Basic opamp based preamp

[Mooly]

Something like this. Your ground connection/s to the amps goes to the star.

Try and imagine and think of each wire and line of print as a resistance... and what will happen if a current (AC... but it's useful to think in DC terms too, to make it easy to follow ) flows along it. Ask yourself "will it influence any other voltage or superimpose itself and interfere with any other part of the circuit"
The good news is that yours is such a low powered circuit it's not really going to be a major issue anyway... but it's nice to get it right.

[Stormrider]

This seems completely unnecessary. Are individual ground runs really needed for each regulator? And one for each 100uF filter cap? Is this the same thing you were talking about, Andrew?

[gootee]

It's definitely not COMPLETELY unnecessary, although some parts are more important than others.

Currents flowing in ground return conductors induce voltages back upstream, at the non-ground ends of the conductors. So, something that should have a quiet ground should not share any length of ground-return conductor with something that is not quiet.

Conductors have at least both resistance and inductance. And while the voltage induced by the resistance is only the current multiplied by the resistance (bad enough; but it gets worse), the voltage induced by the inductance is the rate-of-change of the current multiplied by the inductance. If there are even small-magnitude currents that are changing rapidly, the induced voltage can be relatively large, with relatively-large dynamic fluctuations. And if that voltage appears back at the ground reference point for a regulator, or an opamp or amplifier input's ground reference, the voltage ends up being SUMMED with the input. That's "a BAD thing".

I have even read where people doing point-to-point wiring noticed a difference caused by the order/sequence in which ground lugs were stacked onto a bolt that was the star ground point, because of the added conductor length caused by the stacked thin ground lugs! So they made sure that they put the lug with the noisiest currents on the very bottom, so that those currents would share the shortest-possible path down the bolt with the other ground-return currents.

So, no, it's not completely unnecessary.

Cheers,

Tom

[Mooly]

Originally Posted by Stormrider

This seems completely unnecessary. Are individual ground runs really needed for each regulator? And one for each 100uF filter cap? Is this the same thing you were talking about, Andrew?

If you are aiming for perfection... then this is the way to do it. It's not just being fanatical about this... it's a very real issue and the main reason why DIY amps are not 100% silent hum wise (and a lot of commercial stuff too)
The charging currents in the main reservoir caps can be huge, they can peak in the "several amps" range for a millisecond or two. Now imagine you have a piece of PCB track or wire a few cm long that has only a few milliohms resistance lets say 10milliohms... and the current through those caps peaks at 2 amp. That gives you 20 millivolts of ripple. If you connect an input ground to that print and a feedback return a few mm futher along you have "introduced" a proportion of that ripple into the amp. Result... a harsh hum/buzz.
You can clearly see on a scope this ripple current just by clipping the ground wire to one of the reservoir caps and "sliding" the probe along the caps lead... it's very clear and measurable... and audible if you get it wrong.

In your case, you will achieve 98% of this by making sure you just keep the star ground separate as a spur from the main caps as I indicated earlier. It's all low power stuff... it's all essentially class A, so the current draw is constant. Is 98% good enough ?

When I started building amps getting them silent was the biggest problem... I just didn't grasp this earthing problem... and I was making the same mistakes all the time. When you sit and think, and think in terms of ohms law, and put in nominal resistances for all the conductors, you can then say " if 1 amp flows down there, will it in any way affect the signal"

As you have drawn it here, it's still wrong...

It works

[AndrewT]

Originally Posted by Mooly

As you have drawn it here, it's still wrong...

indeed.
Listen.
Run a wire from the smoothing caps zero volts to the main audio ground.

Never put a string of tappings on the connection between the smoothing caps.

[Corben]

Regarding the ground lift resistor, it's placement seems wrong to me. Placing it between the PSU and audio circuit ground like that introduces a 10 ohm resistance to any circuit taking power from either rail and the ground, which is probably not what you want.

The idea of such a resistor is to prevent ground loops via the protection earth, so it should go between the device ground (the ground shared by the PSU and the audio) and PE. From electrical safety point of view, it's best not to have one and just directly connect to PE. Proper grounding of the equipment ensures that there is no hum due to ground potential differences. But what if the impedance via PE between devices is not much higher than along the interconnects? I would expect this to create crosstalk between (unbalanced) signals via magnetic coupling since PE acts as a signal return path. Also, RF pickup might result. I've never experienced this problem myself, so I can't say how big it is.

[Ted205]

how about isolating the gnd of the reg from the smoothing caps

[Mooly]

Originally Posted by Ted205

how about isolating the gnd of the reg from the smoothing caps

So carry your idea foward
What value would you use ?
Where is the star point ?
If it's to the left of the resistor, then the reference pin of the regs are not accurately defined and with reference to the star would be subject to disturbance as any AC component flowed in the caps of the regs. That would alter both rails relative to the star even if the current draw was only in one reg.

If it's to the right of the star... where would you connect the load too... imagining it were a power amp etc ? The same problems of currents in the ground lead still apply.

Do you see

[Ted205]

rightly or wrongly i assumed the load as being a power amps stage with its own dedicated supply.

in my preamp stage i have used a simple pre-reg from Rod Elliott's site here to drop the supply down to a regulator friendly voltage. I cribbed the resistor from this. Though i realise that this psu has its own transformer and smoothing caps.

Stormrider, I'm interested to know why you have selected a 22uf cap for c11/12

[Mooly]

Hi Ted,
I imagine Rod is "trying to cover all bases" with the PSU in your link... he knows there will be lots of folk building it that don't wire the grounds correctly and that just connect them haphazardly. Ground lift resistors can help in those cases, but it's the basic layout that's wrong in the first place.

The easiest option is always to do it right first time but that has to be designed in to a project as a whole.
Even commercial gear has problems, I have seen PCB's with "options" for different grounding schemes etc laid out on the PCB, and have seen official mods on commercial stuff that involves cutting print and wiring to different points.

It's a very difficult subject to grasp all the complexities of... those three pin regulators... you have to realise where the reference points are (what the output voltage is compared and referenced too) and that "modulating" that pin with ripple in relation to another point (a ground reference) in the circuit degrades the performance.