For sampling flaw type, when we say the sample is large enough, what is the amount in usual?

For instance, for 100 sample, it's enough.

What about 50? Is there a specific amount that we consider to be enough for the survey to be not biased?