This is a true story of patience and perseverance … and getting the right people at the right moment…
You may recall that right after launch, SwissCube experienced a high rotation rate around one of its axis (~ 200 deg/s), which basically prevented the team to take meaningful measurement of the Airglow (its scientific objective). After trying a few things, which did not necessarily improve the situation, we decided to let SwissCube detumble naturally. The planned design lifetime of SwissCube was 4 months (1 month of commissioning and 3 months of science observations).
So after almost one year of letting SwissCube detumble “alone”, we got to a slow enough rotation that the gyroscopes onboard were not saturated anymore (about 80 deg/s). That happened around November 2010. That said, the onboard systems had degraded and we still had a big internal communication problem (I2C bus inside the satellite). So we decided to try to reset the satellite. Of course the satellite was not designed for it… as a safety guard: a failure of a reset system could fully compromise the mission… but hey! that’s what engineers are for (do impossible things!). A way around this shortfall was found: drain the batteries by turning on the communication’s power amplifier (PA) continuously.
Lucky that we had just employed Florian George, the former lead software student working on the project, who knew the satellite in all its details and could implement the right commands. Good software design. This commands activated the power amplifier, which consumes about 3.5 W, and forced it to stay turned on. The average power measured on SwissCube coming from its solar panels while under sunlight is 2.8 W. Thus by letting the power amplifier “ON”, the batteries would be drained to a level where the power system resets and the spacecraft reboots in “safe” mode.
We tested this option a few times on the EQM (qualification model, the twin brother of the flight model in space), and saw various behaviors of the power system, but eventually all behaviors ended with a total reset of the satellite. So we implemented this command to turn on the power amplifier on January 20 and it worked beautifully! The discharge started at 13h06 local and the reset was executed about 2h41 minutes after.
The main consequence was that the I2C bus, which had been stuck and prevented internal communication, did reset and the satellite was back to normal (see figure 1).
Figure 1. The I2C reset brings back SwissCube normal communication.
The attitude control system (ACS) and Payload telescopes were fully useable again!
The next step was to stabilize SwissCube’s rotation. After a few verifications of the power system (the batteries had an equivalent of 3000 cycles at that point), we turned the ACS “On” (January 31st), and downloaded the gyroscopes, sun sensor and magnetometer data. These sensors data confirmed rotation rates ~ 80 deg/s around the Y-axis of the satellite.
We could then implement the detumbling strategy that a talented student from TU Delft, Arthur Overlack, came up with during his Master thesis. He is now working for ISIS, along with Hervé Peter-Contesse, the former lead attitude control engineer on SwissCube. With both their help, we could implement the detumbling procedure as shown in Figure 2 (full technical description of the control strategy in paper reference).
Figure 2. This diagram shows the region of stability of the b-dot controller as a function of its sign and lambda parameter. The detumbling strategy followed the green box path.
This procedure consisted in:
- Uploading new software parameters of the b-dot controller (to actively detumble this time with the magnetotorquers).
- Verifying correct function of the b-dot controller.
- Uploading new parameters when the rotation had reached its minimum value with the previous parameters.
Figure 3. This diagram shows the attitude's evolution.
It is to note that this strategy had never been implemented nor tested before on a satellite. The full success of this procedure had still to be demonstrated in flight! And that worked! Thanks Arthur and Hervé!
In the end, this stabilization of the satellite took about 3 “effective” days, a lot less than our preliminary thoughts.