[Elecraft] K3 RFI in transmitted audio

[Jim Brown]

On Mon, 13 Sep 2010 01:26:43 -0400, Joe Subich, W4TV wrote:

> > Let's say that you bond all equipment to your MAGIC SINGLE POINT,
> > and it takes 3 ft of wire to get there. If two pieces of equipment
> > are only three inches apart and you bond directly between them,
> > CHASSIS to CHASSIS, the wire might be only 6 inches long.

>That's fine with two pieces of equipment that are physically close
>to each other.  However, if one has several pieces stretched along
>a long bench with a power supply at one end, bonding box to box
>could conceivably cause common mode currents to flow through all
>of the boxes causing who knows what kind of abnormal operation -
>and pin 1 problems!

By definition, a common mode current is one that flows on all 
conductors of a SIGNAL cable, usually as a result of a potential 
difference between the equipment at either end, or as a result of 
antenna action. The bonding I've described tends to strongly reduce 
common mode current to very small values, both by diverting it away 
from signal cables, and by reducing the potential between one chassis 
and another to a very small value. Further, I cannot conceive of ANY 
current THROUGH equipment if what is being bonded is the CHASSIS of 
one piece of equipment to the CHASSIS of another!  

> > Providing a low resistance bond between interconnected equipment
> > also puts a band-aid on low frequency pin 1 problems, because most
> > current that flows between equipment takes the lower resistance
> > chassis-to-chassis bonding path rather than the higher resistance
> > path via cable shields and onto the signal return bus.

>But that's only a band aid.  Address the issues by properly 
connecting
>the cable shields to the chassis and bonding the the signal returns
>to the circuit common properly (e.g. removing things like RF chokes
>between signal returns and the chassis).

Often what you suggest is not possible, Joe, because of the way 
equipment is built. As a manufacturer, I'm sure you've run into this 
with your own products. Remember how pin 1 problems are typically 
created -- connectors are mounted to a PC board, the board is stuffed 
into a box, and the connector never touches the chassis. Instead, 
shield current must wander around the PC board to get to the chassis, 
which is required by NEC to be bonded to ground. Sometimes it's 
possible to bond those connectors (I managed to do that with some Ten 
Tec gear and a Time Wave DSP unit), but most equipment is built so 
that you cannot fix pin 1 problems without doing major surgery to the 
gear. 

There is another fundamental principle at work here -- the engineers 
who design a given piece of equipment and do the manufacturing 
engineering to put it in production must verify that the equipment is 
stable in all possible configurations. One of the things that can 
strongly affect that is internal bonding of signal commons (what some 
folks mistakenly call "grounding"). When you CHANGE that by fixing a 
pin 1 problem, you had best be prepared to go through that stability 
checkout. Again, a can of worms best left un-opened. 

Pin 1 problems should be fixed at the DESIGN and MFG stage. When 
encountered in the field, the best approach is a well thought-out 
band-aid. Either choke the current, break the path of the current, or 
divert the current. Ferrite chokes kill the current by inserting a 
high common mode impedance in series, bonding diverts the current away 
from the pin 1 problem, and opening the shield of BALANCED wiring at 
one end breaks the path. None of those measures have are likely to 
affect stability of the equipment, they're all simple, and 
inexpensive. BTW -- these are not only my own judgements, but also 
those of the VERY sharp engineering minds who have written AES EMC 
Standards. 

73, Jim Brown K9YC