https://iesupply.ca/wp-content/uploads/2015/09/Ballasts-and-LED-Photo-Controls.pdf (inrush about 100x)
https://www.lrc.rpi.edu/programs/solidstate/assist/pdf/assist-technote-dimming-inrushcurrent.pdf (article about repetitive peak and dimmer stress)
https://www.digikey.com/en/articles/compact-fluorescent-tribulations (failure claim on relays related to SMPS ballast)
"CSA Standard for electronic ballasts, C22.2 No. 74, has been updated in 2016 to include testing requirements that will demonstrate a means to limit peak inrush current to a pre-determined levels. Since these testing requirements will come in effect on July 1, 2018, industry should be aware of the above recommendations." and since they don't like building separate CSA and UL versions, newer LED ballasts are designed to address the inrush issue.
CFLs and some LEDs, ITEs, and the most basic power supplies operate as a DC 165v or 330v fed load. They all share the same front end which looks like the attached photo. In the most basic form, such front-end has a power factor of 0.5-0.6, which is typical of CFLs and non-dimmable LED. The idea is to use a big ass capacitor behind a fridge and have a 165v DC rail with as little ripple as possible to avoid 120 Hz flicker in light output.
When you have say... 50 14W CFLs of this topology go online all at once at the wrong time (capacitors completely discharged, and switched on just before peak crossing), the inrush current can activate
magnetic trip, not thermal overload. Older power supplies over 1kW often cold started through power resistors which were shorted out with a triac or relay once capacitors have reached charge in order to prevent this kind of trip.
A large number of CFLs or some LED ballasts behave similarly. Once they're turned on, the steady state current is low enough from the perspective of magnetic trip, so you can assume they're not relevant. It's really a non-issue, because you're rarely going to have a setup where 50 CFLs are powered on all at once.
Electronic fluorescent lamp ballasts of 2000s vintage for commercial and industrial installations weren't susceptible, because they used a sizable iron core reactors on the input end to taper off the di/dt in order to raise the input power factor to above 0.9. This inductor is of such size that the ballast often ends up weighing as much as a magnetic ballast. Such a passive PFC also happen to limit inrush current.
Modern L.E.D. lamps often have a permanent 10 ohm series resistor on the input or they use very minimal capacitor and allow considerable 120 Hz flicker in the output. Flicker regulation on LED lighting systems are often dramatically inferior to T8 electronic fluorescent system and this manifests itself as noticeable flicker or stripes in photo/video. You might just see barely noticeable stripe, or the flicker content is so severe that you'll get actual black stripes.
