You’re right: it is quite possible that you are using the light box too early, for too long a duration, at too high an intensity—or combinations of these dosing factors. Let’s assume you are taking a 30-minute treatment session at 545am, while having woken up prematurely at 4am. As a first step, see whether delaying light therapy to 615am resolves the problem, fully or partially. You may have to play with session timing to get it right. You may also need to reduce session duration from 30 to 20 minutes, for example. The AutoMEQ gives a starting point for finding the precise session time to serve your need best—and that time may vary at different times of year. If the sun is rising at 5am in the summer, for example, you may need to reduce the light dose because natural dawn illumination is doing the trick for you. In the fall, if you start waking up later, you would resume treatment. This guideline applies whether or not you experience SAD.
There is no harm trying, but we cannot confidently predict the result. It has been hypothesized that “early types” need a phase delay (from evening light) to show the antidepressant effect, while “late types” will respond to phase advances (from morning light). Thus far, there are no strong data to support for that hypothesis — most people respond to phase advances, regardless of their chronotype.
You might find the following trick helpful: use the light in late evening, right before bedtime, for one week. (You may find your sleep episode delays.) Then, immediately switch to morning light at your new wake-up time. That will give you the desired phase advance, and your sleep may stabilize without premature awakening.