[RFI] Request for comments, Rheem HVAC

[Jim Brown]

On 10/3/2021 11:03 AM, David Eckhardt wrote:
> CMCs would on 31 material should be much
> better at suppressing EMI than the included toroidal chokes.

In addition to Dave's excellent advice, I would add this. First, it's 
important to consider the return path. As Henry Ott taught in his EMC 
workshops, the hidden schematic lurking behind the ground symbols -- to 
study where ALL the current is flowing.

By ALL the current, he meant not only the mains or DC in a system, but 
also the RF. I saw a great demo of this at an IEEE workshop in Chicago, 
about 20 years ago when I was giving a talk about EMC in pro audio. The 
setup connected an HP signal generator to a load separated by a large 
table. Signal was carried by RG58, with a large metallic plate wired in 
parallel with the shield, with current probes in the wires to the plate 
and to the shield. As the generator was swept upward from 30 Hz or so, 
where all the current was in the large plate, around 1 kHz it began 
shifting to the cable shield, and by 20 kHz, nearly all of it was in the 
shield.

A bonding and/or return conductor must be run with the current carrying 
conductors, so that it acts like a transmission line to contain magnetic 
fields created by those currents. It is also good engineering practice 
to make each phase twisted pair, to further limit leakage of the fields.

On multi-layer circuit boards, a continuous "ground" layer runs under 
layers of traces, and a transmission line is formed between the trace 
and the "ground" layer, which limits the strength of fields that can 
crosstalk into other circuits. But this falls apart when the "ground" 
layer is interrupted to fit in a trace that the designer forgot -- it's 
no longer a transmission line, so return current flows wherever there's 
a path, like the chassis, creating a magnetic field for crosstalk.

73, Jim K9YC