KSP Community CubeSat

[Error404brain]

We can still stick with the .5 U idea. We will just test how the smarthphone cpu react when protected (that's quite important for next cubesat launchs). When I am back at home, I will look how we can protect the cpu and if we can do it with an inflatable module or not. We need one face transparent for awesome photo, the opposite to the sun will do the job (and the solar panel on the other side). We should try 3 modified smarthpone so for to have significant data. If that work as intended (same as a spacegrade CPU), we can try other positions and even, maybe expose the glass/plexiglass/acrylic panel to the sun.

If that work it would be awesome enough to get founding for wathever cubesat we would launch next. if that don't, too bad, we tried.

For the plants. Don't even think about it. None of the launchers will accept to launch living biomass in space. Even less an entire ecosystem.

IF we can get a launcher who accept biomass and IF we can actually build it, let's go for mars soil. Nasa can give that to us.

[Aethon]

Also, I was slightly wrong on the cost at WPI. It averages $7,500 per student. But one student is still committed to 21+ weeks of work (estimate about 30-50 hours per week depending on how into it the student is) plus a faculty advisor.

"Dong!?? Grandpa is talking to you!" - Sixteen candles.

<em>

Insert CubeSat for 'Auootomobeeeel'.

Edited July 19, 2014 by Aethon

[Aethon]

Kickstarter start NOW!

Tap tap.

https://docs.google.com/spreadsheets...it?usp=sharing

[Aethon]

http://www.nasa.gov/ames/astronauts-to-test-free-flying-housekeeper-robots/index.html#.U8qZkPldUTo

[K^2]

It's not serious to try and ask people how much they can contribute until there is a concrete plan. Calm down. All in good time.

General RFC. I'm putting together a shopping list. In principle, if we go with off-the-shelf CPU/MCU, we can go pretty fancy. But I think we are much better off by relying on architecture where any part can be swapped out to rad hard if funds become available.

With that in mind, I propose using an 8051 CPU. It's easy to work with. There is a huge amount of tools available. It will make floating-point computations a bit tricky, but honestly, everything we need can be done in fixed point. It also has very low clock rate, so we wouldn't be streaming any video. (Not that I was counting on that.) It will be plenty for navs and attitude control work, and it will be able to handle communications with ground stations and transmit images.

Best part, the off-the-shelf 8051 costs $10. So we can burn through a stack of them during development and testing without costing us. On the other hand, I've found rad-hard version of the same chip (UT69RH051) for $1,275. Expensive, but it's the sort of funds we might even find at the last possible moment, and still be able to swap out the off-the-shelf 8051 for a rad-hard one.

I'm still looking at memory and other components. But it looks like I'll be able to make an absurdly affordable base package which we will be able to expand into something far more reliable with a very reasonable stretch goal.

[christok]

Tap tap.

Patience. We're all still thinking over the numbers. Edit: K^2 is a ninja.

[Aethon]

Just keepin' it fresh.

[Rockhem]

We can't expect to have funds fall from the sky on our kickstarter, we have to aim low to see if we can pull it off, then expand to greater things. We should aim for a 1U cubesat in LEO with nothing more than an instrument or 2 and probably a toothpick KSP flag, We cant expect to raise hundreds of thousands of dollars, we should aim low and if we get more money than we thought we would we could use the larger missions as a stretch goal instead of trying to aim for complicated missions when we don't even have a concrete plan and kickstarter money doesn't from the sky. We just need to stick a box with a toothpick flag and some batteries in LEO, not a complicated mission with ion engines and the like. Maybe a small camera that might even be B&W.

[Rockhem]

It's not serious to try and ask people how much they can contribute until there is a concrete plan. Calm down. All in good time.

General RFC. I'm putting together a shopping list. In principle, if we go with off-the-shelf CPU/MCU, we can go pretty fancy. But I think we are much better off by relying on architecture where any part can be swapped out to rad hard if funds become available.

With that in mind, I propose using an 8051 CPU. It's easy to work with. There is a huge amount of tools available. It will make floating-point computations a bit tricky, but honestly, everything we need can be done in fixed point. It also has very low clock rate, so we wouldn't be streaming any video. (Not that I was counting on that.) It will be plenty for navs and attitude control work, and it will be able to handle communications with ground stations and transmit images.

Best part, the off-the-shelf 8051 costs $10. So we can burn through a stack of them during development and testing without costing us. On the other hand, I've found rad-hard version of the same chip (UT69RH051) for $1,275. Expensive, but it's the sort of funds we might even find at the last possible moment, and still be able to swap out the off-the-shelf 8051 for a rad-hard one.

I'm still looking at memory and other components. But it looks like I'll be able to make an absurdly affordable base package which we will be able to expand into something far more reliable with a very reasonable stretch goal.

How long would our non-hardened equipment survive in LEO, if not very long, we would need to make getting rad-hardening a priority.

[K^2]

How long would our non-hardened equipment survive in LEO, if not very long, we would need to make getting rad-hardening a priority.

