Hi Zoof,
What you have there is a very low frequency version of what I call "jelly beaning". The term "Jelly beaning" is a phrase that I coin to describe when you are using a DPSS laser to do laser graphics projection, and instead of seeing continuous lines, you see dashed lines, as if you lined a bunch of jelly beans up, end to end. You can also get this effect out of NEON, if you use a high frequency transformer. Some transformers are designed to produce the "Jelly bean NEON" effect...
Anyway, what you have there is a very low frequency version of that. What's happening is that when you turn the laser on, it generates heat. That "heat impulse" takes a while to settle. While that's happening, the TEC might kick on to compensate.
The vast majority of DPSS lasers are only designed (if they are even designed at all...) to operate at CW. That is, to produce an unmodulated beam. Also, "in the lab", they are usually only tested while producing a full power CW beam. So then the manufacturer (if we can call some of them that) puts the laser into the hands of a lightshow guy who modulates the power all over the place.
Since this "heat impulse" aspect takes a certain amount of time to occur, you will find that the power level of your laser will seem to change, depending on the graphic images (and thus the ratio and time of on/off intervals) you are displaying.
The the circuit that was described earlier may help you for a few reasons. First, the primary purpose of that circuit is to be give you the ability to adjust the delay of each of your lasers, so that they (hopefully) can appear to be similar to other lasers you are using. For example, a regular diode laser will turn on and off much faster than a DPSS laser. So normally you use one of these circuits with a diode laser to slow-down the diode laser, to be the same speed as the DPSS laser. Also, DPSS lasers may require a different amount of time to turn on, when compared to the time it takes to turn off. So that's why the diode is there -- to allow you to adjust the turn-on time separate from the turn-off time. Even though this circuit will find primary use in the path from the computer to the regular diode laser, we have found that it may help reduce the "jelly beaning" effect, by "softening" the rise time and thus, helping to "soften" the heat impulse effect.
In your case, this is a pretty low frequency effect, so I doubt that the circuit will mitigate what you are seeing. But I guess it's worth a try... In any case, the circuit finds particular use in helping to avoid "colored tails" that appear in graphic images.
Best regards,
William Benner