Sprite satellites are just square printed circuit boards of 35 mm side length, carrying electronic components. During the last decade, there were some campaigns offering the launching of private sprite satellites into low earth orbit; here is a crowdfunding campaign currently in progress:
https://www.kickstarter.com/projects/ambasat/ambasat-1-an-educational-space-satellite-kit?ref=discovery_category_newest
While these campaigns are primarily meant to kindle interest in STEM topics, sprite satellites or even "chip satellites" might be scientifically useful in the future.
One important topic is attitude control: The solar cells have to be well exposed to sunlight during each daylight phase, and some measurements require the knowledge of the satellite's orientation relative to earth.
Active magnetic attitude control is well known but, as far as I know, was never implemented in sprite satellites until now. Here, two flat coaxial coils could be placed on the inner layers of a four layer board, being activated if needed just by the supply current of the microcontroller and the sensors. A simulation showed that with a supply current of 5 mA and 10 windings, the rotational speed can be decreased or increased by 1 cycle per second in about 10 minutes, under ideal conditions.
Common star trackers and horizon sensors cannot be used due to volume, weight, and power demand, but I guess that a single IR sensor mounted on a board rolling slowly, and scanning in the plane perpendicular to the flight path could yield enough information to determine the horizon, and hence the position of earth. The following sensor seems quite suitable if equipped with an additional aperture in order to reduce the field of view:
https://pyreos.com/wp-content/uploads/2018/10/ezPyro-SMD-Motion-Sensor-Short-Form-Datasheet.pdf
After more simulations and preliminary tests, I intend to conduct an attitude control experiment in space; but since failure during the mission means waiting for another opportunity, I would love to read your opinion beforehand: Do you recognize problems I overlooked so far? Do you know of a similar attempt? Do you think this approach is feasible, or can you even devise a better way? (I would be happy to turn this into a collaborative project, if possible.)
Many thanks in advance for your thoughts!
Preston Guynn Thanks for your reply!
Yes, since these satellites are not equipped with batteries resp. accumulators but only with capacitors to bridge short periods of darkness (in the best case a complete night phase of about 45 minutes), the solar cells pointing away from the sun, and the axis of rotation (if any) being parallel to the sun-satellite line would constitute an incurable case because the microcontroller would never even be powered up. However, I assume that all satellites are stored in the "mothership" in the same orientation regarding the solar cells, and that before releasing the satellites the attitude of this carrier will be adjusted in a way which avoids this worst case.
It would be advantageous to be able to determine the horizon even during the night phases, so I think the IR domain is preferable to visible light. The sensor I chose as a first step can take up to 1 k samples/s, so with rotational speed of about 1 turn/s there should be even some margin left. When the sensor is saturated - as it certainly is when pointing to the sun - it will be inoperative for about 40 ms afterwards but with low rotation speed there should be still plenty of time left to the next horizon detection.
Yes, I think the gyroscope will be helpful in measuring the roll speed but determining the yaw speed of about 1 turn per 90 minutes will completely depend on the horizon detection.
Hello Holger, thanks for your reply!
Unfortunately, the satellites are just flat PCBs, about 5 mm in height, so we are far from such advantageous shapes as a dodekaeder. There will be certainly a base station for the CubeSat (the mothership) but at the moment I'm not sure how often the sprite satellites are permitted to contact it, too. (There might be more than 100 of them, at the outset in close proximity, so I guess when contacting base stations we would need some kind of semaphor mechanism to avoid collision of messages, and to ensure fair allocation of communication time slots.)
So, basically I assume that the sprite satellite has to rely on its own sensor signals. This way, I guess it will not be able to determine its geographical position but it should still be able to know the position of the sun and of earth, relative to its own coordinate system.
You are certainly right about the problem above the polar caps but the current plan is to release the satellites into an equatorial orbit, and I expect the additional velocity gathered during release to be so low that the individual orbit does not much deviate from an ideal equatorial course.
With an equatorial orbit, the attitude control could run into a deadlock situation if the geomagnetic field were independent of longitude: If the magnetic field were perpendicular oriented to the plane of the coil, and if the axis of rotation (if any) were parallel to the field, the total force on the coil would be zero. Fortunately, the geomagnetic field changes its orientation considerably during one circumnavigation (please see the attached graph which relies on the World Magnetic Model, basic software provided by NOAA). So, in the worst case the attitude control has to wait for a more suitable orientation of the field.
The choice of microcontroller in the current campaign might pose a problem: Possibly an 8 bit controller (ATmega328), clocked at 4 MHz, is too slow for the control task. I would be more at ease with something like a Cortex core; but my simulation hasn't yet come beyond the omniscient controller (no sensors, no limits to precision (other than IEEE 754 floating point numbers) so far).
BTW, I tried to estimate if it were feasible to use the inhomogeneity of the geomagnetic field for propulsion (That's the reason for the d(east)/d(long) curve in the graph.) but that's no way to go: To increase the speed by just 1 m/s would take weeks. (It might be a viable method when orbiting white dwarfs or neutron stars. ;-)
Best regards,
Joerg
Hello Holger,
your idea to store energy during a certain period, and then to release it during a considerably shorter time opens up the opportunity to combine the steady energy input from the sun with the most advantageous configurations of the geomagnetic field. That's great! According to the current plan, the trace forming the coil is 10 mil wide, and the current is about 5 mA. Since a trace of this width can easily carry 500 mA, one could choose to interact with the best 1% of the field. I have to think about the details, unfortunately about additional volume and weight, too.