Impossible to say. Expectation is on the order of weeks, but it could be much less or much more. Depends on solar activity, magnetosphere dynamics, and pure dumb luck. There are a ton of variations on 8051. I can see if some are going to be better than others at surviving radiation. But only rad-hard CPU will survive for a few months relatively reliably. (A good solar flare can knock out even one of these, so there is still an element of luck.) If no electronics fail, with ISS launch we might be able to count on about a year before orbit decays.

So I agree that CPU should be a priority. And as I think I've mentioned earlier, that would be the #1 entry on the upgrade list. But if we have a ride and all other components, I don't think we should scrub the mission just because we can't make that final push. And a more manageable base goal on a KickStarter goes a long way towards getting funded, and having a chance to get extra funding before the launch.

The other factor is that, like I said, 8051 is easy to obtain, cheap, and has a lot of tools and software developed for. A big part of why I want to go with that CPU is so that we can start prototyping the board and testing things without spending money on expensive equipment and risk ruining it.

[christok]

General RFC. I'm putting together a shopping list. In principle, if we go with off-the-shelf CPU/MCU, we can go pretty fancy. But I think we are much better off by relying on architecture where any part can be swapped out to rad hard if funds become available.

Remember that we must have at least two complete and identical satellites as per testing requirements. I'd like to plan for three because otherwise we have no backup if something were to happen in construction. The test satellite(s) can be used as control(s) and then handed to some Kickstarter backer(s) with too much money on their hands.

Those rad-hard parts could get expensive quickly.

With that in mind, I propose using an 8051 CPU. It's easy to work with. There is a huge amount of tools available. It will make floating-point computations a bit tricky, but honestly, everything we need can be done in fixed point. It also has very low clock rate, so we wouldn't be streaming any video. (Not that I was counting on that.) It will be plenty for navs and attitude control work, and it will be able to handle communications with ground stations and transmit images.

I'm willing to go with your choice of CPU. Anything we pick should be something with free tools so that we can share our work. I understand there already exist community-created cubesat boards with freely available Gerber files. It may be prudent to use or modify something that is flight-proven, even if it isn't our first choice of CPU.

For the plants. Don't even think about it. None of the launchers will accept to launch living biomass in space. Even less an entire ecosystem.

IF we can get a launcher who accept biomass and IF we can actually build it, let's go for mars soil. Nasa can give that to us.

I found nothing in the cubesat specs prohibiting biological payloads. It will complicate thermal bakeout but I don't see that much of an issue. We will naturally start discussing this with Cal Poly and potential suppliers when we have a better idea of what we want. I'm against simulated regolith. An experiment can't study too many variables at once.

Impossible to say. Expectation is on the order of weeks, but it could be much less or much more. Depends on solar activity, magnetosphere dynamics, and pure dumb luck.

In DIY satellite platforms: Building a space-ready general base picosatellite for any mission (2012. O'Reilly. ISBN 978-1-449-31060-8. p. 9), Sandy Antunes claims that for his project in the making (using a BasicX-24 computer), of less than 3 months projected flight time, radiation damage isn't a major concern.

Although single event upsets are likely to scramble the sensors and crash the computer from time to time, I'm perfectly fine with not having 100% uptime in exchange for two orders of magnitude cost reduction if we can get away with it. Is your "weeks" expectation an educated guess, or do we perhaps have data for 8051 survival in LEO/something similar?

[Kryten]

It should probably be pointed out that the guys responsible for deploying sats from the ISS ask that you contact them before starting any kickstarter.

[K^2]

Educated guess. And yes, random restarts are fine. Like I said, we can even add some extra checks to force a restart if something looks fishy. Maybe have canaries in RAM for the simplest check. ECCs for more important stuff.

I did not realize that we need to submit a copy for testing. Does it have to have all functional components? That sounds excessively and unnecessarily expensive. We can build multiple sats, but it feels wrong to waste money on functional, space-hard photovoltaics and ICs. I plan to have off-the-shelf equivalents for every component. I can even come up with photovoltaics that are operational in rad-safe conditions. Will that do for testing? That's an order of magnitude in costs of the test sat, at least.

Could you point me to the specific document that requests second sat for testing? Maybe I can find some loopholes.

It may be prudent to use or modify something that is flight-proven, even if it isn't our first choice of CPU.

8051s have been used. In fact, I think some of the commercial CubeSat CPU/RAM boards are rad-hard 8051-based. But I'd rather build a custom board for this. We can build several copies and even several iterations of the board and do all sorts of testing with it to make sure it's good.

There are tools for doing some virtual testing of the board during design stages, there are good emulators for 8051, and we can outsource printing the actual board. Maybe even have basic components and sockets machine-soldered with something that will have good thermal ranges for the final series of builds. (I don't mind spending a few hours with a soldering iron with the first prototypes.)

It should probably be pointed out that the guys responsible for deploying sats from the ISS ask that you contact them before starting any kickstarter.

We'd talk to a number of potential launch opportunities before putting KickStarter up, and run the actual page by the top choice for sure. We have to know in advance all the options and what we can promise.

And it'd still be nice to try and find a free ride from NASA or someone else.

