On 12/19/2013 11:50 AM, Ed Douglass wrote:
I would like Jim Brown to respond: does running fat copper from each piece of
equipment to a bus bar (as Peter Laws says he is doing) comply with your
direction to run fat copper between each piece of equipment and the one next to
it? (Bonding links become too long at RF? Creating loops which have
inductance and in effect, become receiving antennas? Creating multiple DC paths
between components because of cable shields?)
All three issues are at play -- minimizing hum and buzz requires very
low resistance. Longer conductors begin acting as antennas.
The most common root cause of RFI is "The Pin One Problem," so named
because it was first discovered by audio professionals on the mic inputs
of equipment. Pin One of the XL-connectors used for balanced audio is
the shield contact. The only proper connection for a cable shield is to
the shielding enclosure. A Pin One Problem exists when the cable shield
goes somewhere else. The most common example of a Pin One Problem is a
connector mounted to a circuit board, with the shield contact going to
circuit common rather than to the chassis. When shield current is
present, it wanders around signal common until eventually it finds a
return path, usually the power system green wire, or to some other wire
that takes the chassis to "ground." As that shield current wanders
around, it creates IR (or IZ) drops at random points around circuit
common, and that gets added to the signal. If the shield current is hum
and buzz, it's added directly. If the shield current is RF, it is
detected by the input of every gain stage that sees the IR or IZ drop,
and then amplified.
When we bond chassis to chassis, we are, in effect, shunting current
that could be on the cable shield to the bonding conductor instead, so
it never gets inside the box.
There's an additional coupling method when unbalanced wiring is used.
Here, whatever voltage difference exists between one chassis and another
is added in series with the signal. A common source of voltage
difference is leakage current flowing on the green wire. Most of this
leakage current consists of triplen harmonics (3rd, 6th, 9th. etc.) of
the power frequency, and is heard as buzz. Bonding gear from chassis to
chassis shunts this current away from signal circuits, and to be an
effective shunt, the resistance of the shunt must be much lower than the
resistance of the cable shield. Thus, short, fat is critical. The good
part here is that chassis to chassis bonding is a simple and effective
solution to both mechanisms, and is entirely consistent with good
engineering practice for lightning protection.
For more about the Pin One Problem, see k9yc.com\publish.htm
Interconnecting cables that excite Pin One Problems at RF should always
be as short as practical, but also benefit from an effective ferrite
choke. The choke must have enough resistive impedance at the frequency
of interest to significantly reduce the RF current. That means multiple
turns (at least 4-5 if the frequency of interest is HF.
73, Jim K9YC
RFI mailing list