[sdiy] ground & power on pad-per-hole solder breadboard

[jh.]

>     Grounding is almost more of an art than it is science.  At the very
> least, you are headed in the correct direction.
>
>     Creating a star connection on a PCB can be a big problem, because it
> takes up lots of space.  In general, you only need a star when you a
passing
> "significant" amounts of current in the traces.  The deffinition of
> "significant" is difficult to pin down, because, it depends on the
> situation.
>
>     Example of insignificant current: Current flowing into or out of an
> op-amp input.  Even in bipolar parts, this current is down in the nano-amp
> range.  These currents will not cause any significant voltage drop.
>
>     Example of significant current: Current flowing  out of the ground pin
> on a 555.  Even worse, it is not constant.  Another example of significant
> current is the current flowing out of the ground pin of most logic chips.
>
>     The main problem that drops in your ground may cause is when there is
> something that can use that voltage as it's input.  Sometimes finding out
> how these signals are being amplified by the circuit can be difficult.
>
>     Also, another aspect to all this is proper bypassing, to get rid of
the
> voltage spikes.  To properly bypass the current on a chip, you need to
> provide the shortest path possible between the bypass capacitor and the
chip
> being bypassed.  This, for the most part, can be difficult to do because
the
> power supply pins are generally positioned to make it difficult to put a
> capacitor directly accross the pins.
>
>     Anyway, this is a very big subject.  I am sure others will add their
> thoughts to this as well.  People who specialize in EMC compliance (and, I
> seem to recall we have a few on the list), would know a lot more about
this
> subject than I do.
>
>     -Jim

I fully agree with Jim.

Btw, star grounding taken to the very extreme (which no one does) would
lead to a certain point where inductive coupling actually becomes worse than
resistive coupling of a not-so-star-grounded system. (Or so they say in
EMC literature).

In practice, I don't use much star grounding at all *except* for mixed
signal
applications. (If you have digital and analogue circuits, separating the
GNDs is a *must*.)
I avoid large currents into GND - connect switched LEDs between + and -
supply rails, for instance. If you need GND reference for a LED, use
an opamp to create a "dirty GND" locally. (The opamp will take its current
from the rails even when its output is fixed and 0V.)
Thake care that the GND reference from one gain stage to the other is
a good one : The GND at the positive opamp input of an inverting amplifier
must be "close" to the GND reference of your source or previous stage
(more important than being "close" to any star GND of the whole system).

A good rule is to be aware that there is nothing such as "one GND node"
in the circuit at all - you're always amplifying a voltage difference
between
your two amp inputs - even if one of them is connected to a "GND" trace.
Make sure that nothing can interfere with that GND trace *between* two
gain stages - much worse than a slight voltage that's common to both ends
of that GND trace. (That's the reason why in a chain of amps the final
- powerful - one had to be closest to the PSU in the old days without
split supplies. Still a good rule today ...)

JH.