Historical tone changes

I'm going back to the tone changes in Zireen languages. One of the things I
noticed about Simik tones is that there are fewer than the possible
combinations of tones in two-syllable words. In fact, most of them have a
high tone on one syllable and a low tone on the other. Some of the main
exceptions can be explained by phonetic change -- e.g., after a long low
falling tone, the second syllable has a LOW rising tone instead of a high
rising tone if it's short. This could be a case of assimilation (one
question is if this sort of assimilation is realistic).

So I'm trying to reconstruct successively older versions of Simik; then I
can work forward from Proto-Simik to create related languages. I'm not
concerned right now with working out the relationships between Simik and
Tenai, if they exist. They could be the result of borrowing.

Originally it seemed that the earlier form of Simik (which I'm tentatively
calling Sumig) had a kind of pitch accent system. Either syllable could be
stressed, but the stressed syllable could be rising or falling. But
languages with pitch accent systems generally seem to have some
restrictions on tones depending on which syllable gets the stress (from my
very limited knowledge of Serbo-Croatian and Greek).

Now it seems that the unstressed syllable can also be either rising or
falling. This explains most of the data, and the few exceptions can either
be altered or explained as borrowings from other languages. So what I'm
wondering now is whether Sumig originally had two tones (rising and
falling), and the language somehow developed a distinction of stress (high
and low) on top of that, or if it originally had a stress system, and the
contour tones developed later. High/low distinctions in single-syllable
words might have developed from shortened forms of two-syllable words.

What I really ought to do is learn something about a number of tonal
languages and try to figure out how the tone systems work. It may turn out
that Simik is an unnatural system and needs to be altered in some way. I
can't think of any languages that have only contour tones and no level
tones. But beyond a few of the more common languages, like Chinese, Thai,
Vietnamese, and Tibetan, it isn't easy to find good information on tonal
languages, especially relating to how the tone systems develop over time.
Thai is interesting because the writing system reflects an earlier stage of
development (as is Tibetan). But it's not easy to see how the existing
tones could have developed from the tones represented in the writing
system.