For a start impulse, I think there are two possibilities: The most convenient, of course, is a well designed release method from a well oriented mothership. The second one which comes to mind is a small, exclusively mechanical device near an edge of the top layer (opposite the solar panel) that, if illuminated by the sun for several minutes, ejects a small object like a ball, e. g. by relaxing a spring. That should produce enough repulsion to start the PCB rotating.
To have two PCBs connected at one edge, and unfold them after release would be fine but ... the first project of this kind, initiated by Zac Manchester, was very restrictive (the hardware must not be changed, the software perhaps if Zac understands it and agrees); the current project I'm aiming to participate in is fortunately more liberal (choose your own sensor, write your own program). My 4-layer-PCB is already a step across the given line; I don't expect to be able to persuade Martin Platt, the initiator, to agree to unfolding PCBs. And a real problem (with the given dimensions for the PCBs) were: Having solar panels and electronic devices on both PCBs would about double the total thickness; so it would occupy two slots.
Best regards,
Joerg
heya my team are making blimp drones for use around 40km height. Perhaps we could share some ideas? msg me...
Holger Tuerk
Hello Holger,I'm sorry for a month of silence; I ran into a narrow spot with deadlines while at the same time the campaign I want to take part in seems to look less attractive because maybe the PCBs will be not permitted to fly independently but have to remain tethered somehow to the CubeSat. This would limited the freedom of rotation to one axis, at best.
Many thanks to your suggestion re spring loaded solar cells, super caps, and coil.
I'll compare the Murata super cap to others. Among other tests during the preparation phase, the PCBs will be subjected to strong vibration (I seem to remember something along the lines that the PCB will be attached to a piece of plywood which is connected to an impact drill). For this reason I'd prefer PCB traces to coils wound from wire; while one can fix wires by glue resp. resin I feel that traces are the most robust solution.
Stephen Sheppard Hello Stephen,
I'm sorry for my severely delayed answer!
This whole space business is more of a romantic hobby for me than a serious occupation, so I still have a lot to learn. For example, I have no clue regarding driving and attitude control of blimps at this height.
Could you please point me to a good source of information (article or web site)? Many thanks in advance!
Dear Joerg Fricke ,
these femto-satellites are well known by the community.
As you may have understood the Atittude Control and Determination Subsystem (ADCS) is one of the most challenging subs of a spacecraft.
With these chips-satellite you cannot do much.
You are right about copper coils inserted in the pcb (magnetotorquers or mtq), this is a really interesting option to control your body in space.
However here some challanges that you will face:
- 1 coil can generate a magnetic field on one axis of the body.
- in order to control properly in 3D you need 3 coils, one per body axis.
- there are tons of publications around magnetic control for satellite. I invite you to check the Bdot control strategy (a simple law to despin or spin a satellite). Conference Paper Testing and validation of a Ḃ algorithm for cubesat satellites
However you need 3 mtqs.- Soon you will realize that any kind of magnetic control cannot be reliable since you cannot apply any torque around the vector of the local magnetic field (for instance you cannot have any 3axes control on the body.)
- any type of "variation" of the Bdot law that uses 1 or 2 mtqs instead of three will help the satllite to align its spinning on one or two axes.
- any magnetic control that has a closed loop with a determination that uses magnetometers is extremely difficult. The paradox is that the bdot law needs a magnetometer to collect the magnetic field. However if you turn on the magnetotorquer you generate a magnetic field around the body that will peturbate your measurements. Leading to faulty determination. You will need to deploy your magnetometer or having a control loop in which the mtq stops working while the magnetometer measures (decreasing the performances).
- if you simply turn on the mtq in one direction, soon,will start spinning following the local magnetic field. (see this picture to have an idea) http://ssl.engineering.uky.edu/files/2013/07/Passive-Magnetic-Stabilization.png
- There are much more interesting control laws that can use magnetic control to point the sun (the spinning axis will point the Sun) but you need again 3mtqs and a sun sensor (or using the solar panel as sun sensor)Conference Paper A simple sun-pointing magnetic controller for satellites in ...
with only one mtq, you are very limited. My suggestion is to realize a passive magnetic controller, or modify the bdot law for 1 axis mtq . (for instance, if you have a mtq on the Z-axis you will dump the X and Y axis leading the spacecraft rotating around the Z axis. This with the time will align its spinning axis around the orbital angular momentum).
Have fun!
Dear Stefano Rossi ,
thank you very much for your profound and detailed input! I guess I will need (parts of) some days to process the information behind your links.
Of course, having only one coil is a severe restriction but at the same time an interesting challenge. Attitude control in an equatorial orbit would be an impossible task if the direction of the geomagnetic field were independent of longitude but luckily it's not. My first approach is kind of fuzzy:
Determine the actual axis of rotation (with reference to earth or the sun). If there is a considerable difference between the actual axis and a reference axis, measure the geomagnetic field (mtq switched off, of course), simulate the change of the axis caused by a current (of constant amount but selectable polarity) in the coil during a time slice. If the result is desirable, switch the current on, if not, let the timeslice pass with current switched off. Since the geomagnetic field changes its direction with reference to earth along the orbit, it should be possible to approach the reference axis stepwise. Computation time (on a microcontroller!) for the simulation might pose a problem.
I will come back after following your links!
Well if it is an equator orbit, the magnetic field vector should be aligned with the angular momentum of the orbit (or close to it depending on the orbit inclination). Therefore, tunring on the mtq at full capacity and full time will align the mtq axis to orbit angular momentum.
Basically: at the end your mtq axis will point to "south" or "north" depending on the verse of the current you put in the coil.
Good luck!
- Badges
- Science topic
- Similar topics
- Mathematics