Radio Frequency Interference

My neighbour installed a 3-ph grid-tied 5kW solar system. No optimisers just 17 x 300W panels and 3 phase inverter. I wasn't home the day the system was commissioned but on day 2 I found the system generated significant radio interference for most of the day. In character the interference was broad band similar in character to that produced by HV line insulators tracking.
I got hold of some DC input cable and AC output 5 core with a plan to make some common mode chokes. Meantime the interference has reduced day by day to the point where with a yield of 40-50kWh/day it now makes only very occasional interference.
I am puzzled by the changes in behaviour and wonder whether someone with more solar experience might have an answer? Do MPPT systems learn, initially going through a process of load variation under varying illumination conditions to generate a profile suited to array characteristics? Or has something in the inverter been arcing and now stopped?
Thanks.
Bob, 5B4AGN
Comments
Second system 1890W 3 × 300W No name brand poly, 3×330 Sunsolar Poly panels, Morningstar TS 60 PWM controller, no name 2000W inverter 400Ah LFP 24V nominal battery with Daly BMS, used for water pumping and day time air conditioning.
5Kw Yanmar clone single cylinder air cooled diesel generator for rare emergency charging and welding.
As for why the interference would diminish over time, I dunno. Any loads (eg motors, flourescent lights) that might have changed?
Main daytime system ~4kw panels into 2xMNClassic150 370ah 48v bank 2xOutback 3548 inverter 120v + 240v autotransformer
Night system ~1kw panels into 1xMNClassic150 700ah 12v bank morningstar 300w inverter
In general, RF noise has many sources... From the switching circuits that may give you "birdies" every 10-25 kHz or so, to broad band noise which can be current flow through diodes (as an example--Diodes are surprisingly "noisy").
Reduced emission over time... Variation in current flow (cloudy days, etc.), a poor conduit ground that somehow got "good" (bumped conduit, tightened conduit fittings, broke through paint on box hole, ran system without covers--Then buttoned after a day)... Something that was acting like a resonator or antenna (extension cord, tools stacked near inverters, etc.).
In general, when I did EMI design testing on computers, you could get within about 2 db of repeatability between setups... And frequently as much as 10 db variation between setups, labs, minor changes in wiring/emi control/etc...
What frequency range(s) you run, will have different sources.... Typically, conducted noise below 30 MHz, radiated from cables for 30-180 MHz, and >180 MHz were 1/4 Wave Length slot antenna in cases (doors, card slots, rivet/screws for enclosures, etc.).
-Bill "been a long time since I did that stuff" B.
-Bill