we need details like the product size, the annealing temperature, the concentration of dna added, the cycling parameters and ideally a picture of the failed amplification. I would measure the od of the dna and use about 50ng DNA and run an annealing temperature gradient or if that is not possible then run the pcr cycle with a low annealing temperature to get a dirty result then raise the annealing temperature until the amplification become a clean single band