Ground scheme Q

[Chuck H]

Should the AC mains plug ground (the third prong) be with the PSU or the input? Most amps, commercial and boutique, seem to have it with the PSU. But A LOT of what I've read here indicates it should ideally be with the input. I don't have enough info in my personal knowledge to logic it out. I hope this doesn't open a can of worms or create conflict I've always soldered it with the PSU grounds and haven't had a problem. But this new amp I'm working on is going to be real finicky with the ground scheme so I wan't to idealize.

[Sowhat]

From what I know and do, which may be crazy or sane is that the Mains input ground (3rd wire) be securely grounded to the chassis with a double lock and nut arrangement, locktited down hard as close to the mains socket as possible. I make it a very short run, definitely less than an inch or centimeter depending on how you measure things. This is definitely a PSU related ground but is actually more like a pre-PSU ground if that makes any sense.

[Austin]

I don't think it matters at all. The transformer isolates the ac mains from the rest of the amp and that third wire connects the amp chassis to the water pipes and earth as an extra safety. Maybe I misunderstand the question though.

[Enzo]

CHuck, why would we want to put the mains earth near the input? You never know what small ground currents might flow, but I'd want them in the far corner of the chassis myself.

[Sowhat]

I second Enzo's comment, ground loops are your enemy and the bigger the loop potential (like the mains) the bigger the threat. On most of my more recent builds I do the short chassis run like I mentioned and then take that through a bridge rectifier shunted by a high frequency cap and a surge resistor before I hook to my PSU. This isolates the chassis ground completely from the rest of the circuits while still maintaining a good circuit ground. The bridge is actually wired as back to back diodes and not the way you wire a bridge off the power transformer secondary... a little trick I learned from Rod Elliott, check out Elliott Sound Products. This may violate some politically driven safety laws but who really listens to those guys anyway, they know nothing about good sound!

[dai h.]

Should the AC mains plug ground (the third prong) be with the PSU or the input? Most amps, commercial and boutique, seem to have it with the PSU. But A LOT of what I've read here indicates it should ideally be with the input.

I think the interpretation is a bit skewed and that those recommendations are not about wiring the AC safety ground wire all the way to the input (though in specific cases might help) but keeping all the noisy (yet not well filtered) return paths segments still containing error/noise inducing ripple off the chassis (and not sharing common impedance paths as they should be in a good ground scheme), then making the circuit ground to chassis connection towards the input, then connecting the AC safety ground to the chassis (normally near the AC input) so there isn't anything significant returning through the chassis and it just basically acts as a shield.

[Chuck H]

Thanks all. I agree with just about everything posted so far. And that is how I've always done it. Special thanks to Dai, who brought up differentiating the impedance between ground points to keep them seperate. Because this is at the heart of my confusion. Which ground node should have the lowest impedance to the earth ground? The input or the PSU?

And special thanks to Enzo for bringing up keeping any heavy currents at the PSU node away from the input. This has always been my thinking. But the ideology touted here for "star" grounding (which I never do by the book) would indicate a single ground point terminating at the input. I mentioned here, once upon a time, that I always ground the earth at the PSU node. I reasoned that any heavy currents should have the lowest impedance to earth. And that with the classic star ground any heavy currents in the PSU need to move through the input to get to the earth ground. And that's one reason all these little homebrew Champs hum. R.G. "corrected" me and said (not a quote, so I may have misinterpreted) that the input impedance to earth should be lowest so that it is more resistant to potential changes at the ground node. After that I didn't know what to believe. Especially since my own practices are based in perception and not theory.

[Steve Conner]

Douglas Self recommends connecting the mains ground near to the input jack. The reason is that, when the equipment is used in a larger system, ground loop currents flow between the mains ground and the input jack, so you avoid these currents flowing through the equipment's internal ground system.

This philosophy needn't conflict with electrical safety. Connect the mains ground to the chassis, use a metal input jack bolted to the chassis, and ground your preamp ground bus to the input jack only. Ground loop currents will flow through the chassis between mains ground and input jack ground, but they won't get into the preamp ground bus.

