On the whole I tend towards the coherentist explanation of knowledge. An item of knowledge that is 'coherent' to one person may not be to another, either because they have a wider/smaller circle of knowledge or because they place more weight on some beliefs than on others.
Coherentism would deny that there is foundational knowledge (such as the belief in the truth of a loving God) but I don't think it would deny that people have differently constructed belief systems. NB a coherentist could believe in God, but would not hold this as a foundational belief.
A foundational item of knowledge is held (justified true belief) without further support or appeal to other items of knowledge, either because it is innate (existing from birth) or self-evident. Descartes is a good example of a classical foundationalist because he traces his knowledge back to the foundation of the knowledge of his own existence. I think he would argue that this foundation is self evident, through the power of reason.
Coherentism would also deny infinitism, where the support for a particular item of knowledge is always another item of knowledge (ad infinitum).
The problem with foundationalism seems to me that it is impossible to identify and agree on foundational knowledge or to explain how we come by it. Self evidence isn't really a great argument since, as has already been pointed out it seems to assume quite a lot, to take quite a lot for granted.