KSP Community CubeSat

[bounding star]

Yup. One of the main functions of the fligh computer will be looking after spin axis. This will be acomplished with a suit of sensors and magnetotorquers. We might be able to get additional info from media computer and its camera.

what sort of sensors are we looking at? what if it separates from the launcher at night? also, how will it separate?

[Mazon Del]

Alright, compiled what seems to be a quick list of the questions people had for me.

Henryasia: (PS Mazon Del, if you're reading this, I still think 1U will make a gravity gradient in the experiment big enough to mess with our data, what's your take on that?)

Yeah I've been wondering about it, but it seems mostly that as long as we have a spread of the moss across the enclosed area in question, that we should be fine. We are trying to see how extreme the gravitometric response is, so as long as its growing (even if in funky directions) then we can see what we need to. We can always do some analysis based on the mosses position to determine what their gradient was like.

Henryasia: On the other hand, Mason Del could definitely get some samples like, right now, film them in the dark under IR light at 30 FPS (or whatever an IR camera can do) for some days, then we'll speed up the video at different rates and see which one is acceptable for growth analysis. What do you say?

I highly doubt we need anything beyond an image per hour. I can check with Luis and see what he thinks, but I'd guess that the response is going to end up being that one image every 4-6 hours is more than enough. However, a point to consider here. If we are carrying up multiple moss capsules (they are only going to be a couple centimeters or so to a side), then 1 image every 6 hours could end up meaning 30+ images per 6 hours. Because we'd ideally want to be imaging each set of moss. And this doesn't take into account the possibility that we may desire a different camera angle.

Henryasia: PS: Dammit! SpinSat is taken! MossSat, anyone?

I vote we name it KossSat. K because Kerbal, oss because moss, Sat because satellite. I further vote then that our mascot should be a Kerbal wearing a traditional Cossack outfit.

K^2: So long as we are getting clean readings from all of the sensors, we should still be able to estimate biomass growth pretty reliably.

Caring about the biomass growth is only part of the data we want. We are caring about the gravitometric response of the moss (its tendancy to grow away from gravity [up], towards gravity [down], and perpendicular to gravity [sideways]) based on how they have been bred. We can only determine this based off of image data. The reason we care about the gravitometric response (to those just joining us) is that it gives is a good idea about how the plant cells react under the different gravities. If the plant reacts poorly to the 0.1Gs, but grows well in 0.3Gs, then this tells us that the optimum G load for the plants to be under when cultivated in space and when you don't want to go through the expense of a full 1G rotation system is somewhere above 0.1Gs. It will also provide us with information about the behavior of plants in such conditions, useful if you intend to grow plants in a rotating drum like this.

Extra:

So some thoughts I've been having about the internal camera system. Assuming we are going with several moss capsules, the camera IS going to have to move if we want close up shots (or if we want microscope shots depending on camera). So, given that, the idea that I have is that in the center of the growth chamber exactly on/aligned with the axis of rotation is a pole. This pole can rotate either 180 or 360 (depending on if we have one or two cameras). Extending out from this would be a bar that has the camera on its tip. We can control the rotation of the bar and its position along the axis of the pole.

The way I would implement this is for the pole to be a tube, and the tube has a 2 slits going down through it on opposite sides of the pole. Inside (but not touching the tube) is a threaded rod. One motor controls the rotation of the slit-rod the other controls the threaded rod. The arm for the camera/microscope has a nut at the center which has been threaded onto the rod. You control height by having the tube hold still and rotating the threaded rod (its like screwing/unscrewing something) if the tube holds still, this means that the threaded rod is rotating but the arm cannot, so the arm will raise or lower. Conversely, if you hold the rod still and you rotate the tube you get a rotation. This rotation WILL influence your height, but I've already explained how to adjust that. It is a compact design that isn't very complicated. Whatever happens to the camera will happen to the fake camera-like weight on the other end, or also to that camera (if we have two).

[MBobrik]

what sort of sensors are we looking at? what if it separates from the launcher at night? also, how will it separate?

unlocked GPS unit, magnetic field sensor, gyro, real time clock [if it counts as a sensor], thermometers, solar panels themselves can double as photodectors, and so can the media CPU's external camera function as a star (more like sun, earth and moon) tracker. I would guess, that cubesats are intentionally released always on the day side, but even if it were launched just after entering shadow, it would just run the 45 minutes on batteries like it will be doing on all next orbits. How it will be released ? they will be just pushed out at ~1m/s relative to the container on ISS.

Edited December 4, 2014 by MBobrik

[MBobrik]

So some thoughts I've been having about the internal camera system. Assuming we are going with several moss capsules, the camera IS going to have to move if we want close up shots (or if we want microscope shots depending on camera). So, given that, the idea that I have is that in the center of the growth chamber exactly on/aligned with the axis of rotation is a pole. This pole can rotate either 180 or 360 (depending on if we have one or two cameras). Extending out from this would be a bar that has the camera on its tip. We can control the rotation of the bar and its position along the axis of the pole.

This would require too much continuous free space for the camera to move around. I would suggest, we stick the sample capsules together like beads on a string, and then we would just move them around to the camera back via pulleys. Requires much less continuous free space.

[LordQ]

Well, I did a few calculations myself. but not about power usage. I tried to compute the maximum angular acceleration we can squeeze out of the magnetotorquer. And came to some very interesting numbers and very interesting ideas.

The BIG number first. If we use some very unorthodox solutions, we may get angular acceleration up to 50 RPM/min at 10 W power and the magnetotorquer coils weighting 5 % of the total mass of the spacecraft ( assuming that the sat got relative permeability ~1 ). Which means we would be able to change the spacecraft's axis of rotation within a few minutes, and thus be able to reorient the sat from its usual sun basking orientation, do high speed transmission via a a directional high gain antenna, and then turn back. Now comes unorthodox solutions part.

It would require beryllium wire, and the magnetotorquer would have to have a very unconventional shape - three orthogonal circular coils with diameter 2.25 times larger than the cube's side. Which means it would have to be folded, and expand to its full shape after satellite release. Either by spring action, or something more exotic like shape memory polymers.

Now the maths. I won't write the long tedious derivation here, just the resulting formulae.

AngularAcceleration = B * √(P) /√(Ic) * 1 / √(20*ÃÆ’*ÃÂ) * Q

where

Ic is the inertia momentum of the rest of the satellite without the magnetotorquer Ic = 1/6 Mc * L^2, where

Mc is mass of the cubesat without the torquer and

L is size of the cubesat

B is magnetic field

P is input power

ÃÆ’ is coil material electrical resistivity

ÃÂ coil material density

Q is a dimensionless constant that describes how well is he magnetotorquer matched with Ic of the rest of the satellite. It runs from 0 = useless to 1 maximum = perfect match.

Of course we are aiming at perfect match Q = 1

Q = 2 * √(q) / (1+q)

where the small q is the ratio of magnetotorquer's inertia momentum It to the inertia momentum Ic of the rest of the satellite q = It / Ic

We can see that Q reaches maximum 1 when q is also 1. Of course we want not that the torquer weights half of the satellite. Therefore we will have to make up with its diameter.

It turns out that the torquer coil diameter can be easily computed from the mass fraction of the torquer.

d/L = √( 4/15 * (1/f -1) )

where

d is torquer coil diameter

L is cubesat's size, same as above

f is the torquer mass ratio f = Mt / M

we say f = 5 % so we get that d = 2.25 times cubesat's size, which means it will have to be stored folded and unfold around the sat, but it's not as dramatic as if we needed it to be say fifty times sat''s diameter.

Now back to the original equation. We can see that in the 1 / √(20*ÃÆ’*ÃÂ) factor both resistivity and density of the coil play equal role, so we should not simply choose the least resistive material, but the one which has the least ÃÆ’*ÃÂ. After a short search, it turns out, that the best feasible material is beryllium, which beats copper with 157 % more acceleration, all other things being equal.

Now, let's put all the numbers together

B = 4e-5 T

Mc = 1Kg

L = 0.1 m

P = 10 W

ÃÆ’ = 3.6e-8 Ωm

ÃÂ = 1690 kg/m^3

Q = 1

we get AngularAcceleration = 0.014 rad/s^2 = 51 RPM/min

.

Now back to waiting for K^2 and his numbers. If someone is interested in verifying my numbers, I would appreciate that very much, because, well, it look too optimistic to me.

Very nicely done. I'll be more than happy to check your numbers once I have some time. I'm not too sure about having the magnetorquer fold out like you said, as it would be exposed to radiation and potentially complicate thermal management. It's definitely a cool idea though. Again, when I get some time, I'll look into how fast we can precess the satellite once it is already rotating.

Having had a look at passive spin stabilisation, I managed to remind myself that spin around the major axis of rotation ( the axis of rotation with the highest angular momentum) on a semi-rigid body (like any spacecraft, including our cubesat) is asymptotically stable. If there is any spin around one of the other axes of rotation that results through a disturbance of some sort, this spin will eventually dissipate, again leaving us with just major axis spin. Considering we'll already be spinning up our sat to high speeds, this sort of passive spin stabilisation will be best. The questions then are time and power requirements for:

1) Initial spin up

2) Latter spin up/down if necessary

3) Pointing at the Sun or ground station, once the satellite is already spinning.

MBobrik has already given a pretty good answer to 1 & 2, and I'll have a look at 3 later today.

[MBobrik]

MBobrik has already given a pretty good answer to 1 & 2, and I'll have a look at 3 later today.

Actually, it's just vector addition. changing spin axis is no different from spinning up and down.

[henryrasia]

MBobrik, wow. Just wow o_0 . I guess I should take a crash course on aerospace engineering these vacations, I'm sure I'll find one somewhere... (any aerospace engineers out there who can advise me? )

Now, onto the maths:

three orthogonal circular coils

That's unorthodox indeed. What does that even mean? I'm a visual kind of guy, so the lack of 3D models is making this thinking quite a challenge of imagination

change the spacecraft's axis of rotation within a few minutes, and thus be able to reorient the sat from its usual sun basking orientation, do high speed transmission via a a directional high gain antenna, and then turn back

That sounds awfully complicated to me, and is it really necessary? Our plan is to have the thing spin continuously for the duration of the moss's life. This means there will be an axis of rotation that, thanks to space weightlessness and vacuum, can be ANY axis! So we can get creative with exactly what axis will it be spinning around in order to get our solar panels, antennae, and cameras placed in equally creative positions to work their best. Again, 3D visualization would be very helpful here.

51 RPM/min

Really? I think that's really overkill. According to this artificial gravity calculator we'd only have to increase 20 RPM every "step", assuming we start with 0.1 g and work our way up, as was the experiment iirc. Again there's the discussion if we should add any deployable solar panels / directional transmitters along the rotation axis and have them "counter-rotate" to stay still in relation to the sun or ground station.

coils with diameter 2.25 times larger than the cube's side

Weren't the magnetotorquers supposed to double as our heater for thermal control? If so shouldn't they be inside? Now that I think of it, how will we fit them inside a 1L cube? Also, are we going to have them for all 3 axes? or just the main one? Because if it deploys tumbling then we wouldn't be able to correct it, which could prove disastrous.

Now to answer bounding star,

what sort of sensors are we looking at? what if it separates from the launcher at night? also, how will it separate?

Sensors: That's a question for Mazon del and his professor Luis, who knowabout the experiment. I personally would guess CO2/O2/sucrose concentrations and maybe pressure just to be safe.

Launch at night: The battery should already be made to survive orbital nights, so as long as we can get it up there charged, and maybe leave the sat on sleep mode just the first orbit, we should be safe.

Separation: Don't quote me on this, but from what I remember personally it can either be sent to the ISS and launched from there or it can ride along a fairing or outside of the ship as a counterweight of sorts that the gets ejected into space. This last option they don't let the sat be on until after the main rocket has ejected it and is far away.

PS: Sorry I can't answer the other posts that came before this one, it's really late where I live *yawns* I'll get to them tomorrow

Edited December 4, 2014 by henryrasia

[LordQ]

Actually, it's just vector addition. changing spin axis is no different from spinning up and down.

Sadly not. We don't want to change the spin axis, we want to change the direction the spin axis points. Which we need to do with precession. Which isn't a difficult analysis given all the inputs at all, but it's not very intuitive. I just did a course that dealt with this sort of thing, so I think I understand it well enough to attempt an analysis. Later today once I get home.

[K^2]

MBobrick, did you compute that with assumption that you basically pack a certain volume/weight fraction with coil? That's... Kind of insane for Q = 1. But it's nice to know what the upper limits are, at least. Well done.

Keep in mind that we can't simply pack the whole thing with a single coil. We will need multiple coils with different directions. I was thinking of winding a coil along each of the faces. That would give the sat 3-axis control with some redundancy. I was also looking at a much lower weight fraction. Finally, 10W is definitely doable in bursts, but it will cook experiment if used for any significant amount of time, or require a lot of extra power to pump that heat out.

All in all, you can probably see how that 1 minute quickly becomes closer to 1 hour for something a bit more practical.

what sort of sensors are we looking at? what if it separates from the launcher at night? also, how will it separate?

We'll have magnetic field and GPS to work with even if the sat is totally blind. The separation is up to the launch provider. They usually have launch tubes that simply kick out 1-3 cubes when target orbit is reached.

[bounding star]

i found this, we could use something like this for the experiment sensors http://www.consumerphysics.com/myscio/scio.htm

[LordQ]

Alright, my report from what I've investigated. Your calculations seem to be alright MBobrik, but some of the underlying assumptions don't quite work. I think it would be unreasonable to have a magnetorquer the coils of which are larger than our spacecraft itself, especially since it wouldn't then have a mass ratio of 5%, as K^2 pointed out. Pretty high amounts of power would then be needed to push enough current through all that coil as well, which would lead to heating issues. Instead, I think it would be best if we bought or made a magnetorquer like this one.

So going off the specs for that magnetorquer, I did a quick calc going off its magnetic moment (mu) spec of 0.2Am^2. I then used T = mu*B, where T is torque and B is Earth's magnetic field strength. I used MBobrik's value of B for my calcs, which was 4e-05 T. Thus I got T = 8e-06 Nm. I then calculated moment of inertia in the same way as MBobrik, though I upped the mass to 1.3kg, as I think that's more accurated. This allowed me to calculate angular acceleration, which is just 0.0037rad/s^2, or about 2.1RPM/min. Going off the 60 or so RPM we need to achieve 0.16g (given a radius of 4cm), this means it'll take approximately 28 minutes to achieve this spin rate. Certainly a long fraction of our orbital period, but not unreasonable.

So next I went to analyse how long it would take to change the orientation of the spin axis. However, the analysis technique I know makes the assumption of instantaneous impulses, like you might see with monopropellant thrusters, and the magnetorquer simply isn't powerful enough to be considered capable of providing "instantaneous" impulses. So I need to figure out how to analyse the precession that results caused if the magnetorquer is kept on for a given period of time. I'll get back to you guys when I figure it out.

[K^2]

Oh, precession stuff is pretty easy. Picture the angular momentum vector as radius of the circle, and torque moving it along the circumference. So given some torque T, to change angular momentum L by angle phi (in radians), you need to apply it for t = T/(L * phi). So a 90° turn takes pi/2 times longer than it would take to spin up the sat to that RPM rate to begin with.

This works just like changing inclinations. The difference is that you can't exploit law of cosines, because we need to maintain constant angular velocity for the experiment, so you have to go with normal torque.

All in all, if we stick with a dipole antenna, this limits communication windows to an hour or two just after sunset or just after sunrise. Making major adjustments on the day side just doesn't make sense in terms of power usage and heat production. But on the night side, we'll probably be using the coil heat to maintain temperature anyways, so we might as well make attitude corrections.

This might still be a better option than omnidirectional. After all, the sat will pass terminator twice in every 90 minutes. If we have more than one station around the world, and we get something like ISS orbit to start with, this is plenty. Of course, all of these will have to be scheduled in advance. But this is just to beam data down. Sat will be able to receive messages in any orientation day or night.

[Endersmens]

So happy to see this still kicking!

So, I still don't have a computer. Still no .25 for me. Still no .24 for me actually. I think i'm gonna see if I can work on the site from my school computer. Can someone bring me up to speed? Anything Important happen that I should know about?

[Mazon Del]

This would require too much continuous free space for the camera to move around. I would suggest, we stick the sample capsules together like beads on a string, and then we would just move them around to the camera back via pulleys. Requires much less continuous free space.

Hmm. We could do something like that, it is probably easier to make a ring with a gear tooth placed on it and then a small motor to rotate it around. I wouldn't do the beads on a string idea, mostly because that sort of problem is already difficult enough on Earth (for predicting where very piece is) which means it will be harder up in space where a bit of it scrunching up means the sat is unbalanced. A rigid platform of some sort that can rotate would be best.

[MBobrik]

Seem I screwed up the constants last time, and LordQ got a completely different number.

So I will have to re-derive it from scratch, and LordQ will check.

let's define first a few parameters

a maximum angular acceleration

Lc diameter of the circular coil, side of the rectangular coil

r radius of the coil defined simply as Lc / 2 for both shapes

Sc = coil enclosing area

Cc = coil circumreference

L diameter of the satellite

Mt mass of the torquer

Ms = Mt / 3 mass of a single coil

Mc mass of the rest of the satellite

M mass of the satellite

Ix x component of torquers inertia momentum

Iy y component of torquers inertia momentum

Iz z component of torquers inertia momentum

It inertia momentum of the torquer

Ic = 1/6 * Mc * L^2 is inertia momentum of the rest of the satellite

f = Mt / Mc mass fraction of the torquer

N number of turns

A wire cross sectional area

Sw = N * A total coil cross sectional area

P max coil power

I max coil current

R coil resistance

B magnetic field

t torque

ÃÆ’ resistivity of the coil

ÃÂ density of the coil

q = It / Ic ratio

Q torquer to sat match value

----

now, let's define a few custom fancy constants

S1 = ratio of enclosing area of the basic shape to enclosing area of circular coil of the same dimensions

S2 = ratio of circumreference of the basic shape to circumreference of circular coil of the same dimensions

of course circular coil will have S1 = S2 = 1

but rectangle got area Sc=Lc^2 where Lc = 2*r, thus Sc=4*r^2 and relative to circle's À*r^2 we get S1 = 4/À

similarly rectangle got circumreference Cc=4 Lc where Lc = 2*r, thus Cc=8*r and relative to circle's 2*À*r we get S2 = 4/À

S3 = constant from the inertia momentum formula

for circle we google standard values Ix = Iy = 1/2 Ms*r^2, Iz = Ms*r^2, now we consider 3 coils 1/3 of total mass each, oriented towards all 3 axes,

and we thus get

It = Ix + Iy + Iz = 1/2 * Mt/3*r^2 + 1/2 * Mt/3*r^2 + Mt/3*r^2 = 2/3 * Mt*r^2 and thus S3 = 2/3

for rectangle we google standard values Ix = Iy = Iz = Ms*r^2, now we consider 3 coils 1/3 of total mass each, oriented towards all 3 axes,

and we thus get

It = Ix + Iy + Iz = Mt*r^2 and thus S3 = 1

--------

Now,

t = B * I * N * Sc

P = I^2 * R

Sc = S1 * À * r^2 because of how we defined S1

t = B * I * N * S1 * À * r^2

R = ÃÆ’ * N * Cc / A

Cc = S2 * 2 * À * r because of how we defined S2

R = ÃÆ’ * N * S2 * 2 * À * r / A

P = I^2 * R

P = I^2 * ÃÆ’ * N * S2 * 2 * À * r / A

I = √( P * A / (ÃÆ’ * N * S2 * 2 * À * r) )

we plug that into t formula

t = B * √( P * A / (ÃÆ’ * N * S2 * 2 * À * r) ) * N * S1 * À * r^2

= B * S1 * √( P * A * N * À * r^3 / (ÃÆ’ * S2 * 2) )

= B * S1 * √( P * Sw * À * r^3 / (ÃÆ’ * S2 * 2) )

= B * √( P ) / √( ÃÆ’ )* S1 / √( 2 * S2 ) * √( Sw * À * r^3 )

It = S3 * Mt * r^2 because of how we defined S3

Mt = ÃÂ * Sw * Cc * 3 because we have 3 coils

= à* S2 * 6 * Sw * À * r

Mt = It / ( S3 * r^2 )

It = q * Ic

Mt = q * Ic / ( S3 * r^2 )

q * Ic / ( S3 * r^2 ) = à* S2 * 6 * Sw * À * r

q * Ic / ( S3 * à* S2 * 6 ) = Sw * À * r^3

t = B * √( P ) / √( ÃÆ’ )* S1 / √( 2 * S2 ) * √( q * Ic / ( S3 * à* S2 * 6 ) )

