(2) dirty/clean ground again [sdiy]

Neil Johnson nej22 at hermes.cam.ac.uk
Sun Aug 18 21:48:31 CEST 2002

Just to jump into this thread...

Decoupling is a typical engineering issue---there is no one magical
solution that fits all problems.

Firstly, the situation.  Moving *any* current about a circuit, be it on a
PCB trace or down a wire, produces a voltage differential across the
conducting medium (unless you happen to be using superconductors with
zero impedance :-).  Copper has a finite, if small, resistivity, and its
that which generates the voltage difference.

Supply rails are not perfect.  The only part of a voltage regulator that
is regulated is the output terminal (or point where the sense line is
connected).  Anywhere else along the supply rail will be dependent on (a)
the output voltage of the regulator, and (b) the instantaneous current
being drawn along the wire.  So regulating the ground would not solve the
problem, only add to the complexity and impedance of the ground circuit
(not good).

Now, look at op-amp circuits.  They draw (or dump) current through their
supply pins *only*.  Any ground connected to an op-amp is used purely as a
reference voltage.

I personally don't like decoupling op-amp rails to ground, because
basically all op-amp current flows through the supply rails, nowhere near
ground.  Indeed, placing two caps across an opamp, via ground, actually
halves the rail-rail decoupling capacitance:

	+V ------||-----*-----||------ -V
                 C1     |     C2

Now, any signal flowing from +V to -V sees C1 and C2 in series, which now
looks like half of C1 (or C2, assuming they're the same value).  So
although you've put two expensive 100n monolithic ceramics next to your
op-amp, it actually looks like 50n across the rails.  Mmm....

And the large electrolytic down the end of the supply line isn't there for
fast pulses due to the impedance of the track.

Also, unless the +V and -V rails are carefully balanced, you'll pass some
current down the ground wire, introducing a voltage spike on ground.  BAD.
Unless its for safety (in which case you have a separate circuit) ground
should only be used for reference, not carrying current.  That's what the
supply rails are for.

Obviously, in the case of single-rail stuff you have to be more careful,
but the same principles apply, and its usually good practice to have two
grounds...one for supply, and one cleaner one for signal references.

Of course, we also want to reduce the amount of crap on the supply rails,
because op-amps only have so much PSRR, and we don't want to screw
ourselves.  Hence we put hefty local decoupling (10-100uF) to keep the
supply noise down.

That resistor/capacitor trick does two things: it filters out any crap
coming in from other op-amps, and also reduces any local crap getting out
and affecting anyone else.

> I'm under the impression that using "extra" decoupling caps can't hurt.
> For a one-time project, the additional cost seems negligible.

>From a cost perspective, it could.  From a circuit perspective, it
shouldn't normally, unless you have so much capacitance that the voltage
regulator is unable to cope with the capacitive load.  There's also the
issue that placing them further away from their respective device reduces
their effectiveness.

Ideally, a multilayer board with a single, massive ground plane is heaven,
and then it doesn't matter so much where the decoupling caps are located
as there's such a low impedance from any supply pin to a decoupling cap.

Then again, in HF/RF PCBs I've used multiple ground planes (one for the
micro, one for the RF generator, one for the sensitive receiver) which
only connected at the board's power connector.

> So the trouble is that ground is not an infinitely large pipe -- too much
> water too fast and it will back up.

Sadly, welcome to the real world.

> So how can we solve this??  Can ground be regulated??  I've never seen a
> 7800 :)  However, a voltage  can be regulated, and if that regulated
> voltage is held on one side of a cap, wouldn't that locally "re-zero" the
> ground on the other side??  Perhaps this goes back to JH's point b) above,
> by stabilizing the local ground of a single-sided amp, you stabilize the
> amp.

But, see above.  The local ground used by an op-amp is only there for
reference purposes, not for carrying any current.  For instance, in a
differential amp, both op-amp inputs are connected to signals.  And
indeed, there is no need to use ground at all if that's what you need.

> Going back to what Roman was saying about ground being zero, even at
> equilibrium without any transient currents, if PCB traces have resistance
> and curent is flowing through them, then there has to be a voltage
> difference between one end of a ground trace than the other, right??  If
> there is an I across R then V can't be zero at both ends.  Either that, or
> I'm totally screwed and I need another hobby.

Correct--"ground" is never, ever zero!  Even the real, brown, muddy ground
is not a perfect ground (read the recent thread on lightning strikes and
ground surges).

For very sensitive circuits, ground is hardly used at all, as its way too
noisy.  Check out chapter 7 in Art of Electronics for more insights.


Neil Johnson :: Computer Laboratory :: University of Cambridge ::
http://www.njohnson.co.uk          http://www.cl.cam.ac.uk/~nej22
----  IEE Cambridge Branch: http://www.iee-cambridge.org.uk  ----

More information about the Synth-diy mailing list