The Drive/Recoil subsystem interface a single tensile force exerted along the transmission line.  Our challenge is to have the D/R receive energy when the transmission line force is high and going outwards, and extend energy when it's lower to pull the transmission line, and in turn the kite lines, back in.

First, we want to introduce a degree of freedom in ratio of the rotational velocity of the drive/recoil system and the kite lines.  We want this so that as our reciprocating cycle size increases (lines pull in and out greater distances) relative to the rotational velocity of the DR system, or vice versa, we can accommodate.

In other words, we want to be able to adjust drive/recoil to steering system gear ratio.  Thankfully, this is a common problem, and solved for bicycles with a chain-wheel/dérailleur setup.  Our use is equivalent to a bicycle rider with one exception.  In additional to the "pedaling" use case, we'll have the cog set pulling back on the chain.  A dérailleur setup can accommodate, so long as we take care to not attempt to shift gears while the system is pulling the kite line back in.

Now, where will we get the energy to reel the transmission line back in.  Some of the energy imparted to the system by the kite will go to the generator.  If all of it went to the generator, we would need to use a motor or some such to pull the line back in, at a significant energy loss (mechanical to electrical back to mechanical is pretty inefficient).  Instead, we keep the energy we use to reel the kite back in in mechanical form.  The most viable choice is a flywheel.  But this part of the system can be stationary, and when we say flywheel, we don't need something terribly refined.  Think a reasonably balanced, concrete filled car wheel rotating at < 100 rpm.

Our final challenge is coupling this flywheel, with our drive-recoil input/output, and with our resistive, energy consumer, i.e. generator.  Like decoupling of drive/recoil in steering, we have a problem of differentials, but this problem is to combine torques exerted via the drag of the generator, the pull on the kite line, and the flywheel's moment of inertia.  As in the last problem of differentials, we again opt to use a planetary gear set.

In the D/R diagram, the flywheel, generator and their chain wheels will only rotate clockwise.  The Green chain wheel positions correspond to kite line length, and these will rotate in both directions.  We couple the components as shown in the exposed view of the planetary gear, the generator and freewheeling chain wheel on the cog hub to the planet ring, the recoil chain wheel to the sun gear, and the flywheel to the outer ring.

Imagine the planet ring is fixed, and the outer ring is rotating clockwise with the inertia of the flywheel.  Then, the sun ring must be rotating backwards with a torque exerted via the flywheel's inertia.  In fact, we don't need the planet ring to be fixed in order for the sun to rotate counter clockwise.  When the planet ring's rotational speed is < 3/4 that of the outer ring, the sun gear must be rotating backwards.  The torque on the sun gear is proportional to the relative speeds of the planet and outer ring, and the resistive torque of the generator on the planet ring.  It may be the case that we have to vary the counter-torque of the planet ring in order to get the sun to enter recoil mode.  In the worst case, we can apply a braking force, but ideally this can be done electro-mechanically by increasing the load of the generator...if that makes sense.  Does it?

Unfortunately, the drive and recoil states are not exhaustive.  The transition between drive and recoil, before or after the freewheel is engaged (and the sun gear has started to or is still rotating clock-wise) corresponds to kite line being let out, but no energy being imparted to the system.  We want to minimize the amount of time the system is in this state, since line let out has to be reeled in at an energy loss.  Fortunately, we get to choose the ratio of the sun gear to planet ring when the freewheel is engaged, and so we choose something large.  In other words, we ensure that the sun gear only moves at 1/N the speed of the planet ring before the freewheel engages.  Providing the kite is steered so that the transmission line force is large when this happens, this factor of N is roughly proportional to our energy profit margin.  For the same amount of kite line let out (corresponding to rotation of sun gear), we increase the velocity of the flywheel by N more than we require to reel the line back in.

The following animation shows a hypothetical drive/recoil cycle.  The thick columns correspond to rotational velocity of the planetary components, while the thin red/blue lines correspond to the length of line let out (proportional to sun gears rotational position).  When the planet ring is red, the freewheel is engaged and we're imparting energy into the system via increasing the kinetic energy of the flywheel (while driving the generator).  Otherwise, the resistive torque of the generator translates into counter-clockwise torque on the sun gear (and pulling back on the transmission line).

To Do:

Thankfully, there's a lot of overlap with the steering system in the mechanical tasks, annotated with (overlap)

• Acquire or construct spool (overlap)
  
• Spool and coupled to chainwheel rotating about fixed axis (overlap)
• Sturmey Archer, pillow-block mounted with freely rotating axle (overlap)
• Sturmey Archer configured internally, with chain wheels coupled to both outer ring and planet ring (overlap)
• Chain wheel mounted to rotating axle/sun gear of S/A
• Chain tensioners for each of the chains, on low tension side, between transmission spool and kite spools
• Acquisition or construction of flywheel.
• Framing

Hand drawn version of the diagram: