I have seen some examples and read explanations of vowel harmony , but I still can't get a good grasp on how it works and how one might use it in a conlang. Anyone good at explaining it or can point me to somewhere that can?

Basically it means that a word cannot contain two vowels that "disagree" in a certain feature, like height, nasalization or rounding. For ease of illustration, let's talk about rounding harmony, and assume a language where all vowels in a word are either unrounded [i e a] or rounded [u o]. This leads to two main effects:

1. No roots have vowels that disagree in the given feature, i.e., there are no roots like */tagon/ or */kubel/. Because of this, you can characterize the root in terms of the feature value they show, e.g., /liben/ is a [-round] root and /postuc/ is a [+round] root.

2. Affixes must agree with the feature value of the root, i.e., you can't put a [-round] affix (like /-im/) on a [+round] root (no *[postuc-im]). Because of this, affixes are underspecified in terms of the given feature value, e.g., an affix could be [-round] [-ek] with [-round] roots ([liben-ek]) and [+round] [-ok] with [+round] roots ([postuc-ok]), so the UR of the affix would not be specified as either [+round] or [-round].

Does that help? I would suggest looking at the morphophonology of a language with vowel harmony like Turkish or Finnish (there are tons of resources on these). If I have time (which I won't), I could walk you through rounding harmony in Yokuts. But you could look up Yokuts on your own, too.