KSP Community CubeSat

[Mazon Del]

MBobrik: The information I was given from Luis about the size of the moss capsules (which can be custom made somewhat arbitrarily small) indicates that given our low data rate for data transfer, we are probably more limited by how many images we can send down than we are by the amount of moss we can carry. I believe that a given capsule might be 1.5cm by 1.5cm, maybe with a depth of 1.5cm. In a ring formation we can fit quite a few in if we want. We definitely want at least 3, ideally in multiples of 3. This is because we can breed moss that grows away from gravity (up, or in this case inward), toward gravity (down, or outward), or perpendicular to gravity (sideways, or spinward/antispinward/left/right). For every multiple of 3, we can have redundant samples of a given type.

Newt: You misunderstand the purpose of the ring. The ring is not to act as the centrifuge itself. The ring stays still relative to the cubesat unless we need to point the cameras at a different moss capsule. So if it is just sitting in a growth cycle and nothing is happening, the ring is spinning at zero RPM relative to the cubesat because the servo is off. (Meaning if the cubesat is moving at 5 RPM, then so is the ring). But now we need to take a picture of a given sample, and it isn't the one the cameras are looking at right now. So the ring rotates to move the samples in front of the cameras. Now, you do bring up that this will change the rate at which the sat is rotating depending on which direction we turn. But this is not a big deal because of equal and opposites. Example: For simplicity of the example, we will declare the ring to weigh the same as the rest of the cubesat. We need to see a sample 180 degrees away, so we start the ring spinning at 1RPM clockwise. Because the two weigh the same, this means that the ring will be spinning at 0.5RPM clockwise, and the sat is now spinning faster/slower by 0.5RPM counterclockwise. But now 30 seconds later the moss is in the right position, so we need to stop the ring. We do this by applying the opposite force. So this brings the ring to 0RPM clockwise and the sat is now faster/slower by 0RPM counterclockwise.

This is important because it means any time that we are spinning the sample ring, the cubesat itself will be spinning faster/slower depending on the direction of the ring-spin. Therefor the system that determines our RPM is going to have to realize that this is happening and either ignore it or fight it. Someone should do some calculations on this, because I'm struggling to do it in my head here at work, but other than the jarring from the servo accelerating up and down, will the force on the moss be constant? Instinctively it feels like after the cubesat is at the RPM for say 0.1Gs, then as long as the servo is rotating the ring at a constant speed, the forces should equal out, causing the moss to STILL feel only 0.1Gs. But this instinct also has the feeling of being wrong about that.

Regardless, the actual changes in gravity will be rather minute and short-lived, so if there IS an imbalance across the time the ring is rotating position, then we likely don't care to burn the energy needed in the torquers to keep the gravity constant. But maybe if we are efficient enough with power then we do?

[Newt]

It is essential to look at how these things will all work together, though. Mr0nak raises a good point in that we should be well, a little more down to earth. All of the parts will interact, and if we at least had a basic sketch of what the hypothetical satellite looks like, how big everything should be inside of it, and a ballpark guess of what it will mass, that would keep things a bit more in perspective, and keep us from wandering too far from possibility. Maybe it can even give us some ideas?

I would be willing to try to put together a model of all of this, but really do not know if there is enough decided on at this point. Perhaps we should try to get that sorted out, so if we are too big, we figure that out sooner rather than later. The document linked at page 1 has a list of parts more or less, but does not get into size, really (yet).

EDIT: Mazon, I was addressing the proposal that the ring itself spin, which seems mostly to have been shot down now. I had the page open for a while, and got somewhat lost in what was being said while I was not pressing f5. Still trying to sort out all the order....

Regarding gforce changes during camera spin, it was said that we would probably be spinning very fast (more than 50RPM). At .05m radius, at 50RPM, we should have about 1.37ms^-2, at 51RPM, 1.426ms^-2. Not too much difference, but worth noting. Also, as has been mentioned, we can take our time taking pictures and swiveling the camera, so 1RPM might be excessive. If you are refering to just the speed increase for the ring, my understanding is that the ring will be affected by the satellite rotation. But again, we are usually in no hurry to orient.

