Essentially, words are really made of sounds and not letters. The weak forms of have and of are homophones and can both be pronounced as /əv/—or in rapid speech as just /ə/ or /v/:
- /'ʃəd əv 'left/ "should uv left"
- /'ʃədv 'left/ "should v left"
- /'ʃəd ə 'left/ "shoulda left
- /'bɒtl əv 'wɔ:tə/ "bottle uv water"
- /'bɒtlv 'wɔ:tə/"bottle v water"
- /'botl ə 'wɔ:tə/ "bottla water"
In the position after the modal verb (would, should, could etc), the following word, /əv/, is very unlikely to be stressed in normal speech. Because the orthography of the word is just a convention, if this was always the way this word sounded, then it would be much of a muchness whether you spelled it have or of.
However, notice that I said that this word is unlikely to be stressed in normal speech. When this word is stranded, appearing with no following verb phrase, it may sometimes be given its strong form by some speakers. So here, the strong form of the word will disambiguate between have and of.
Not so fast, Ethel. The fact of the matter is that while many speakers will use the strong form /hæv/—"hav"—here, a significant number of English speakers, at least in the UK, will use /ɒv/—"ov". So for the string "I should have/of", you might hear either of the following:
- /aɪ 'ʃʊd hæv/ "I should hav"
- /aɪ 'ʃʊd ɒv/ "I should ov"
Now, English is always changing and evolving, and over time items are reanalysed grammatically and so forth. For speakers who use the second form there, the word arguably is the word of for them. The fact that a large number of speakers use have in their grammar does not mean that these other speakers are mispronouncing the word have. It just means that there is a dialectical difference where their grammar stipulates an ov not a hav in this position. If we see this spreading into writing, it is hardly surprising.
The Original Poster's question
So, in answer to the Original Poster's question, for some people should of might always be considered legitimate. However, for the majority of standard English speakers modal verbs like could and so forth always take a complement headed by a plain form of a verb. And these particular speakers don't happen to have a verb of in their vocabulary.
However, if the question is merely one of whether it's possible—for speakers whose dialects use have after the modal—to have a sequence of a modal verb, the preposition of, and then a past participle form of the verb the answer is not usually. The reason for 'not usually' is that the preposition of must take a complement. It never takes past participle clauses as a complement. It will only accept noun phrases, preposition phrases, and gerund participle clauses (-ing forms of a verb) as a complement. [However, it may be possible for a noun phrase following the preposition of to have a past participle form of the verb as a modifier. I can't think of an example right now, though.]
As to the question of how many ways one might have a string consisting of a modal verb followed by of, the answer is that there is no limit. If part of the verb phrase after the modal is deleted or omitted for some reason (and there are many reasons why it might be) then any adjunct or complement of a noun or adjective beginning with the word of might occur afterwards. And there is no limit to the number of adjuncts or complements beginning with the word of:
- I couldn't find the same number of spiders as I could of beetles.
- You shouldn't be as scared of rats as you should of mosquitoes.
- I would take pictures of my girlfriend. I wouldn't of you.
However, having said all of that, what people like to be righteous about is not the occurrence of the word of after a modal verb. They decry the use of an of-phrase being used as the grammatical complement of a modal.
I suppose I could of told you that at the beginning. Maybe I should of ...
Here are some Language Log posts about could of et cetera: