Look for the stuff John Pursley did with gimballed rockets... he did at least two NARAM projects (large model rockets) including a finless Vanguard rocket with engine-gimbal stability, a Mercury Redstone, and a Saturn V (IIRC). I've seen/held the Redstone and Vanguard... You can check out his website at
www.accur8.com
What John did was take a model airplane 'stabilizing' system (sort of an autopilot) that is equipped with four horizon sensors arranged in a horizontal 'ring'-- like the four cardinal points on a compass. The horizon sensors feed their input into a microcontroller, which then sends signals to the pair of servos controlling the engine gimbal mount... one for "pitch" and the other for "yaw". IIRC he said it was an off the shelf model airplane unit, slightly modified to fit in a rocket.
I asked him how it worked. He told me that the four horizon sensors look out from either the nosecone (in the Vanguard) or from a ring at the top of the body tube (in the Redstone) and sense the horizon... The sensors detect carbon dioxide, from what he told me, so they see the atmosphere below the horizon as "white" and the atmosphere above the horizon as "black", since it's looking out to infinity through hundreds or thousands of miles of atmosphere, whereas below the horizon there's only a couple miles at most of air between the sensor and the horizon, so that part of the sensor receives light that is blocked by the C02 in the miles and miles of atmosphere above the horizon... so each sensor "sees" a white and black bar, intersecting at the "middle" of the sensor's field of view. The microcontroller gets the feed for all four sensors, and compares the opposing pairs... if the rocket tilts substantially, the sensor on one side of the rocket will see mostly "ground" (a lot of white with just a little bit of 'black' at the top of the sensor's view) while the opposing sensor on the other side of the rocket will see mostly "sky" and very little or no "ground" (that sensor will see mostly black "sky" with just a little bit of "white" "ground" at the bottom of the sensor's view). The microcontroller simply compares the two signals and measures the discrepancy between the two, and then issues the appropriate directional command and proportion to the servo to "steer" the rocket engine which pushes the rocket back on course... The hardest part, he said, was "clocking" the system to ensure that the sensors and servos were 'in synch' and the microcontroller was steering the engine nozzle toward the sensor measuring more "sky" than "ground" to correct the trajectory. The other pair of sensors measure the yaw and the microcontroller feeds info to that servo in proportion as well. VERY slick system!
Now, the Redstone had scale fins, and was very light... it's a VERY large model but designed for "G" motor power and under the weight limits, so it's classified as a "large model rocket" and doesn't need a waiver to fly. It's true that the fins DO create some corrective force, which certainly helps during the coast phase, because basically the gimballed engine is of course most effective under thrust. BUT, a rocket with a certain amount of INTRINSIC STABILITY can fly with the same system without fins, even during coast... I asked John to explain it to me...
His Vanguard rocket (scale) of course had no fins, just like the prototype. Yet it flew perfectly with just the engine gimbal system for stability. What makes this possible is that the rocket's design has a large "fat" first stage about half the length of the rocket, transitioning down to a smaller "upper stage" for the other half of the rocket's length, so this helps to create stability similar to a cone or plate type stability... the smaller "upper stage" and larger "lower stage" tends to move the CP back, so with some careful tweaking of the CG location, you can achieve basic stability even without the gimbal system. John flew a subscale model of the Vanguard to test the hypothesis without guidance and it flew fine, so he told me. The gimbal system just gives it a lot more stability, since the aerodynamic stability without fins is VERY weak due to low corrective forces and moments of inertia... At liftoff, the gimbal system provides rock-solid stability through the thrust phase, and so long as the rocket doesn't enter the coast phase with high rates of rotation (swerving) the "intrinsic stability" and momentum keeps it going straight. Also, as John explained, thrust is NOT zero during the coast phase, contrary to popular belief... while thrust is ESSENTIALLY zero, as in not contributing to the acceleration of the model, the rocket engine during the coast phase IS putting out gas from the nozzle in the form of tracking smoke. While this effect is nil in regards to pushing the rocket, the gas coming from the nozzle DOES create a SLIGHT force that, when the engine is gimballed, acts (weakly) to correct the trajectory, so essentially the gimbal system continues to work, though at low efficiency, during the coast phase, to some degree...
Now, that said, the rocket needs at LEAST neutral stability for this to work... an INTRINSICALLY UNSTABLE design, like Ares I, which has a LARGER UPPERSTAGE and a SMALLER FIRST STAGE, will act just the opposite of the Vanguard, since the stage sizes are reversed... the larger upperstage moves the CP FORWARD instead of BACK like the Vanguard does, which is destabilizing. The gimbal system will NOT be able to cope with the aerodynamic forces inducing unstable flight after burnout-- not with the whisper of gas coming from the nozzle as the ONLY corrective force, and the rocket will become unstable at burnout. BUT, for "non-hammerhead" rocket designs (most rockets) knowing the limitations of the system, a workable solution can be achieved using a gimbal-thrust stabilized system.
Hope this helps! OL JR