With the new aerodynamic model in 1.0 it looks like the terminal velocity calc in mechjeb is broken. Once you've launched it seems to plummet down to 40m/s and then only gradually goes up to around 80m/s and "limit to terminal velocity" is pretty useless to check now.

Sorry if this already got reported, but I couldn't find a relevant closed github issue via search.

Confirmed. Ascent speeds are extremely low with this enabled, and the 'atmospheric drag' value from the vessel info stats shows values way less then 1G when the limiter kicks in.

One more thing you might need to watch for when fixing this is that drag is now dependent on angle of attack and we might need to relax the limiter to prevent it from kicking in too strongly at the start of a gravity turn. Perhaps consider only the portion of drag as projected along the vessel's orientation.

Suggest you use something like reduce_throttle = atmospheric_drag *cos(angle_of_attack) > local_gravity

Sorry haven't tried the dev build yet. Looking at commit 9c6f0cf I think that will solve the problem of the calculation being way off, and a brief experiment seems to confirm they still model drag as speed-squared.

I still think you'll have a problem with throttle cutting out at the start of a gravity turn when the AoA increases and the new aerodynamic model will spike the drag. A sudden cut of throttle at that moment I suspect will lead to many rockets flipping over.

This was fixed in the dev version already. Did you try that ?

As for the limiter I thinks that I ll move it on real surface speed instead of the vertical component.

Drag shouldn't spike too high on the gravity turn if the AoA isn't too high. If your AoA is too high then TV limiting will likely not be your greatest issue.

After playing around with it a fair bit I think the new version is good enough for all but extremely tall/skinny rockets. If the AoA is causing that much of an increase in drag you have bigger stability problems than MechJeb being over-eager with the throttle limiter.