I’ve started construction of the weights that will be swung around by the Servos. They each have the following properties

• Built from Steel Rod with diameter 50 mm
• Thickness of 15 mm
• Angular distance is 90 degrees
• 10 mm cut out from centre
• Weight is approximately 46 grams each

A picture of these is shown below.

By having these weights positioned so that their curved edge is 26 mm from the centre of the rocket, we should be able to move the Centre of Gravity by 1.25mm. This will be a significant achievement if we can complete this payload. This will mean with a 100N thrust motor, we will get a moment of 0.125Nm. If the rocket has a Moment of Inertia of 0.25 kgm^2, then we are looking at Angular acceleration of 0.5 rads-2. This means that for a burn of 1.5 seconds in a vacuum, this would result in:-

• Angle of Rotation = 0.5625 radians = 32 degrees
• Rotation speed = 0.75 radians per second = 43 degrees per second

During tests on ground (in atmosphere), we would experience aerodynamic forces that would help to counteract this movement, so the effects would not be so great.

In this post we focus our discussion on the device used to adjust the Centre of Mass. This device consists of two masses that we rotate independently to produce a resultant change in Centre of Mass. We split the masses in two to:-

• Halve torques required to rotate them (if they were combined)
• Allow us to obtain neutral position (by having masses opposite each other)
• Allow movement of Centre of Mass around…by moving masses
• Reduce movement of Centre of Mass by moving masses apart

This isn’t a new idea. We solved this problem with stepper motors. Unfortunately the stepper motors have problems. The are:-

• Too heavy
• Have insufficient torque
• No means of verifying the position
• Large current requirements
• Require specialized electronics – Driver board

So I put some thought into how we could make this lighter and have greater torque. The only solution I have been able to come up with is a geared system using high torque servos.

The trade-off is that we can’t continuously rotate around. But this shouldn’t be necessary because we only need this system to operate for the first 1 to 2 seconds of flight. We do not expect the rocket to rotate about it’s Y-Axis significantly. Put another way, we don’t expect erratic motion, instead some clear rotation in one plane that requires some correction. There may be some rotation as the rocket builds up speed and the fins (with their imprecise alignment) results in some rotation about y-axis.

We have started work on a “pre” prototype system using Perspex. Below are a few pictures of system connected to Arduino.

 

We ultimately aim to have 3-D printed components to form the framework. Below is a preliminary design for the bottom section that the servo is joined to.

 

[stl file=”framework_v0.02.stl” ]

 

We have ordered components from ServoCity. This looks like a very good supplier of quality components; certainly comprehensive.  Below is a picture of the parts we have recently purchased.

 

We had problems with the 48P 36 tooth gear. It was larger then the Spline on our Servo. So, I drilled out the servo gear carefully and then press-fit a Servo attachment into the gear (which I’ll later glue). Then I carefully drilled holes in the Perspex and press-fit the bushes. The final mechanism looks like:-

The gear ratio is 3:1.

This means:-

• Travel distance is now 3 x 135 = 405 degrees
• Travel speed is now 60 degrees in approximately 0.017 seconds
• Torque is reduced from 3.7kgf.cm to 1.23kgF.cm

We will ultimately look at adjusting the gearing,by inclusion of another 12 tooth and 24 tooth gear to give us a gear ratio of 6:1. Initial design work suggests that this should be doable. But for the purposes of the next flight test, we do not need to work on such a system. This system should be adequate for characterizing the effect of small movements of the Centre of Mass.

 

A lot of progress has been on the stabilisation system. We have managed to:-

• Get the Stepper motors working in 1/4 steps. We needed 1/4 steps because full steps were causing vibrations and we would miss steps.
• Use Interrupts and clever coding to calculate (and buffer, using a Ring Buffer) timing values. We didn’t have enough memory to store timings in array and we could calculate just one time interval per step.
• Managed to implement Hall Effect sensors to detect when Smoothers are close to pre-determined position. This allows the system to calibrate the smoothers, i.e. move them into an initial starting position.
• Managed to get the Gyroscope working error free. We did this by getting to to check the status register to confirm data has been written to the registers. We also implemented a single Wire Read to get all the gyro data, rather then individual reads. This should give us extra cpu cycles while the stepper motors move. Very very critical!
• We calculate HIGH/LOW values for X, Y, Z. We also calculate average and variance of these values. We use the high/low values for x, z axis to set the values to Zero, should they fall within the range.