One issue I just thought of was that of our Center of Mass. If that is not spot on, we might get serious gravity differences from what we had hoped for.

Edited December 8, 2014 by Newt

[deljr15]

Beryllium costs cca $748 per 100g. we need 13 g of it. If necessary we can make a wire from it by ourselves.

I know this is a old post but I feel it should be brought up. Beryllium is very health hazardous to work with. Most manufacturing facilities that work it have very specialized ventilation systems. One shop near me has all the lathes sealed in air tight chamers and have all employees wearing hazmat suits

A quick google on beryllium heath effects turned up http://www.ccohs.ca/oshanswers/diseases/beryllium.html

[deljr15]

Also a few pages back K^2 mentioned bearing lubrication problems. This is simply avoidable by using brass bushings. Brass being self lubricating makes it a excellent fit for this use and is found in many things you use in your daily life.

[deljr15]

I would be willing to try to put together a model of all of this, but really do not know if there is enough decided on at this point. Perhaps we should try to get that sorted out, so if we are too big, we figure that out sooner rather than later. The document linked at page 1 has a list of parts more or less, but does not get into size, really (yet).

I completely agree. I have a file server dedicated to this project. Message me for login info. I have some frames uploaded to it at the moment. I will admit to slacking on the generating of other models, but with everything still up in the air. Many of the items listed in the doc (cameras at least) are not feasible. As we come to more of a agreement on things I will start adding more. The rotating mechanism for the samples will be something I will start to take a look at. Will post some ideas when done.

[MBobrik]

A quick google on beryllium heath effects turned up http://www.ccohs.ca/oshanswers/diseases/beryllium.html

Only inhalling beryllium dust is dangerous. Wire drawing does not involve any machining that produces dust.

Besides, beryllium outperforms alluminium only by 11 %, and after we dropped the idea of quickly turning to/from ground tracking, we don't need extra performance that much. We can just go with aluminium.

Many of the items listed in the doc (cameras at least) are not feasible. As we come to more of a agreement on things I will start adding more.

That doc is hopelessly outdated. There have been major design decisions made.

Camera and CPU heve been already choosen -

We go with dual CPU

- flight CPU: rad hard 8051 + 1Mb MRAM (weight unknown, but most probably < 10g )

- media CPU: dedicated to camera handling. Raspbery Pi compute module, weight 7 g

Camera will be

2 x Raspbery Pi camera module, weight 2x3 g

.

Not saying I have a better idea, I'm just pointing out that maybe working within a given space budget and then see whether you actually can fit more than a single experiment in is more fruitful than the other way around.

Having a mass/volume budget and adding all parts to it is definitely a good idea. And we should, as default value estimates for all systems/parts that haven't been choosen yet, add the weight/volume of the corresponding cubesat shop parts.

And then replace them with our values once we specify/design that part by ourselves.

This evening I'll create another spreadsheet for this purpose.

[MBobrik]

Took a little longer than expected, had other things to do, but the sheet with currently specified components is here. All major components that are not specified are filled with cubesat shop placeholders.

@K^2 could you add this link and the torquer calc sheet to the main page ?

[Nicholander]

A Google Docs component list sheet, great! Just a suggestion, but maybe somewhere where it added up all the costs on the sheet would be good, just saying.

[MBobrik]

A Google Docs component list sheet, great! Just a suggestion, but maybe somewhere where it added up all the costs on the sheet would be good, just saying.

Cost, Mass were already in. Dimensions will come.

[Nicholander]

I mean like ALL the costs of all the things added up to give a total cost of the entire CubeSat, or maybe I'm just not seeing it. Where do you say it's on the sheet?