You don't need to ground everything to the input jack. Once the signal gets big enough, you can pull a "change of reference". I like to do it between the PI and the power tube grids: not only is the signal huge here, but it's balanced, so noise induced by the change of reference will cancel. What more could you want?

Things get really fun when you have a whole bunch of inputs and outputs all connecting to the outside world.

[Chuck H]

Douglas Self recommends connecting the mains ground near to the input jack. The reason is that, when the equipment is used in a larger system, ground loop currents flow between the mains ground and the input jack, so you avoid these currents flowing through the equipment's internal ground system.

So that's pretty much the same thing R.G. told me. I won't finish assembly of this amp until sometime around New Year. But I'm definitely going to try it both ways. And I'm gratified that my question didn't end up being a bonehead noob issue.

[Steve Conner]

You can tie yourself in knots endlessly with grounding questions. The bottom line is that the ground system will always be imperfect in some way, but with practical experience you'll get a feel for which imperfections you can get away with. For example, the old Fender ground schemes were a reasonable compromise between signal integrity and cost of parts and labour.

I'd say that if you ground the cathode of your first stage to the input jack, and keep the charging current from the rectifier to the first filter cap separate from everything else, the rest is just icing (frosting? ) on the cake.

[J M Fahey]

Yes, it's basically that.
Personally, I have simplified my grounding a lot by considering each individual gain stage a system in itself, and grounding its input and its feedback to a common point, so any hum there is dealt with differentially.
Solves *most* of the problem and simplifies later building a "chain" of ground reference points.
I've been making amps for 43 years now, before star grounding became fashionable, and never had big problems.
Star grounding may give you a false sense of security; I prefer to deal pair by pair, and, of course, use a separate path for dirty grounds, which mainly are the big filter caps returns, and speaker returns.
So far, so good.

PS; I design mainly with Op Amps, where the system I use is a natural, but same can be applied to discrete transistors, Fets or Tubes.

[R.G.]

This may violate some politically driven safety laws but who really listens to those guys anyway, they know nothing about good sound!

The problem is, lawyers and judges listen to them. Worse yet, juries listen to lawyers talk about those laws.

It is supposed to be cool to think that any regulation you don't happen to like was "politically motivated", and in fact over-regulation and over-criminalization are big problems.

However, what's really being violated are safety standards. You don't want to be in the position of sitting in a courtroom explaining to the widow's attorney why ignoring safety standards was reckless endangerment by not knowing the standards and not willful endangerment by knowing that standards exist and violating them anyway. That's very, very expensive, and if the local DA is looking for a campaign issue might be construed as criminal. Yes, district attorneys probably don't know anything about good sound either, but that only makes it worse.

Using diodes and resistors to lift grounds can cure the symptoms of other problems without curing the actual condition. It's a trick that may be useful as a last resort, but why go to the extreme of making an amp arguably (by lawyers, yet!) less safe if you could have fixed the issue by moving some wires around?

@ the rest of the audience:
The essence of grounding is to know what currents flow in what current path and wires, and arranging that the currents flow without upsetting what should be happening to the signal. J.M.'s process of considering each stage as its own amplifier circuit is one way of knowing what currents are in what wires, and a good one. Star grounding is not the be-all and end-all of good grounding, as it forces the elimination of cancelling currents where those might help. It's just the only way I've ever seen where slavishly doing this.then.this.then.this.then.this.then.this.then.this.t hen.this.then.this.then.this results in low ground noise.

That is, it's difficult, nitpicky, hard to run the wires, frustrating, and arduous - but it avoids the necessity to have to think long enough to know what currents flow in what wires. And thinking is always the most difficult work, right?

[Steve Conner]

I like good sound as much as the next guy, but I'm not prepared to die for it! I use balanced inputs and outputs on all my DIY hi-fi gear.

Sowhat's bridge rectifier trick is used commercially, so I can only assume that it doesn't violate any safety laws. It depends on your defintion of "exposed metalwork", I guess. To me, the strings of the guitar you plug in are part of the "exposed metalwork".

