Neovibe - a question about the GGG board

[yeeshkul]

Guys i know this is well known and tested Neovibe board, made by R.G. himself if i am not mistaken. That's why my question is probably silly but: can the ground connected into a loop around the circuit cause some of the ground-loop troubles?



i haven't run into any troubles, just asking ....

[R.G.]

Guys i know this is well known and tested Neovibe board, made by R.G. himself if i am not mistaken. That's why my question is probably silly but: can the ground connected into a loop around the circuit cause some of the ground-loop troubles?

Well, first of all, after hundreds if not thousands of those boards being assembled with no problems - what ground loop problems?

The real proof of any board is how well it performs. And this one does perform well; so there are no "the ground-loop troubles" with this board that I know of.

But that's not what you asked. You asked about the theory. There are really two grounding systems on that board. One is a loop around the signal section of the board, and the other is a no-fooling star ground setup on the PCB itself. Both are obvious on the diagram you posted - which, by the way, would you take down when we're finished with this, so I can keep some kind of control if the board ever needs updated?

The star is in the lower left. Notice that
(a) the incoming rectifier (-) wire, which is NOT a ground wire, is a single, separate wire leading to the (-) side of the first filter cap. This keeps rectifier pulses off the signal and power ground nets entirely. The rectifier pulses travel on the single wire from the rectifiers to the first filter cap.
(b) the (-) wire from the first filter cap to the second filter cap still contains some rectifier pulse content, but it is separated from the signal wiring again. Only after the second filter cap does ground go to the circuitry. You'll see a common thread here - what matters on ground wires is WHAT CURRENTS TRAVEL ON THE WIRE?
(c ) the (-) wire from the second filter cap, now separated from the noisy (-) wires in the rectifier circuits connects to the star point. The star point has branches going to (i) the LFO/bulb circuit, (ii) the three terminal regulator, (iii) chassis ground, if wired separately from the jacks (iv) the signal path loop.
(d) you're worried about the signal path loop; it is, after all, a violation of my often repeated advice about not forming grounds into loops, and not mixing reference grounds with sewer grounds. A lot of thought went into that.
(di) It's a loop because a copper loop can shield the inside of the loop from exterior electromagnetic pickup if the wavelength of the radiation is more than four times the largest dimension of the loop. Since the wavelength of 60Hz radiation is L = 3E8/60 = 5E6 meters (five MILLION meters), then the loop looks pretty much like solid copper to 60Hz radiation, and also to all the lower harmonics of power-line frequency. It rejects power line and rectifier induced hum and shields the circuit. I judged this was important in a pedal with an AC power transformer in the box.
(dii)The loop does mix input reference ground with signal path sewer ground. There is the possibility of this causing parasitic oscillation in the circuit, if other conditions are met. I think this may be the mystic hoodoo grounding issue that JC was alluding to but refused to support and admitted by his silence that he could not support when I challenged him to do so. Three factors keep this from being an issue.
(dii1) the loop gain is low, less than unity; this is a common complaint about the unmodified univibe
(dii2) the input impedance is low, under 100K; again, a complaint in some quarters about the unvarnished univibe; and
(dii3) the currents are low and vary in phase from stage to stage, so there is partial cancellation of the current-induced voltage.

You'll notice that the input is as far away from the AC power inlet as it is possible to get on the PCB. That was not an accident. I considered and rejected making the board larger in order to star ground the signal path in addition to using the ground loop as a shield, and actually built a version of that. I could tell no difference in the sound from the samples, and this boosted the size of the PCB to bigger than what would fit inside a wah pedal housing. Since many effects pedals have far worse grounding issues, I bagged it and left it as you see it.

Questions? What have I confused you about?

[yeeshkul]

R.G thank you for your time and the big explanatory answer - i was really just curious as i did not notice the loop before. Everything is clear but one thing:
"The loop does mix input reference ground with signal path sewer ground. " - what is difference between input reference ground and signal path sewer ground, i can't find translation for sewer ground.

Jan

[R.G.]

My apologies - you're reading this through a translator program. Let me explain.

Ideally, every place where there is a "ground" symbol in a schematic would connect to a perfect source of 0.000000V through a zero-resistance wire.

That's not possible with real parts, since all wires have some small resistance, and all "grounds" have some resistance as well. So we make distinctions between all the things that we think that "ground" (the concept) does. These are:
1. A reference voltage, a perfect source of 0.000V so that things like inputs can have one quiet place to refer to.
2. A shield covering or protecting some delicate parts of the circuit from outside interference and draining that interference energy to someplace where it won't interfere with the circuit.
3. A return-pipe for taking the electricity which the circuit has already used back to the power supply so the power can flow in a complete loop. "Sewer ground" is my own made-up term for this version of the concept of "ground", making an analogy between the return pipes for used (and dirty) domestic water and the return wires for used electricity.

