Pegasus/SciSat Launch Cost

Derek Lyons wrote:
Ian Stirling wrote:
If you just allow the addition of sun sensors, and magnetometers, even
basic commercial grade $5 each accellerometers, magnetometers and cameras
are perfectly adequate to get you into some sort of orbit.

The problem is... Almost nobody wants 'some sort of orbit'.
Instrument performance, antennae coverage depend on altitude, thermal
environment depends on altitude and sun angle...

If you then add a GPS reciever, it can be as accurate as the gold-standard
one, for a very small fraction of the mass and cost, with the caveat that
if GPS goes off during the launch, it probably won't make the orbit you
want.

AFAIK there are no GPS receivers outside of government hands that will
perform in an accelerating booster. (There are hard limits inside
commercial chips specifically to prevent provide cheap guidance
systems to folks we'd rather not have.)

The limits of the GPS chipsets arn't really that hard.
A lot of it is software.
Some of the lower level chipsets that are more or less only front
ends have little inherent limitations.
And even building the whole ball of wax from scratch is getting significantly
less complex year on year as electronics gets faster.
However, I was considering the case where this was an approved attempt,
and you could get one of the existing vendors to twiddle a couple of
bits, to make it easier.

--
http://inquisitor.i.am/ | | Ian Stirling.
---------------------------+-------------------------+--------------------------
All I want is a warm bed, a kind word and unlimited power -- Ashleigh Brilliant.

George William Herbert wrote:
Ian Stirling wrote:
George William Herbert wrote:
Alex Terrell wrote:
Aren't avionics mostly electronics - in which case why aren't they all
on a $100 chip?

Even if you use MEMS gyros and acellerometers it's not quite that
compact yet.

It's not far off.

It's further off than you think, unless there have been some
advances in the last week or so.

The guidance package can probably be fit in 10g, with quite a lot of
effort, I think to go an order of magnitude lower would need at the least
making it into one multichip module.
100g is easy, and possibly achievable using just standard parts, without
much design.

A guidance package the size of some chips (say the same size as the
athlon processor I'm using to compose this post) is quite possible.

MEMS by itself isn't good enough yet. You really want FOG, and those
are kilogram by the time you have all three axies going.

MEMS magnetometers can easily get to a degree resolution, sun sensors
can almost trivially get a quarter of that, without trying hard.
This gets you attitude information (with a model of the magnetic field,
and some assumptions about trajectory, magnetic field measurements around
the trajectory before liftoff may be helpfull)

MEMS gyros and accellerometers are comparatively bad.
MEMS accellerometers, neglecting attitude errors and not looking further
than ADs web page, seem to be able to give under 10m/s at the end of
a 5 minute flight. For an error of +-2Km or so.
MEMS gyros have a place, but more for attitude stabilisation than
determining long term attitude.
Assuming that the rocket is around a thousand kilometers downrange at
MECO, that ~1 degree attitude error gives an error of +-16Km.
The 10m/s error results in around another +-15Km or so uncertainty in
perigee.

This is fine for some things, though completely unacceptable for others.

Aside from that, the best way would be to get the volume up, which
won't happen till the price comes down.

Or, design to lower priced components from the beginning...

In some cases it's a case of relaxing the specs, and getting by using
combinations of sensors.

The gold standard is certainly an inertial guidance system that can
get to a tightly specified orbit without needing any outside help other
than its position before launch.

This pretty much requires fiber optic gyros and other complex items,
which are neither light, or cheap, some of which for pretty fundamental
reasons.

Oh?

FOG are light, cheap, etc by comparison to old gyros.
$10k range for a good unit, a bit over a kilogram, etc.

Sorry, I was using cheap and light in comparison to MEMS devices,
in answer to the "$100 chip"

However, while this may be entirely appropriate for some things
(ICBMs where it may be launching through a nuclear fireball, and maybe
manned flight), is it really needed in all cases?

I think you want to study the problem a bit more.

COTS IMUs and GPS are just fine for the mission, now.
Really.
snip
I totally agree, for some missions.

And possibly also design and manufacture.
Someone who makes trumpets for a living just might know how to get

On a hundred thousand pound thrust rocket motor?

Do you know how *big* those are?

Yes, I was just meaning that other fields might have things to add,
rather than just looking to find out how boeing/... have done it, the
particular example given was hyperbole, rather than a realistic
expectation.

--
http://inquisitor.i.am/ | | Ian Stirling.
---------------------------+-------------------------+--------------------------
Lord, grant me the serenity to accept that I cannot change, the
courage to change what I can, and the wisdom to hide the bodies
of those I had to kill because they ****ed me off. - Random

Colonel K wrote:

"Ian Stirling" wrote in message
snip
A guidance package the size of some chips (say the same size as the
athlon processor I'm using to compose this post) is quite possible.

Is your Atlon single event upset-resistant? Companies that build launchers
require that, not because it's nice, but because it's needed to successfully
complete the mission.

