[RFI] SolarEdge Finds New Source of Solar System RFI

[Jim Brown]

On 6/3/2019 9:21 AM, Hare, Ed W1RFI wrote:
> There are a number of possibilities here.

I got involved in trying to help a Solar Edge applications engineer 
optimize a couple of choke designs to be applied to the loop of 
optimizer outputs on the roof. He provided me with lengths of the cable 
they are required to use (by code). In the process, we had discussions 
at a couple of long lunches (he lives in Oakland, CA, about 70 miles N 
of me), and I made several major points to him.

First, that the most cost effective way to solve these problems is to 
design the electronics to minimize MF and HF content, and to practice 
good EMC practices in mechanical/electrical product design. This would 
include interior layout of everything in the path of potentially noisy 
current, careful shaping of the switching waveform to minimize harmonic 
content (as first Elecraft, and subsequently Flex) have done with CW 
keying), and, of course, suitable filtering.

Second, he described a bonding architecture for the panels in the 
particular installation in question whereby lightning ground and power 
ground were on opposite sides of the building, and said that had been 
dictated by the inspector. This arrangement could be producing both 
magnetic and EM fields. As part of troubleshooting, I suggested that the 
lightning ground temporarily be moved so that it follows the same path 
as the switched DC conductors.

Noise generated by the panels themselves can, of course, be radiated by 
conductors within the panels, and by wiring external to the panels. In 
addition, the system can, if there is a sufficiently large loop, produce 
a magnetic field which can be quite strong in the near field. There are 
also electrical/mechanical designs that can minimize both the magnetic 
and EM fields.

An example of what I'm getting at is careful implementation of 
microstrip techniques, either on circuit boards, or by routing 
conductors in very close proximity to a continuous conducting surface 
(like a metallic enclosure or frame). This causes return current to flow 
in the adjacent conducting surface, forming a transmission line, which 
in turn minimizes both the magnetic and EM fields.

I've not heard from that systems engineer since my choke design was 
applied to the installation in question, but I've heard separately 
(perhaps here?) that they were redesigning some of their electronics.

And, of course, there are multiple potential sources of EMI -- the 
panels, the optimizers, the inverter, and any control circuitry.

> A common-mode choke is only going to be effective if it includes hot, neutral and ground for ac wiring and +, - and ground for dc.

Exactly right. But for purposes of clarity of understanding, I prefer to 
use the word "return" in place of "ground" when talking about DC wiring. 
:)  For the same reason, I don't like the word "balun" -- I can think of 
nearly a dozen VERY different things commonly described by that word, so 
using obscures understanding of what they are and how they work. :)

Another point that I emphasized to that applications engineer was that 
it is NOT, DC wiring, it is, by virtue of the continuous switching of 
the waveform, both DC and RF wiring.

73, Jim K9YC