= B * √( P ) / √( ÃÆ’ * à)* S1 / √( 12 * S2^2 * S3 ) * √( Ic ) * √( q )

= B * √( P ) / √( ÃÆ’ * à)* S1/S2 / √( 12 * S3 ) * √( Ic ) * √( q )

a = t / ( Ic+ It) = t / ( 1 + q ) / Ic

a = B * √( P ) / √( ÃÆ’ * à)* S1/S2 / √( 12 * S3 ) * √( Ic ) * √( q )] / ( 1 + q ) / Ic

= B * √( P ) / √( ÃÆ’ * à)* S1/S2 / √( 12 * S3 ) / √( Ic ) * √( q )] / ( 1 + q )

= B * √( P ) / √( ÃÆ’ * à)* S1/S2 / √( 48 * S3 ) / √( Ic ) * 2 * √( q )] / ( 1 + q )

we set Q = 2 * √( q )] / ( 1 + q ) so we get

a = √( P ) * B / √( Ic ) / √( ÃÆ’ * à)* S1/S2 / √( 48 * S3 ) * Q

which we can split into

input parameters part √( P ) * B / √( Ic )

coil material part 1 / √( ÃÆ’ * à) and

geometry part S1/S2 / √( 48 * S3 ) * Q which can be further split into

shape part S1/S2 / √( 48 * S3 )

match part = Q

--------

q = It / Ic

= ( S3 Mt * r^2 ) / ( 1/6 * Mc * L^2 )

= 6 / 4 *S3 * Mt /Mc * Lc^2 / L^2

q = 2/3 *S3 * f / (1-f) * (Lc/L)^2

and optimal Lc/L where q = 1 will thus be

Lc/L = √( 2/3 * ( 1 / f - 1 ) / S3 )

which is 3.559 even for S3 = 1 and f = 0.05

so we won't get Q = 1 under any reasonable circumstances

now, let's take the opposite approach - fix Lc/L = 1, coil fits into the sat exactly and let's see what performance we can get out of this geometry

and compute it for f = 1 %, beryllium coil and two coil shapes

Now, let's put all the numbers together

B = 4e-5 T

Mc = 1.3 Kg (total mass will be thus 1.313 Kg )

L = 0.1 m

P = 10 W

ÃÆ’ = 3.6e-8 Ωm

ÃÂ = 1690 kg/m^3

Computing the geometry independent parts we thus get

a = 0.348393 * S1/S2 / √( 48 * S3 ) * Q

and computing the geometry factor we get for the circle 0.035178 and 0.035003 for the rectangle.

So the circle is marginally better, but would take some place inside, so we take the rectangle.

a = 0.0122558 rad/s^-2 = 44 RPM/min

which is unexpectedly high, and thus either I made some mistake, or just fiddling with the parameters, we can get an order of magnitude more from the torquer.

I spent three hours writing this together, my fingers hurt. It turns out, rocket science is hard, even when no rockets are directly involved.

Edited December 4, 2014 by MBobrik

[MBobrik]

This works just like changing inclinations. The difference is that you can't exploit law of cosines, because we need to maintain constant angular velocity for the experiment, so you have to go with normal torque.

Wait a minute. that would just change the general direction of the rotation axis in space. Would it equally change the direction of the rotation axis relative to the satellite as LordQ pointed out ?

All in all, if we stick with a dipole antenna, this limits communication windows to an hour or two just after sunset or just after sunrise.

If we can reorient the craft quickly enough, we can transmit directionally at any given point of the orbit. We will just to have to know in advance when. If we could send just twice per orbit, we would have to chase after the points around the world. If it can turn and communicate any time, each of us can connect with it twice per day without traveling too much ( some travel will still be involved to get the longest communication window )

EDIT : now I myself pushed my previous long and far more important post to previous page. @ LordQ, K^2 and anyone interested in the gory details of magnetotorquer theory, previous page, 174, the most bottom long post.

Edited December 4, 2014 by MBobrik

[henryrasia]

Yeah, my post was apparently swallowed by others in the previous pages... I would think it'd be a good idea for everyone to read all posts they missed, not just the latest page. Definitely for a project like this.