[R.G.]

Guitar strings are indeed "accessible metal" under the terms of safety standards.

It's a Real Good idea to give yourself some distance from hard-metal ground inside your guitar. The wire that attached the bridge to the ground lug of the output jack can be isolated by a 100K-220K resistor paralleled with a 1KV ceramic disk. This adds enough conductivity at high frequencies to keep buzz from echoing down to hum frequencies, and isolates enough that the strings and bridge are no longer going to conduct enough current at power line frequencies to kill you in many cases. It's a protection against good grounds on YOUR amp killing you when you grab some other amp/guitar/microphone with line voltage on its "ground".

In combination with a good cavity shielding job, this is hum free for normal situations. You'll probably still feel a power-line buzz in bad situations, but won't die from it the way a solid hold on grounded strings could do.

Sowhat's bridge rectifier trick is used commercially, so I can only assume that it doesn't violate any safety laws. It depends on your defintion of "exposed metalwork", I guess.

Yeah, well. Also the glasses you used in reading the "standard". The privatization of safety work has produced some odd results. I know this from experience.

Depending on the standard you're using, as determined by the classification and use of the equipment, the resistance between any accessible metal and the earth ground wire must be less than 0.1 ohm or 0.2 ohm, perhaps others in standards I haven't seen. One of the standards provides for testing this resistance by applying a source of 12VAC at up to 25A (!) or 1.5 times the current capacity, which I believe to be but am not certain of being the primary fuse rating, then measuring the voltage across the connection. Two diodes is 1.4V, so you'd have a resistance of 1.4V/test current.

1.4V/25A = 0.056 ohm, a clear pass. 1.4V/14A = 0.1 ohm, a marginal pass. If the test is made at 1.5 times the rated current on a 5A rated device, then R = 1.4/7.5 = 0.187 ohms, a fail if the standard applied is 0.1 ohms. If the test is 1.5 times 4A, resistance is 1.4/6 = 0.233 ohms, a fail.

The thing is, the testing lab decides how this should be interpreted. The situation I quoted is indeed perverse, in that it doesn't make sense to fail my equipment when it clearly meets the intent (i.e. low voltages on the user in the case of an AC fault) but fails one limited reading of the standard under some strained logic. Safety labs operate in the mode not of saying "This equipment is safe.", but instead of certifying "We could not find any way that it doesn't meet our reading of the standard. Might be safe, might not, but we couldn't find it if it's not."

And since a "fail" produces another test pass in most cases, these being $2k to $20k passes, there is a strong underlying push to be REAL SURE before you award a "can't prove it didn't pass" certificate.

Oh, yeah - AND a corresponding push among designers to keep track of which labs are ... um, most helpful ... with advice about how to make this a one-pass certification and not take a whole lot of market time to do it.

[Chuck H]

Guitar strings are indeed "accessible metal" under the terms of safety standards.

It's a Real Good idea to give yourself some distance from hard-metal ground inside your guitar. The wire that attached the bridge to the ground lug of the output jack can be isolated by a 100K-220K resistor paralleled with a 1KV ceramic disk. This adds enough conductivity at high frequencies to keep buzz from echoing down to hum frequencies, and isolates enough that the strings and bridge are no longer going to conduct enough current at power line frequencies to kill you in many cases. It's a protection against good grounds on YOUR amp killing you when you grab some other amp/guitar/microphone with line voltage on its "ground".

And for anyone sneering at this precaution, I assure them that it's all too common. I've worked a lot of dive bars and clubs. The sort of gigs where you move your gear to the stage through an alley. Stepping over sleeping bums and broken Thunderbird bottles. These places, as often as not IME, might never have been serviced by a qualified electrician. Differing poles at the plugs and omitted ground connections are the norm. Back when I was gigging I learned to be very careful and always used a spit guard on the mic. You figure out things like "first I must turn down my guitar so that I can remove my hands from the strings. Only then can I adjust my mic."