Marty Watt asks why putting his 38S into a case caused instability.
The answer is mostly "ground loops". Marty found that if he attached
the board to the case with a metal standoff (close to C29, which
is also grounded) his rig put out an unstable signal.
So the question arises "if its already connected to ground,
why should connnecting it to ANOTHER ground cause problems?"
And the answer is: At R.F. frequencies "ground" is not always what
you think. We always think of ground as zero volts, zero signal, and
zero impedance. In practice, this is very difficult to achieve....
printed circuit traces exhibit inductance, and resistance too - even
the traces that are supposed to be ground traces. So the result is
that ground in one part of your rig isn't the same as ground in
another part of your rig.
Now this usually isn't a problem when you're dealing with a
receiver, where circuit currents are low, and circuit impedances are
high. Those small ground impedances are quite small in comparison.
But in a transmitter, R.F. currents can be large, and circuit
impedances are low, so that those same ground impedances now become
much more important.
So what can you do about it?
Its difficult to do much if there are significant ground impedances
on the printed circuit board. But its very easy to add extra paths
for R.F. current to flow when you start adding controls, switches
and connectors (many of which require that at least one lead go to
ground). Some of these extra paths can divert R.F. current to
low-level parts of the transmitter, causing feedback - and instability.
There are too many possibilites to consider, but some general
principles might help (I'll probably miss a few):
(1) Isolate the high level amplifier ground - only allow ground
currents to flow through one path - the path that the designer
intended (thru the printed-circuit-board ground traces).
As Marty has found, this means isolating the board from the
case with insulated standoffs...play it safe and isolate all four
corners...maybe only necessary to isolate that one corner where the
output amp sits.
Another potential bad ground loop can be eliminated by
using an isolated BNC connector for the antenna. A short run of
coax cable grounds the connector to the board.
(2) Single point grounding - the case SHOULD be grounded, but ONLY
AT ONE POINT. Often, you don't have a choice here: the 12v power
connector shell might be non-isolated - that's your single-point-
ground. You'd be wise to isolate everything else.
(3) Consider a plastic case - since it may be difficult (or
impossible) to isolate everything from ground if you're mounting
to a metal case, a plastic case might be the answer. If your rig
worked fine while sitting naked and scrambled on the bench, but
is unstable in a metal case, chances are that a plastic case will
allow stable operation. However, plastic cases can have problems
too: ungrounded metal shells of pots should be grounded, lest
they act as a small antenna, and relay R.F. to the pots' innards.
(4) bring in power thru an isolated jack. A nasty ground loop can
possibly form if the power-jack negative lead is connected to the
(metal) case. Try to use a fully insulated connector here.
[
QRP-L Archive |
]
[
1993 |
1994 |
1995 |
1996 |
1997 |
1998 |
1999 |
2000
]