In short, the UKRoC event didn’t go nearly as well as we had hoped.
This is an extensive post. You have been warned.
Report of the competition
We went into the competition with high hopes and expecting to do reasonably well, potentially qualifying for the next round. We originally experienced some technical difficulties which we quickly resolved, and resumed the setup of our rocket on the pad.
One inconvenience we experienced here was that there was a point, during the setup, at which the electronics package could no longer be accessed. There were many steps left in the setup phase after said point. This presented a problem: connecting the sections of the rocket would cause small pressure spikes, enough for the electronics to think it was just launched into the air, triggering the parachute to deploy, and forcing us to repeat the setup.
We got around this problem by adding a timer at the beginning of the program so that we would have 5 minutes before the electronics would boot up properly. This was an OK fix, but left us guessing whether the system had booted, or whether the calibration (after the timer) had finished, or if the electronics just simply broke somehow.
We then proceeded to launch the rocket. Reviewing the shaky video footage we took, it is clear that the rocket bent off course immediately after it left the launch rod.
Since the altimeter system we were using was fairly inaccurate, we could not trust it to make accurate speed measurements, since any altitude inaccuracies would simply be exaggerated when converted into speed. This meant we had 3 triggers of deployment: either above 236.25m, going to be above 236.25m in 0.5s, or 7 seconds passed since launch. We could not use a “negative vertical speed” trigger, since that would often cause false deployments.
The rocket – due to a multitude of factors – did not reach its target altitude, and due to the altimeter problems discussed above, the rocket did not deploy until a significant amount of downward speed had built up. This meant the prevailing wind, which was blowing from the buildings to the launch site, was causing the rocket to be guided to the buildings, rather than blowing the stages under parachutes into the opposite field.
This meant the lower stage – containing the motor mount and fins – landed on the roof, and was irretrievable. The upper stage parachute was slightly larger, so the prevailing wind blew it just shy of the roof, saving the more expensive part of the rocket.
With the kind help of the DT teacher of the school, we quickly cobbled together a bottom stage within 50 minutes.
This time the flight went worse, with both the upper and lower stages landing on the roof.
The upper stage did not get blown down until several weeks had passed.
Needless to say, we were not happy.
Speculation over what went wrong
There were two general problems that caused the failure of our rocket:
- The rocket didn’t go nearly as high as expected;
- The rocket consistently landed on the roof of the nearby building.
These problems were not observed before the day of the competition, and so were extremely surprising when they did appear.
What caused these problems can be put down to the following points:
- Poor quality altimeter – had an accuracy of +-1m.
- Too many separable parts of the rocket.
- Connections between these parts were not cut square.
- Poor fin alignment.
- Rocket was unnecessarily long.
- Electronics calibration was on a timer.
- No backup rocket built.
- Gas cylinder was unnecessarily large and – by consequence – heavy.
- Electronics mount could have been sturdier.
- Components inside rocket were free to move about – not secured in place.
- No way to diagnose issues with electronics during / post setup.
- Parachute took too long to unfurl.
- Launch lugs were not aligned properly.
- The paint was not perfectly smooth.
Quite a list.
Solving the problems
However, since the competition, and after our reflection on the problems of the rocket, we have made leaps and bounds in the effort to remove these problems.
Problem 1 has already been solved, with the help of a new altimeter with a faster update rate, and a new way of parsing the data from the altimeter. We have now implemented a basic Kalman Filter, which finds “the signal from the noise”, per se. We now have an altimeter system which is accurate to +-2cm (which is 50x better than what we had before). I have attached a video below:
Problem 3 was compounded by problem 2, which in itself was caused by problem 5. We have redesigned the rocket, such that it is much shorter and has only 2 joints (one for the parachute section and one for the nose cone). We will take extra precautions when constructing the rocket to ensure that the joints are perfectly parallel.
Problem 4 is being managed with the help of 3D-printed fin alignment jigs. They will allow us to make markings accurate to 0.1mm about where the fins should be positioned, as well as assisting with cutting the joints parallel to each other.
Problem 6 has been solved with the help of a radio transmitter and receiver chip. The transmitter chip will be incorporated into a remote-control of sorts, and the receiver will be housed on-board the rocket. When we press and hold a button on the remote, the rocket will calibrate itself. With the press of another button, we can make the rocket reboot.
In order to solve problem 7, we will ensure that testing is complete several weeks before the competition itself, so that we have plenty of time to order and construct a backup rocket that is identical to the original.
The gas cylinder has been made significantly shorter, such that it now weighs 22g as opposed to the 60g it weighed before, so now problem 8 is solved.
The electronics mount has been redesigned and printed. The mount keeps the electronics far more organised and secured. It also does not wobble nearly as much as the old model – even without supports. With supports, all wobble will be eliminated. It has also been redesigned such that the gas cylinder, servo valve, electronics package and gas syringe are all connected, secured, and packed far more compactly, such that the rocket is less long, needs less joints, has all components secured inside. Problems 5, 9 and 10 have thus been solved.
Simply adding a small piezo-electric buzzer has made diagnosing problems far faster and easier. It will also allow us to be certain that the system is functioning, and has no significant problems, solving problem 11.
Problem 12 can be solved with the help of some talcum powder. The powder will be added to the chute before folding it, so that it will deploy several times quicker.
Problems 13 and 14 can be solved with the help of the previously mentioned alignment jig and a healthy dose of patience and carefulness.
We failed in the competition, with our entire rocket ending up on the roof of the building. However, all of the problems we thought could have contributed to the failure have now been solved.
All that remains is to construct the rocket and proceed with testing.