I have not actually looked at the datasheet of this athlon, however,
the AMD processor I was using three years ago could, in combination with an
identical processor detect errors, by one snooping on the bus, and raising
an error pin if the other processor was misbehaving.
Add a small circuit to power down both processors, and then power them
back up in the event of an upset, and software that knows it may have
to restart computation, and it's possible.
To make this into a complete system, you'd need SEU resistant RAM too,
which ECC RAM helps with, but does not quite get there.

Besides the guidance system, there's also avionics like command
receiver-decoders, telemetry transmitters, power controllers, flight
snip
Depends. Do you want to put your $300M communications satellite on an LV
that is built to be low-cost, not highly reliable?

No, of course not, one would have to be utterly barking mad to do so.
I should have made clear that by all cases, I was meaning to include very
small payloads and low cost rockets.

--
http://inquisitor.i.am/ | | Ian Stirling.
---------------------------+-------------------------+--------------------------
"When I use a word," Humpty Dumpty said in a rather scornfull tone, "It means
Just what I choose it to mean - neither more nor less." -- Lewis Carrol

(Derek Lyons) wrote in message ...
Ian Stirling wrote:

If you then add a GPS reciever, it can be as accurate as the gold-standard
one, for a very small fraction of the mass and cost, with the caveat that
if GPS goes off during the launch, it probably won't make the orbit you
want.

AFAIK there are no GPS receivers outside of government hands that will
perform in an accelerating booster. (There are hard limits inside
commercial chips specifically to prevent provide cheap guidance
systems to folks we'd rather not have.)


That may be true for fast-flying rockets. On the other hand, it is
interesting to see what *can* be done with cheap GPS. For example,
a model airplane reportedly flew 1,888 miles across the Atlantic
Ocean recently:

"http://story.news.yahoo.com/news?tmpl=story&u=/ap/20030812/ap_on_re_eu/ireland_model_flight_1"

I hate to say it but, times being what they are, this event is a
bit worrisome when you consider what a terrorist could do with a
slightly larger airplane model ...

- Ed Kyle

Ian Stirling wrote

Derek Lyons wrote:


AFAIK there are no GPS receivers outside of government hands that will
perform in an accelerating booster. (There are hard limits inside
commercial chips specifically to prevent provide cheap guidance
systems to folks we'd rather not have.)

The limits of the GPS chipsets arn't really that hard.
A lot of it is software.
Some of the lower level chipsets that are more or less only front
ends have little inherent limitations.

That's correct, (AFAIK, also). Several years ago some people interested
in missile proliferation (I think it was the CHOP operation at Kirtland)
looked into this and found that the altitude/velocity limitations in
a number of commercial GPS systems were implemented in the display
drivers, not the GPS engines themselves.

Ian Stirling wrote in message ...
Colonel K wrote:

"Ian Stirling" wrote in message
snip
A guidance package the size of some chips (say the same size as the
athlon processor I'm using to compose this post) is quite possible.

Is your Atlon single event upset-resistant? Companies that build launchers
require that, not because it's nice, but because it's needed to successfully
complete the mission.

I have not actually looked at the datasheet of this athlon, however,
the AMD processor I was using three years ago could, in combination with an
identical processor detect errors, by one snooping on the bus, and raising
an error pin if the other processor was misbehaving.
Add a small circuit to power down both processors, and then power them
back up in the event of an upset, and software that knows it may have
to restart computation, and it's possible.
To make this into a complete system, you'd need SEU resistant RAM too,
which ECC RAM helps with, but does not quite get there.

You would not know which processor was misbehaving, only that they
disagreed with each other. Instead of taking a power on reset after
the disagreement, you may want to take a high proirity interrupt and
attempt to execute a recovery routine and/or save the context of the
computation where the error occurred. Otherwise, you would not be
able to restart the computation after the power on reset if the
recovery routine failed.

This might work earthside where the SEU rate is low. The radiation
environment in space can increase the SEU rate noticeably for non
rad-hard parts. It would not be good if the above setup took a
disagreement error every few minutes or so. Also, consider radiation
damage can be cumulative, eventually causing a permanent instead of a
transient failure.

Rad-hard processors are COTS these days, just more expensive that non
rad-hard parts. Google "rad-hard processors".

Back on your original post of putting a guidance package using
standard part onto an MCM the size of an Athlon package "without much
design". I shudder at your last 3 words.

Mike Chan wrote:
Ian Stirling wrote in message ...
Colonel K wrote:

"Ian Stirling" wrote in message
snip
A guidance package the size of some chips (say the same size as the
athlon processor I'm using to compose this post) is quite possible.

Is your Atlon single event upset-resistant? Companies that build launchers
require that, not because it's nice, but because it's needed to successfully
complete the mission.

I have not actually looked at the datasheet of this athlon, however,
the AMD processor I was using three years ago could, in combination with an
identical processor detect errors, by one snooping on the bus, and raising
an error pin if the other processor was misbehaving.
snip
You would not know which processor was misbehaving, only that they
disagreed with each other. Instead of taking a power on reset after
the disagreement, you may want to take a high proirity interrupt and
attempt to execute a recovery routine and/or save the context of the
computation where the error occurred. Otherwise, you would not be
able to restart the computation after the power on reset if the
recovery routine failed.

