You'd be hard pressed to ever directly observe something that is astronomically tiny and emits extremely low frequency radiation, if any at all.
When an accretion disk is actively ripping apart a celestial body the plasma is some of the hottest material in the current universe. It's extremely violent and bright. You're seeing a gravitational well converting a huge quantity of mass to energy.
Yes, as I understand it, the “turn-on” effect is from the plasma generated while the star is falling into the gravity well. The bit of the star that passes the event horizon would be tiny.
Moreover, I believe that the radiation from an object falling into a black hole is redshifted so rapidly, that it effectively disappears from view in a very short time. An outside observer would not see anything “lingering” near the event horizon.
It didn't actually "eat a star." A star passed close enough to it that the tidal forces ripped the star apart and the star became a ring (accretion disk) around the black hole. Matter in an accretion disk gets very hot because of gravitational acceleration and friction -- so hot it can shine more brightly than the original star.
And what probably really happened is that the star was massive enough that it caused the black hole to start emitting jets. There are always two of these pointed opposite one another, and one happened to be pointed toward us.
I think this is good evidence in favor the plasmoid model from Eric Lerner as a replacement for black holes. Some key differences.
* Instead of a black hole eating a star, a plasmoid is having an increased load. * Black holes are "gravitational" machines while plasmoid are "electromagnetic" machines.
* With the plasmoid model there is no time dilation
* With the plasmoid model, the load is any source of plasma, not necessarily a star
It might be good evidence for something else if it were correct. It’s not correct, though. As I mentioned in another comment, infinite time dilation (from the reference frame of an external observer) is only an issue exactly at the event horizon. But the events that lead to a supermassive black hole “switching on” are happening at the accretion disk, far enough away from the event horizon for time dilation to be an effect we’d have to measure carefully to detect.
Also, the idea that you’d want a model that avoids time dilation here makes no sense. We know gravitational time dilation is a real effect because we’ve measured it, it matches the predictions of general relativity, and GPS would be very inaccurate if they didn’t correctly take the effect into account. To say that time dilation doesn’t occur near a black hole is completely inconsistent with well-verified facts.
From this, we can conclude that Lerner’s conjecture is wrong. We can’t even call it a theory, because it doesn’t match the evidence.
As stellar matter is consumed by a black hole, some of it is ejected through gravitational flinging (think gravity assist maneuvers). It isn't ejected after consumption, but during. In some cases, the acceleration of matter causes energetic radiation (x-rays... etc) in the opposite direction. This is often how we detect black holes in the first place (gravitational lensing, detecting of accelerated matter are others).