[MBobrik]

Cost is column E. I wrote numbers only where I knew the price. Which is mostly the cubesat shop placeholders, because except media CPU and camera nothing was specified in detail. At this stage, it makes no sense to make a sum of them.

[deljr15]

So I have been kicking around some ideas for the sample indexing mechanism. Instead of going at this totally blind. I want to get some feedback on mass limitations. Obviously we want it to be as light as possible, but what is the upper limit?

Edited December 11, 2014 by deljr15
. to ?

[Newt]

Well, the NASA CubeSat Initiative page (linked in the OP) says that they weigh ''about three pounds''. That concurs with mass limitations on pages linked by that page (the fact sheets): No more than 1.3kg per 1U cubesat.

Of course, if we want to reduce that to less than that, it is up to us. But we are not the ones launching the thing; we get no more control over our orbit by dropping mass.

Many thanks for the table, MBobrik, it seems to put this in a lot better perspective for me.

And, Nicholander, so far that is about 91,330USD. Bear in mind that that fails to account for many parts, and that at this stage many of the costs are as far as I know, conjecture. I have heard that a common lower threshold cost for one of these is about 250,000USD.

Edited December 11, 2014 by Newt
Wording and Grammar changes, new section

[deljr15]

I am looking for the upper limit we are willing to budget to the experiment indexing mechanism.

[MBobrik]

And, Nicholander, so far that is about 91,330USD. Bear in mind that that fails to account for many parts, and that at this stage many of the costs are as far as I know, conjecture.

As a mater of fact, as we specify and design our own parts instead of the insanely expensive placeholders, it will get cheaper. For example the ground station is basically a glorified ham radio. If we gave 10 % of the cubesat shop price tag we would overpay it. The same goes for structure, antenna, power supply and batteries. We will most probably have to pay up for solar panels, and if we don't manage to find someone to design a custom transceiver for us, then the transceiver too.

Edited December 11, 2014 by MBobrik

[MBobrik]

I am looking for the upper limit we are willing to budget to the experiment indexing mechanism.

What do you mean with "experiment indexing mechanism" ? you mean, how the moss samples will be packed and how they will be moved into the camera view ? I imagine a hollow cylinder with 90 mm outer diameter, 73 mm inner diameter, 15 compartments, and 1mm thick acrylic walls. Servos and cogs that rotate it will be inside. The problem with thickness is, that if it has to be 15 mm thick, the camera depth of field has to be 15 mm too. and a quick calculation shows, that the viewing distance has to be 150 mm. so we would have to bounce the image once through a mirror to get the distance from the camera. If we don't want this, then we will get maximum cca 80 mm view distance, and thus only 5 mm depth of field, so the sample should not be thicker. and thus question @ Mazon Del. Can the samples be cca only 15mm x 15mm x 5mm ?

[MrOnak]

@MBobrik: That list will definitely help once more blanks get filled. I realize that it's a bit moot to fret about off-the-shelf component prices until you all know what you will buy off the shelf and what you will / have to make yourself but the prices in the list are definitely on the high side (not that that's a bad thing in terms of budgeting).

Cubesatkits.com has an extensive pricelist available. A 1U kit complete with structure, solar panels and 8051 CPU costs 7500 USD for example. Those who are interested in the details should read the fineprint.

Oh and since this is the second time I'm mentioning Cubesatkits: I'm not affiliated with them nor do I get anything out of it by mentioning them.

[MBobrik]

Cubesatkits.com has an extensive pricelist available. A 1U kit complete with structure, solar panels and 8051 CPU costs 7500 USD for example. Those who are interested in the details should read the fineprint.

The red "placeholder" values in the sheet are just the first what I found. I's just a starting point, a placeholder put there so that the field is not blank. The sheet is editable. Anyone can, and should, if he find something cheaper, or more suitable, just replace the part.

[NERVAfan]

have heard that a common lower threshold cost for one of these is about 250,000USD.

Well, other Kickstarted cubesats have asked for much less than that. They might have had other funding sources... I really don't think it needs to be nearly that much though.