You can, you just need to write everything so that it either stores all
state in error checked RAM, or never overwrites the information needed
for a computation with partial results.

If you add an external write wait state, you can catch errors before they
are written and restart.
Restart would be nothing like it would be in a PC, more akin to a task
switch, back to the beginning (or latest checkpoint) of the task that
was in progress when the event occured.

This might work earthside where the SEU rate is low. The radiation
environment in space can increase the SEU rate noticeably for non
rad-hard parts. It would not be good if the above setup took a
disagreement error every few minutes or so. Also, consider radiation
damage can be cumulative, eventually causing a permanent instead of a
transient failure.

I would be aiming for restart timings well under a millisecond, though it
may be a little longer for all the caches/... to totally fill.

Rad-hard processors are COTS these days, just more expensive that non
rad-hard parts. Google "rad-hard processors".

I totally agree, I was just surprised to find (when I looked at it a couple
of years ago) that it was possible at all.

Back on your original post of putting a guidance package using
standard part onto an MCM the size of an Athlon package "without much
design". I shudder at your last 3 words.

I seem to have missed "custom silicon" out there, the bits all
exist seperately, putting them together is 'just' a matter of wiring them
all together with some glue logic.

Anyway, the bits (accellerometers, magnetometers are on order, so we'll see.)
Unfortunately I currently lack a rocket to guide with them, I suspect
I'll be trying them for the (rather dissimilar) task of model aircraft
guidance to work the bugs out, at least until I've got some thrust vectoring
biprop engines with decent performance on the bench and reliable.

--
http://inquisitor.i.am/ | | Ian Stirling.
---------------------------+-------------------------+--------------------------
Tad Williams has an interesting new fantasy: http://www.shadowmarch.com/

Allen Thomson wrote:
Ian Stirling wrote

Derek Lyons wrote:

AFAIK there are no GPS receivers outside of government hands that will
perform in an accelerating booster. (There are hard limits inside
commercial chips specifically to prevent provide cheap guidance
systems to folks we'd rather not have.)

The limits of the GPS chipsets arn't really that hard.
A lot of it is software.
Some of the lower level chipsets that are more or less only front
ends have little inherent limitations.

That's correct, (AFAIK, also). Several years ago some people interested
in missile proliferation (I think it was the CHOP operation at Kirtland)
looked into this and found that the altitude/velocity limitations in
a number of commercial GPS systems were implemented in the display
drivers, not the GPS engines themselves.

For the pretty fundamental reason that you've got to have a position
solution in order to work out if you should blank the display...
(well, there are some limits, but only well above the posted limits)

Add to that that some recievers use well known microprocessors with
the capability to upload new firmware, and it may only be a matter of
a little hacking. (My GPS 12 uses a 386, for example)

--
http://inquisitor.i.am/ | | Ian Stirling.
---------------------------+-------------------------+--------------------------
Among a mans many good possessions, A good command of speech has no equal.

Ian Stirling wrote

Derek Lyons wrote:

AFAIK there are no GPS receivers outside of government hands that will
perform in an accelerating booster. (There are hard limits inside
commercial chips specifically to prevent provide cheap guidance
systems to folks we'd rather not have.)

The limits of the GPS chipsets arn't really that hard.
A lot of it is software.

Also see http://www.weblab.dlr.de/rbrt/pdf/IONNTM_02.pdf

The Allstar is representative of common single frequency receivers
for terrestrial applications and received no particular modification,
except for the disabling of the common altitude and velocity
constraints that would otherwise inhibit its use for space
application.

"Colonel K" wrote in message . com...
Besides the guidance system, there's also avionics like command
receiver-decoders, telemetry transmitters, power controllers, flight
termination system components, relays, and numerous components I can't think
of at the moment. The reason this stuff is expensive is the effort that goes
into engineering, development, manufacturing and qualification testing to
ensure it will function as a high reliability unit. Usually the hardware is
built to Mil Std 1540, not necessarily because a contract specifies it, but
because it makes good sense to do it that way. There's a reason why - that
Mil Std is based on lessons learned from fifty years of rocketry. It's when
someone tries to show they can do it cheaper that accidents result
(remember Delta III?).

I think a lot of the above (not all) could be digitised, which would
reduce cost and improve reliability. For telemetary transmitters,
modify a satellite phone.

Depends. Do you want to put your $300M communications satellite on an LV
that is built to be low-cost, not highly reliable?

If launch costs were lower, people would design cheaper, less relaible
satellites. Increased volume would again push the price down).


If you then add a GPS reciever, it can be as accurate as the gold-standard
one, for a very small fraction of the mass and cost, with the caveat that
if GPS goes off during the launch, it probably won't make the orbit you
want.

Precision delivery to a specific orbit is always desired by the customer.

Not needed (within a km or so) if you can correct the orbit at your
leisure.