Grounding. Simple questions...or not?

[brett]

Hi.
I've built many stompboxes, amps and things, staying with tried-and-true practices.
The results have been mostly good.
But my current build has me thinking (overthinking!) about practices and theory.
Q1.  If we like to avoifd ground loops, why on most pcbs do we have a ground next for the input signal that goes through the board to a ground for the output signal?  It's a loop, and the opposite of "star" grounding that seems best practice. 
Q2.  When using a shielded cable (such as for guitar socket input) is it better to ground to the pcb ground or to the star ground?  Although the pcb ground seems to be what's used (or at least accomodated) it must follow a trace and the power supply ground to get to the 'true' ground (usually at the power supply).  Especially at the input, that's asking for extra resistance and noise, to my way of thinking.
Q3.  An esoteric one, but I'll ask because someone will know.  Any effect or amp with an output transformer has a primary that's got a signal across it, usually tied to ground at one end and "live" at the other.  The secondary is likewise tied to ground at one end and "live" at the other.  For an amplifier output transformer, the ground is also the speaker -ve and the "live" is the speaker +ve.  My question is - what purpose does the ground serve on the secondary?  Would it make any difference if it wasn't there? (Isn't a current pushed back and forward to the speaker regardless of a ground?)
Q4.  In the case of an output transformer, with one end of the secondary grounded via the board with the OT on it (and back to the power supply) ...is it a good idea or a bad idea to take the speaker -ve to the power supply or star ground too?  It creates a loop, but surely it's a lower restance to ground than going via the board.
These seem such simple things, but in practice they get kinda complicated.
Yes, in my current (pun) build, I'm using fairly heavy wire for grounds, but the pcb tracks are only about 3mm wide.  For convenience, I'm using the tone pot and steel chassis to ground the tone control section.

Thanks for any and all help.

[GibsonGM]

1. Probably limitations of space on the PCB, and also the fact that at the voltages we're using, any noise currents that might be floating on the ground on the PCB are very tiny, not a big factor noise-wise. So, they get away with it, and it's a matter of practicality.

2. By grounding one end of a shield, we're just extending the enclosure 'around the signal wire'.  The most direct connection to chassis seems to be the best practice, if one could do it....back to the battery ground in a stomp enclosure at the input jack seems logical...often I connect it along with the many other grounds already at the input jack.    In an amp, I always have a heavy copper wire ground bus running along the front, and I ground it to that; it connects to the chassis right there 1" away usually.

3.  Can you post a pic of the OT question?  I'd make an assumption that primary and secondary aren't electrically connected, so the gnd on the secondary may be just a reference point for the output side.  EDIT: OR NOT REALLY CORRECT (See Merlin's comment below)

4. the shortest, most direct point back to chassis ("main star point") is generally the best way to go, IMO and experience.  Can't see how your tone pot works into things, but if 'tone' is a ground and the chassis it's sitting in also is (and they're connected), that may not be the best way to go.   Of course, it may work fine, too  

HTH Brett.

[merlinb]

Q1.  If we like to avoifd ground loops, why on most pcbs do we have a ground next for the input signal that goes through the board to a ground for the output signal?  It's a loop, and the opposite of "star" grounding that seems best practice. 

Our PCBs are quite small, so if you squint the whole PCB ground plane is basically one big(ish) star. We're not handling megahertz signals.

Q3.  The secondary is likewise tied to ground at one end and "live" at the other. My question is - what purpose does the ground serve on the secondary?  Would it make any difference if it wasn't there?

It shouldn't be there at all. Transformer secondary should go straight to the output jack with no ground connection at all. But a lot of homebrew designers are locked into a 'signal plus ground' way of thinking out of habit, so you sometimes see it done wrong, I made the same mistake as a beginner.

[amptramp]

Again for Question 3: Designers do not want any possibility of an insulation breakdown that could put B+ voltage on the speaker leads.  Better to have a ground connection, particularly if the amp goes to a separate speaker through external cables.  The cables could go up to the B+ voltage in that case and you don't need a direct short - even a few megohms of resistance could put hundreds of volts on the speaker leads.  If you have a speaker in the same enclosure as the amp, it is not necessary.

One side of the speaker is also usually grounded to allow negative feedback to come from the speaker connection itself.  The negative feedback stabilizes the gain and reduces harmonic distortion.

[Dormammu]

It shouldn't be there at all. Transformer secondary should go straight to the output jack with no ground connection at all. But a lot of homebrew designers are locked into a 'signal plus ground' way of thinking out of habit, so you sometimes see it done wrong, I made the same mistake as a beginner.

If you get with  ungrounded amp in the rain or "successfully" thrown beer - you risk your health and life from electric shock. Especially with tube amp.

[brett]

Wow.  Such good and helpful answers.

The "protection against a live speaker cable" answer makes great sense.  Not something I'd have EVER realised on my own. 

Re: tone control.  I'm using the body of the tone pot (500kB) as a local ground, as it's screwed into the chassis (my star ground is bolted onto it a few cm away). Should be trouble-free.

Thanks so much.  I'm feeling happier about my my methods now.

[GibsonGM]

I'm not saying it's a true "no no", Brett, but using a pot body as a ground sort of violates the 'good grounding code', lol.   If and when it rotates loose or the connection corrodes, it's NO ground time.  But, many old amps did this successfully since amps were invented.   If ever there's a problem, just run a pigtail to that chassis ground...your call.

[brett]

I hear what you're saying Gibson GM.  Not a good idea for a regular amp*

The tone control is from the BMP, built on the pot, as per beavisaudio.com.



*this isn't much like a regular amp.  This is a loungeroom amp that won't move from the top of the speaker cab.  It's a 0.4W 'beast' 😄.  Hence my desire for very quiet but no need to be rugged.  The 12AX7 preamp has gain ~60, so there'll be a big signal relative to ground noise in the tone section.  Up to 80V!

[GibsonGM]

I don't think you'll have any problems, Brett...EVEN if you're BREAKING THE RULES, lol.   That was standard for decades..."proper grounding" says no, but as long as the connection thru pot remains tight, it won't do anything. 

When you get done with it, I'd be interested in seeing what you built (schematic).  I like low watt amps too, for home recording...can't wind 'em up the way we used to these days (at 51 yrs old).  No place to isolate the amp anyway, so it's next to you! LOL    Many of them sound really good, and don't need too many bells or whistles to do so.