Pretty much any AI that I would design. An AI gains a “subconscious” by first having to juggle it’s own priorities and then by not being able to sufficiently accomplish its goal. The most efficient use of mental capacity then requires a division between what you call a conscious and a subconscious. The “conscious” portion builds a completely different imagery to represent the surrounding reality, including the urging to attend more to this or that issue as directed by the original priority juggling. Since those urgings are separate from the conscious, perceived as “remote” from the conscious yet within oneself, they are sensed as “feelings”; joy, hate, love, depression, frustration, …
A social example would be the advent of activist groups in Congress. An AI properly formed, similar to Congress, would naturally form its own activist urgings (the priority juggling) so as to persuade the Senate (the conscious). Those activist urgings are what “emotion” is.