The important thing to remember about ground is that you should be aware of what currents flow in which ones of the ground wires. One of the classical mistakes in power amplifiers, for instance, is having one ground wire for the speaker return to the power supply, and connecting the input ground to that same return wire. The heavy speaker currents on this "sewer-ground" wire cause voltages to appear on the wire because of the wire's resistance. If the input ground wire, which is a reference ground, is connected to the speaker end of the speaker "ground" wire, the input reference then adds the speaker ground wire's voltage to the input. If there is enough gain, the amplifier now oscillates. If it does not oscillate, it has funny response artifacts or uneven frequency response.

This is an extreme example because of the large currents on the speaker output. But in our effects, we use huge gains in distortion pedals. That means that the voltages on the "sewer-ground" return conductor can be much smaller and still cause problems. In a distortion pedal, I might decide to change the grounding scheme so that I did have an independent input ground. But the univibe is low gain, so it works.

[yeeshkul]

I have to admit that my head-translator has certain vocabulary limits and most of the made-up words are beyond those. I however should have realised what the "sewer ground" means.

Again, thanks for taking your time and putting some light on these basic things. I understand now  

[Eb7+9]

My apologies - you're reading this through a translator program. Let me explain.
In a distortion pedal, I might decide to change the grounding scheme so that I did have an independent input ground. But the univibe is low gain, so it works.

love it ... an independent input ground

[R.G.]

Yeah, pretty cool, huh? 

[yeeshkul]

Just making sure - can the (above described) oscillation happen(meeting other conditions) when i connect the output of a high gain circuit (Fuzz Face) to another circuit that has a low input impedance (another Fuzz Face, speaker ...)? Is there any schematics around showing the separate grounds(return-pipe separately from the reference ground)? Is that done by a transformer? Can't imagine that.

[R.G.]

Just making sure - can the (above described) oscillation happen(meeting other conditions) when i connect the output of a high gain circuit (Fuzz Face) to another circuit that has a low input impedance (another Fuzz Face, speaker ...)? Is there any schematics around showing the separate grounds(return-pipe separately from the reference ground)? Is that done by a transformer? Can't imagine that.

I can't quite parse out the questions, so if the answers mystify you, ask again.

A. Can you get oscillation because of the interactions that happen when connecting a high gain circuit to a low impedance circuit ?
Not just from the inputs and outputs. If they have a common power supply, possibly. The way to prevent this is to keep the AC signal components which could cause oscillation from flowing through any common wires. You do this with decoupling networks inside each pedal. A 100uF or even 1000uF capacitor across the +9V inside each pedal usually stops interactions through the external power lines. It does this because the capacitor selectively shunts the AC signal components on the power supply and ground to one another, where there is usually some cancellation; in any case, it keeps the power supply noise inside the individual pedal.

B. Schematics showing the different grounds?
Not that I know of. If you'll mark one up, I'll review it and help you get the idea. Grounds that go to inputs are generally reference grounds. Grounds that go to the emitters of bipolars, sources of FETS, and V- supplies of opamps are generally sewer/return grounds. Grounds for tone controls are reference grounds.Grounds for clipping diodes are problematic. You want them to be reference ground so only the diode's voltage is transmitted to the rest of the circuit. However, the diode currents are large, and could pollute other low-current reference grounds. A pair of clipping diodes really should have a separate ground back to the star ground.

The thing is, inside a single pedal, the ground wires are so short that in many cases you can get away with ignoring grounds and just connecting up the nearest ground wire. In pedals with high current loads and especially digital switdhing, grounding gets more critical. In pedals with JFET and MOSFET  inputs, the impedances on the gate are so high that grounding gets more critical. For a Fuzz Face, you almost have to work at it to get it wrong.

[yeeshkul]

R.G. i am back with one more question: does the term a "separate ground" mean the ground conductor (wire, PCB line) that goes straight to the star and by the other words it is not connected to the sewer part any other way but through the star point? Geeeee, I was just tangled in missleading imaginations of double-ground circuits . Shame on me 

[R.G.]

... does the term a "separate ground" mean the ground conductor (wire, PCB line) that goes straight to the star and by the other words it is not connected to the sewer part any other way but through the star point?

If you mean the place where I said:

A pair of clipping diodes really should have a separate ground back to the star ground.

then yes. That would have been clearer if I'd said "separate ground trace" or "... path". US English usage gets imprecise easily. Especially -my- use of the language.