ground be a separate bolt to the chassis, used for nothing else.

You then have the choice of connecting the chassis to the One True Ground at any point. I usually cheat and once I've made the safety ground bolt secure, take a wire from the (-) terminal of the first filter cap to a completely separate ring terminal/star washer/nut on that same bolt. That pretty much says to insulate input jacks from chassis, or count on luck. Often you get away with luck, but sometimes you don't. Insulated input jacks and a wire from the One True Ground to some chassis bolt works. So does making one (or counting on luck, several) input jacks' ground lug being the connection to the One True Ground.

Bad idea to connect speaker return to the chassis. It ought to go on a wire back to the output transformer. It can, if you insist, be connected to signal ground through a large resistance, but no other signal ground connections should connect to it.

As for making two, three, or more stars on various filter caps, you have to remember: all wires are resistors. There is a voltage across every wire which is V = I * R. If you ground some circuits to the first filter cap (-) and others to the second filter cap (-), then the two "grounds" are offset by a DC *and* AC voltage equal to V= I*R. Maybe that's OK-ish, maybe it even cancels some ugly stuff depending on your circuits and the physical layout of the ground/wire/resistors.

Maybe.

There is one useful way to think of using the second filter cap (-) as the One True Ground. This is if the first filter cap (-) has exactly two wires from it: one to the PT CT, the other to the second filter cap (-). If you do this, for the lowest noise and so on, the first filter cap (+) side should also have exactly two wires, one from the rectifiers and one to the second filter cap (+). This makes the first filter cap a "prefilter" cap, where it eats the pulse currents of the rectifiers and lets only pre-smoothed currents flow to the second filter cap, which now acts like the first filter cap to the rest of the signal circuitry.

This is different from simply paralleling two caps to the extent that the wires between first and second filter caps have resistances. Otherwise, it's just like paralleled caps.

Remember: all wires are really resistors.

[Chuck H]

Oh man! Grounding has been covered here A LOT. And the reason is that so many philosophies seem to contradict. R.G. has done a pretty good job of employing caveat emptor. Explaining that there are different things that are KNOWN to work... or not

But the important thing R.G. did is explain the different pitfalls that can occur even when you believe you've idealized. His info wasn't introduced that way, but if you read between the lines it's there. And that can give you a head start on troubleshooting ground issues more quickly. Even if you've done this sort of thing before, and I have, you can still come across quandaries. My latest build uses the same tubes but swaps duties to employ either a BF type amp OR a Vox TB AC type amp. Right down to moving the TS, adding a cathode follower, the "high cut" circuit and lifting the NFB. The amp also has reverb and bias modulation trem!!! Grounding was carefully considered before I started this build. With all the 0V references for the switching, a small transformer within the preamp (reverb) and an oscillator I had to make a few concessions. My point is...

I did end up moving a couple grounds. The amp is dead quiet. The sort of spooky quiet that makes you wonder if it's broken when you flip the switch. Success! And I promise that everything you need to know about tracing and fixing ground issues has been covered by R.G. in this thread.

Good luck.

[defaced]

Once you've digested R.G.'s incredibly helpful posts, here's a few things that helped me when I was still struggling with grounding:

Merlin's grounding PDF helped me get the basic "picture" in my head when reading about ground a 100 times didn't help: The Valve Wizard

Once I had the picture, reading helped take it the rest of the way home. I found Tim William's Circuit Designer's Companion to be very helpful. The first chapter is dedicated to grounding (but that's not all, this books is packed full of practical information that makes it very worthwhile). You can preview most (if not all) of the grounding chapter through Amazon and Google Books.

And lastly, I found the attached ground scheme helpful in seeing the "whole picture" at once as it relates to an amp. It's based around the Soldano SLO, but can be easily adapted to other amps. One would call this a "bus" ground scheme. I find it easy to apply and effective. Is it the "best"? Probably not, but with an iron in your hand and an evening you'll have tried enough grounding techniques from the ones outlined in this thread to give you a good feel on what I'll call "the order of magnitude" of how different grounding techniques effect the end result. Sketches like this will also be helpful to keep in your notes for future reference so you can keep track of what worked and what made more noise than a kid with pots and pans.

Image courtesy of Joey Voltage.

[Enzo]

And I'd like to add that hum is not generic. Hum can come from a whole bunch of different sources, and each one has its own cure. Grounding will do nothing to cure hum from radiated fields, nor will it help poor filtration.

One needs to be sure the grounding is the actual problem. Otherwise we can spend hours messing with the ground arrangements only to find out later we had heater to cathode leakage, resulting in hum.