Edited July 19, 2014 by K^2

[indroth]

Weeks sounds right based on this:

http://www.nasa.gov/centers/ames/engineering/projects/phonesat2.html#.U8rc8_ldWyU

“We expect PhoneSat 2.5's orbital lifetime to be as long as six weeks"

[Rakaydos]

Best part, the off-the-shelf 8051 costs $10. So we can burn through a stack of them during development and testing without costing us. On the other hand, I've found rad-hard version of the same chip (UT69RH051) for $1,275. Expensive, but it's the sort of funds we might even find at the last possible moment, and still be able to swap out the off-the-shelf 8051 for a rad-hard one.

l

For the rad hard version, do you think that would be viable for a 1 year or 3 year mission, going solo? or would the solar orbit maneuver and phobos missions need to pack redundant systems?

[Nicholander]

I think, if what christok said is true, we should only have some Java Moss and some bacteria, because it would be intresting to see how bacteria interacts with other biological matter in microgravity. (And also other gravities, because it will create artificial gravity by spinning)

[Sax Man Aeronautics]

To add to the cubesat, we should put a colored Titanium model of the Mk1-2 Command Pod with Bill, Jeb, and Bob inside with their facial reactions.

[dharak1]

This is starting to Co, e together pretty good. I never would have expected to use a 10$ cpu on a space mission. What type of battery and capacity would we want? I suppose we need to get components together first to check power consumption but a general estimate wouldn't hurt.

[Three1415]

After expending a ridiculous amount of time reading all sixty-two pages of this thread, I must say that this is quite an interesting project; I would most certainly support it.

I was sufficiently intrigued to do some minor number-crunching in terms of temperature management on the night side of the planet, and I wondered as to the feasibility of using a water-based heatsink. The requirements for such are minimal; assuming the CubeSat's is kept at a temperature of 25 degrees Celsius upon passing into the night side, has an aluminum exterior, and follows the roughly ninety-minute orbital cycle in LEO, only about 62 grams of water heated almost to boiling while in sunlight would be able to stave off heat loss entirely, while reducing cooling requirements. This is also the absolute worst-case scenario for the night side; I used the maximum values for aluminum emissivity and a probably unreasonably-low surface temperature for the cube. This method of temperature regulation could easily be more weight- and volume-efficient than battery-based systems.

[K^2]

While standing by, it won't draw more power than an old flip phone, and the battery only needs to last for 45 minutes between charges. A pair of 100mAh+ LiPo cells should do just fine.

Considerably more power will be drawn during attitude adjustments or while transmitting. But these operations can be carried out on the day side. Or we can add a bit more batteries to allow for limited night-side operations.

For the rad hard version, do you think that would be viable for a 1 year or 3 year mission, going solo? or would the solar orbit maneuver and phobos missions need to pack redundant systems?

Anything that leaves Earth system might as well have a backup just on principle. Given costs of propulsion and solar array required, backup computer isn't going to break your budget.

[dharak1]

Will such a low cost processor be able to support 2 Cameras? Will we have to have some sort of secondary chip just for handling image processing?

[K^2]

We'd need a camera with a buffer. Once it has taken a still, we can process and beam it at a leisurely rate.

[Rakaydos]

We'd need a camera with a buffer. Once it has taken a still, we can process and beam it at a leisurely rate.

Not too leisurly, though, unless we rad harden the buffer as well.

[K^2]

Hm. If camera can produce a JPEG (something like this), I can probably get it to about 50-100kbps over an 8051, streaming Huffman codes with reset points. The total size will be a bit larger than original JPEG because of the reset codes (by a few %), but if a block gets lost due to noise or any other problems, it should be just a block that's missing. Not the rest of the image. The loss of baud rate is due to having to do some processing on CPU to insert the reset codes, and 8051 only does 1MIPS. At typical camera JPEG quality, we're looking at 10-20s to beam the image down. That should be fine.

Of course, camera like that is going to buy it pretty fast, since it's not rad-hard. I'll see if there are rad-hard options out there that are reasonably priced and wouldn't require a dedicated processor just for that.

Alternatively, we can look at cameras that simply have addressable buffer on board, without any processing, but then we'll have to encode the image, and that will be considerably slower. JPEG might not be the way to go at all there.

Edited July 20, 2014 by K^2

[christok]

Could you point me to the specific document that requests second sat for testing? Maybe I can find some loopholes.

http://www.cubesat.org/images/developers/cds_rev13_final.pdf

Section 4.5.1:

Qualification testing is performed on an engineering unit hardware that is identical to the flight

model CubeSat. Qualification levels will be determined by the launch vehicle provider or P-POD

integrator. Both MIL-STD-1540 and LSP-REQ-317.01 are used as guides in determining testing

levels. The flight model will then be tested to Acceptance levels in a TestPOD then integrated

into the flight P-POD for a final acceptance/workmanship random vibration test. Additional

testing may be required if modifications or changes are made to the CubeSats after

qualification testing.

Edit: I may have misunderstood. It looks like you need to do either qualification testing or protoflight testing, not both.

Edited July 20, 2014 by christok