Yes indeed, I have seen that.
They use a fixed 250kHz instead of a jittery clock like used in many supplies for better EMC performance. A jittery clock could induce unwanted audio artefacts.
Note that besides powering the control circuit, the 15V control voltage, after being divided by R613 and R614, is also compared to Vref by the TL494.
I believe you when you say the unit starts working at 14V and stops below 12V but that is part of the feed-back scheme they use. The switch noise you hear is the TL494 trying to restart the supply but being unable because there is not enough voltage coming from the auxiliary primary winding to create the feed-back loop. May be something as stupid as C634 not being able to store the energy. But my first guess was excessive load on the secondary side.
In SMPS everybody is a suspect.