Module 1 showed what a CME is. This is the machine behind it: where the energy to launch a billion tons comes from, and what finally lets it go. Start with a real one.
Play from the flash and pause the moment the bright new wedge appears at the Sun. How long after the flash did the cloud show up? Press I see it and the page reads the clock for you, or type the minutes yourself.
The gap is real physics, not a delay in the data. The flash is the field snapping. The cloud then has to accelerate out through the Sun's atmosphere before anyone can measure it. NASA's catalog time-stamps the front about 15 million km out, which is where the forecast model takes over.
The Sun isn't solid, so its equator spins faster than its poles. The magnetic field is frozen into that plasma, so a tidy pole-to-pole field gets dragged, stretched and wound like a spring, year after year. Where the wound-up field bursts through the surface, its footpoints are sunspots, and above them it arches into loops that hold plasma down.
Loops of magnetic field arch over a sunspot group. Under them sits a twisted rope of field holding cool, dense plasma. It can sit like this for days.
Flares and CMEs run on this stored magnetic energy, not on heat. That is why they come from sunspot groups, and why a tangled group is watched more closely than a tidy one.
This is not a drawing. On August 31, 2012 a long rope of plasma lifted off the Sun and left at about 1,450 km/s. Play it, or drag the slider.
Earth, to scale
the rope from step 2, for real
The pairing of flash and cloud isn't automatic. A confined flare flashes under a strong overlying field that holds the cloud in, so no storm follows. A stealth CME slips away with almost no flash, which is a forecaster's least favorite kind. That is why "X-class flare!" in a headline tells you almost nothing about storms by itself.