[Tom Phillips]

And similarly, if the problem is a ground loop no amount of messing with the heater circuit, such as adding an elevated center tap or using DC on the pre-amp tubes, will solve the problem. I recently quieted a Vibrochamp and it turned out that the stock heater circuit that uses the chassis for the connection to one side of the heaters did not need modification. That will be heresy to those that always say that you must change to a center tapped twisted wire scheme. It surprised me too but in the course of working on the amp I left that step to last and it turned out not to be necessary. There are many variables.

[Wombaticus]

One other point about star grounding that is worth repeating is the fact that the presence of a star point in no way implies a chassis connection at that spot. I suspect that many beginners harbour the misconception (as I did) that local stars imply that there must necessarily be a connection to chassis at that point, when in fact it will as often as not compromise the whole scheme if this approach is taken. These days I almost always end up employing a "bussed stars" ground scheme approach with a few local stars -- yet with only one actual chassis connection point (usually at the input). It hasn't failed me yet. The key thing to remember is that in well-conceived ground schemes, the signal ground chassis connection point turns out to be an almost an incidental afterthought... no current should ever flow through it except in a fault condition except the microscopic amount that allows it to serve as a reference... the other grave misconception of course being that the ground return for the whole circuit flows through this connection. This misconception, I think, is why you see people sometimes trying to locate the circuit reference ground near the safety ground connection -- following the third great mistaken belief that the earth/safety ground is "the well from which all electrons must spring". But of course, there should also be no current flowing in earth/safety ground except in a fault condition.

[txstrat]

For those who are interested. Tim Williams' companion can be found here: http://diagramas.diagramasde.com/otr...0Companion.pdf

[Steve Conner]

There is no such thing as an ideal grounding scheme, but there are a lot of satisfactory ones. RG does a great job of explaining all the possible pitfalls, but it takes practical experience to know which ones are significant and which are just theoretical worries. In guitar amp work, only a few of them are significant. Tubes have lots of internal noise that masks a multitude of sins, and we're just listening to the output, not doing 24-bit measurements on it.

Bottom line: The charging current between the rectifier and first filter cap, and the cathode current of the power tubes, have to be kept out of the signal ground system, and I forgot the other things.

[Chuck H]

Bottom line: The charging current between the rectifier and first filter cap, and the cathode current of the power tubes, have to be kept out of the signal ground system, and I forgot the other things.

WRT guitar amps, that's it in a nutshell. Well, that and avoiding ground loops. What my ground topology has morphed into, and it's worked for me so far, is:

One star point at the power supply end of the chassis where the PT CT's, main filter and heater grounds go. Then another star near the input where everything else goes. And I run a wire from every circuit ground to the star point. No daisy chains. Safety ground is bolted to the chassis near where it enters. And all jacks are isolated from the chassis. That's basically it. Sometimes I have a third star at center chassis where things like reverb drive, NFB and the PI are grounded. It's A LOT of black wire!!! So it can look messy if your not careful. I locate the star points under the board with the leads (and extras) already attached. Then when the board is installed I just trim and solder them in place. Looks tidy. I shrink tube the ends of any unused leads from the star and tuck them under. So if I need a ground for a repair or modification I can fish it out and I don't have to lift the board.

Not that anyone asked... But I hope this helps.

[EETStudent]

Man, what a great list of replies. R.G., thank you for getting the ball rolling and for giving me something to ponder. I admit a lot of information went well over my head at first, but Ive been able to digest much of it. I SHOULD have mentioned that prior to my post, I was noticing the noise, which I believe to be subtle buzz due to a grounding issue, with only the rectifier, power, and phase inverter tube installed. So to Enzo, I feel pretty confident that it wasn't another issue. Im excited to go through all the info and figure out what works for me

[R.G.]

Here's some other points to ponder.

You say "buzz". Is there any way you can get that displayed on an oscilloscope and see what it looks like?

As Enzo mentioned, there are a large number of things that get labeled "hum" or "buzz". Seeing it on a scope lets you see if it's
- mostly 60 cycle sine-ish, in which case it's power line hum, and this can come from several places
- mostly 120 cycle stuff, meaning it comes from a full wave rectified source somewhere
- mostly spikes at 120 cycles, which is a sub set of the ripple/rectified stuff.

Spikes, or bursts of RF noise at 120Hz is often from bad connection of the transformer/rectifier return getting mixed into the signal ground. However, it can also be blats of RF from solid state rectifiers slamming off at the end of conduction pulse and resonating with the wiring loops in the amp. Some rare times, it can be a loop of wire from the PT to rectifiers and back coupling magnetically to another loop of wire in the signal line. Can also be bum/faulty fluorescent lights, computer power supplies, etc. on nearby AC power line circuits.

Ripple is usually triangular or sawtooth. Ripple at power line frequency instead of two times power line means that only one rectifier polarity is running; I found a bad full wave rectifier this way once.

True sine wave hum cannot be coming from rectifiers and filters. And if it's not at power line frequency or some multiple, it's not power line related at all.