[MrOnak]

All hardware except the experiment will cost you around $50,000 US if you're going off-the-shelf with most bits. Less if you handle your own PCB design and soldering.

Everything involving the experiment is on top, as is work-time, logistic and eventually the launch cost.

[MBobrik]

Well, other Kickstarted cubesats have asked for much less than that. They might have had other funding sources... I really don't think it needs to be nearly that much though.

As I said above. Most of the cost comes from the PLACEHOLDER parts, which we most probably will replace with regular, more than order of magnitude cheaper components. Even if we have to buy the solar panels, there is a chance we need not that many of them. I believe K^2 said he will calculate the total power needed for heating, but he seems to be occupied. My own back of the envelope calculations show we need only 3 watt max if we want to heat the entire cube, which translates into 6 watt of solar panels.

I personally think that we can send the bird flying for cca $20 000 + launch cost.

[technion]

Raspbery Pi compute module, weight 7 g

Are there thoughts on how you would power that? I've spent a lot of time on battery power for my RPi and I get less than a day out of this setup.

Switching power supply, downsteps to 5V. Way over built to provide smoothest most reliable power.

UPS with battery. The battery on its own wouldn't fit in a cubesat.

[K^2]

Yup, sorry about the delay with heating stuff. Some imminent deadlines at work, and I've been catching up on sleep today. Tomorrow's looking good, though.

In terms of panels, the only way to get 6W is to have 3 10x10 facing the Sun at the same time. That already calls for unfolding panels. If so, I'd feel better having two pairs of unfolding faces which work independently. That maintains stability of the sat even if one pair doesn't deploy, gives us extra power for comms if all goes fine, and gives us some redundancy if it doesn't.

Edit: Re: Battery. We can get 4.5 - 5V from panels. That's enough to charge a LiPo cell. Two cells gives us 7.4 - 8.2V to power the sat. I can provide stable 5V from that easily, and I could probably find a way to not waste the difference.

Edited December 14, 2014 by K^2

[MBobrik]

Oh, welcome back!

In terms of panels, the only way to get 6W is to have 3 10x10 facing the Sun at the same time. That already calls for unfolding panels. If so, I'd feel better having two pairs of unfolding faces which work independently. That maintains stability of the sat even if one pair doesn't deploy, gives us extra power for comms if all goes fine, and gives us some redundancy if it doesn't.

Well, I assumed the entire sat will be heated to 20 °C, and the surface reflectivity will be that of aluminium. If we heat and insulate only the samples and batteries, and keep the rest of the sat at say -10 °C heating requirements will drop to less than 2 W. And maximum cube projection area is, I believe sqrt(3) * one side, which would give us 3.5 watt with 3 non-foldable panels. But that would require the sat to rotate around its diagonal, which is awkward at best. And is a waste of panel efficiency. So I too think that the unfolding the panels will be unavoidable. In the calc sheet I thus counted 4 foldable panels and 1 fixed giving us 11 Watt total.

And when I am already mentioning the google sheets, could you please add

- the toquer calc sheet and https://docs.google.com/spreadsheets/d/13vQUH8vK43XcgI-o--lxWRhwMzHNJw0X-DgyFmitKKs/edit?usp=sharing

- the part weigt/volume/cost sheet https://docs.google.com/spreadsheets/d/1rU0J4dMLqgI2b2MDWhZML-ffgUhhBwcTbRDssW3zcZc/edit?usp=sharing

to the first post, so that they don't get lost ? And while adding links,

- a comfortable and free 3D modeling tool http://www.sketchup.com/products/sketchup-make

It's not a professional CAD, but still better than drawing in MS Paint .I think we should use it to visualize our ideas, and even for the final design.

edit: fixed swapped links

Edited December 14, 2014 by MBobrik

[Nicholander]

- a comfortable and free 3D modeling tool http://www.sketchup.com/products/sketchup-make

Or, for a more professional (And complicated) option, SolidWorks.