If we follow the treatment of stress in SPE, when part of a phrase comes to have stress, stresses on other elements are not lost, but merely become subordinated to the new stress. If there is no stress at all, a non-diphthongal vowel will reduce to schwa, so if a vowel fails to reduce to schwa, we can take that as evidence that the vowel does have some stress, though the stress may be at a low level due to having been subordinated to other stressed syllables. (There are complications to this that I won't mention for now, and, of course, many don't accept the SPE stress theory at all.)
So a first question to raise about your examples is: do the first vowels of "not" and "funny" reduce to schwa? My opinion is that they do not. So, given that, in "It's funny" and "It's not funny", both "fun" and "not" are stressed. What the stress levels are, will depend on whether the "not" is taken to be contrastive: "It is funny" versus "It's not funny". If so, the highest stress will be on the "not": "It's n1ot f2unny" (meaning that the most stress is on "n1ot" and a secondary stress is on "f2unny".
If there is no contrast, primary stress will be on "f1unny", following the Nuclear Stress Rule, which give the last primary stress in a phrase (here, the verb phrase) the most prominence. Then, we'll get "It's n2ot f1unny".
I have left the stress of "it's" in the examples out of account. Pronouns are a special case, and I think it might either be stressed or not, here.
In my speech, the vowel of uncontracted "not" always has some stress.