[MBobrik]

Yeah, my post was apparently swallowed by others in the previous pages... I would think it'd be a good idea for everyone to read all posts they missed, not just the latest page. Definitely for a project like this.

I think I answered some of your questions in the long post on the previous page.

[K^2]

If object is spinning about a maximum or minimum principal axis, applying normal torque causes precession of the craft. That is, axis of rotation will turn in space, but stay put relative to the craft. Think gyro. Any other possibility yields tumbling, which we want to avoid once experiment is running.

[Nicholander]

Maths

That's a lot of maths. Thanks MBobrik, even though I unfortunately don't understand like 95% of that.

Anyway, thanks to your maths we got our number, 44 RPM/min. But well, LordQ said it would take about 28 minutes to spin up to 0.16G if it was at 2.1 RPM/min, so how long would it take at 44 RPM/min?

Also, henryrasia, how's the little experiment with the moss going?

[MBobrik]

If object is spinning about a maximum or minimum principal axis, applying normal torque causes precession of the craft. That is, axis of rotation will turn in space, but stay put relative to the craft. Think gyro. Any other possibility yields tumbling, which we want to avoid once experiment is running.

So it is really that simple as I initially suspected. I used the maths about rotating objects long ago ... LordQ got me confused a little.

Then I don't see why we should not go for the megatorquer from the previous page

total weight 13 g, 3 coils wound directly around each mid section of the sat surface, each 33 turns of 0.5 mm diameter beryllium wire, I = 2 A, U = 5 V

will get us 44 RPM/min

[Mazon Del]

We probably don't want to go with beryllium wire unless we get super over funded. A 1 METER segment of wire costs $785. http://www.sigmaaldrich.com/catalog/product/aldrich/gf52902840?lang=en&region=US

[MBobrik]

We probably don't want to go with beryllium wire unless we get super over funded. A 1 METER segment of wire costs $785. http://www.sigmaaldrich.com/catalog/product/aldrich/gf52902840?lang=en®ion=US

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

[henryrasia]

Also, henryrasia, how's the little experiment with the moss going?

It's Mazon Del who has access to the moss samples. So I forward the question to him. Can you do it? Like an IR camera filming a small sample of the moss (doesn't need to be fancy at this point) for a few days with the most FPS (one frame picture per hour?) So we can then see if that's overkill and, if it is, speed it up through post-editing to see what's the sweetspot of least FPS with good video for analysis.

MBobrik, you seem to have overlooked over my main concerns:

1) That's unorthodox indeed. What does [orthogonal circular coils] even mean? I'm a visual kind of guy, so the lack of 3D models is making this thinking quite a challenge of imagination

2) [stopping the spin, transmitting, and spinning back up again] sounds awfully complicated to me, and is it really necessary? Our plan is to have the thing spin continuously for the duration of the moss's life. This means there will be an axis of rotation that, thanks to space weightlessness and vacuum, can be ANY axis! So we can get creative with exactly what axis will it be spinning around in order to get our solar panels, antennae, and cameras placed in equally creative positions to work their best. Again, 3D visualization would be very helpful here.

3) [50 RPM/min] Really? I think that's really overkill. According to this artificial gravity calculator we'd only have to increase 20 RPM every "step", assuming we start with 0.1 g and work our way up, as was the experiment iirc. Again there's the discussion if we should add any deployable solar panels / directional transmitters along the rotation axis and have them "counter-rotate" to stay still in relation to the sun or ground station.

4) [Magnetotorquers outside of the cube] Weren't the magnetotorquers supposed to double as our heater for thermal control? If so shouldn't they be inside? Now that I think of it, how will we fit them inside a 1L cube? Also, are we going to have them for all 3 axes? or just the main one? Because if it deploys tumbling then we wouldn't be able to correct it, which could prove disastrous.

[rhoark]

Rotation about the intermediate principal axis of an asymmetric body is an unstable equilibrium, so the axis of rotation will be changed with the slightest perturbing force. In theory, carefully controlling the mass distribution could allow controlled changes of the rotational axis without anywhere close to the energy expenditure of applying torque directly against the undesired axis. When you're done you just make yourself axisymmetric again. I can't tell from the last few pages what the motivation for changing the axis is though.

Here's an animation, and further in a link to some equations: http://fouriestseries.tumblr.com/post/91685028535/rotational-stability