Phi 29 has a rolling amplification capacity and mtDNA is over-amplified (or any circular DNA) versus rest of samples.
Phi29 prefer's long template (if linear) versus short (linear) one.
If anyone faced similar technical challenge, please share possible solution/ideas.