 

Below is Youtube movie of the system being testing. The video was slowed down because the smoothers move so fast, one cannot pick it up with ones eye!

 

Keeping a rocket pointed up when launched from a balloon is not an easy feat. Countless hours of thought have been put into this and several ideas have been considered and then dismissed. Some of these ideas are presented below.

Have a rocket with Long Launch Rod

Can we use a launch rail, like we do on Earth?

No.

We need to first realise that while there is very little air at an altitude of 30km, there is still sufficient air to allow stabilisation using fins. This is provided we are travelling fast enough. The reduced air density means that the velocity required is substantially more then at sea level.  This means we would need a longer launch rail to ensure that the velocity is sufficiently great when it leaves the rail tip. Let’s consider the equations that give us lift.

Lift = Cl * density x Velocity ^2 * Area/2

Density of Air at sea level is 1.225 kg/m^3

Density of Air at 30km is  is 0.01841 kg/m^3

Let’s assume that 20ms-1 is the minimum velocity required at sea -level. With this we can estimate the minimum speed required at an altitude of 30km

velocity = sqrt(density-at-sea-level * velocity-at-sea-level^2/density-at-30km)

= 163ms-1

Let’s assume the rocket is accelerated at 10g  (98ms-2) up the rail.

v = a * t

t = 163/98 = 1.66seconds

s = 0.5 * 98 * 1.66^2 = 135 metres

This is clearly impractical. The launch rod would be very long, heavy and unwildy. We would need a massive balloon and someway to stabilise the rod. Then we would have the issue of the launch rod coming back to Earth. Totally out of the question!

Spin stabilisation

Another option considered was spin stabilisation, where by we attach rockets to the side to spin it up. This makes the rocket act like a spinning top. Simulations have been done to see how attaching two D-engines to a rocket, to see how fast it will get and how much the spin is affected when engines are slightly mis-aligned. Unfortunately. even with very small manufacturing tolerances there is sufficient wobble to cause issues. We tried to increase the Moment of Inertia in in some axes to reduce the amount of rotation in those directions, however this adds alot of extra weight which is extremely undesirable.  There is also the problem of how we suspend the rocket when we spin it up, or if we have to release the rocket and then spin up.

This solution is extremely unpractical and complictes the launch sequence considerably. There a lot more possible modes of failure.

We have decided against this option.

Thrust Guided Rocket

The idea here is that one can either:-

– Move the nozzle, which will adjust the thrust direction

or

Move some vanes that are downstream of the thrust that re-direct the thrust direction

With the solid engines, the former is inpractical. The nozzles are fixed.

With the latter, the design is not simple, or easy to simple and some of the engineering required is not so simple. We would need some external actuators, ones that can provide a massive torque (to combat the force of the rocket thrust), while still keeping the rocket streamline, and at the same time using the limited ‘realestate’ at the bottom of the rocket. For the vanes, we would need special material that would not be destroyed by the thrust, be light and have a specific design to induce the right sort of adjustments.

We have decided against these two options.

Centre of Mass Guided Rocket

This is not a commonly used method, however, it has been mentioned a little in some of the rocketry literature on the Internet. The concept is that you adjust the Centre of Mass (moving it laterally) so that the thrust, in combination with this off-axis centre of Mass produces a resultant moment.

Imagine yourself standing up and someone pushing you backwards. You instinctively put your arms/hands outreached in front of you to try and move your Centre of Mass forward.

This method could work, but it would need to move masses quickly and efficiently without producing too much un-wanted disturbances to the rocket’s motion. The whole system would need to be relatively light, so as to not be a burden on the rocket.

Summary

The latter option seems the most likely to succeed. We have decided to put a more thought into such a solution. In particular, we decided it would abe a good idea to try and simulate such an engine.