LannyRipple

Ideally, like time, you'd split the difference. Thought experiment, you are orbiting a body. We've stolen the third dimension so can say you are orbiting clockwise such that your ship is oriented to 3 o'clock when your position is at 12 o'clock. I.e., You're ship points to the right. You wish to give an impulse at 12 o'clock in your prograde direction but your thrust is so low you calculate you'll have to start at 9 o'clock and end at 3 o'clock. As such you should start the burn at 9 o'clock when you are radial -, burn through the intended 12 o'clock (now halfway through your burn and prograde), and finish at 3 o'clock when you are radial +. The vector change will be the same as if you instantaneously burned prograde at 12 o'clock. In KSP just set up a maneuver node and burn on it half the time (plus a touch for acceleration) before the node encounter time and half after. PS - Hmm. Thinking on it the total vector would be off but not sure how you'd do better with the stock navball.

Your extra weight is all the fuel you are using. You could get by with a smaller setup. I noticed you mentioned you haven't done any research so without knowing your stage deltaV or TWR it gets a lot trickier (coming down to luck and experience from what you've tried before). Dave -- You should probably also ask folks if they use design tools like KER (or by hand) or just eyeball it. You are going to get radically different values for the two approaches.

Launchpad weight is 482.1 tons with 122 parts. LKO lifter is 67 parts and will put 43 tons into LKO with 400-500 dV remaining in last stage. I could redesign the second stage so it wasn't such overkill but so easy to just grab the assembly and apply. My Minimus manned uses the same lower 3 stages but has a launchpad weight of 498.4 tons with 182 parts (yay! Panels!! :/ 59.2 tons into LKO with lower three stages consumed. Both Mun and Mini use an Apollo style lander so 2 Kerbals sent on each mission. I have T7 filled in and the cupola node in T8. (Which means I need to redesign landers for the Biometric Sensor Array.) PS - Just played around a bit with T5 + 3x T6 (lv909, heat-shield, landing legs). I missed getting home by 70-100 dV but got a full landing with Jeb, Bill, and Bob and back into orbit only failing to get all the way out of Munar orbit for a transit home. Launchpad weight was 472.1 tons with a part count of 168. PPS - I use MechJeb for weights, deltaV, and TWR.