[Ground-station] Satellite program

[Phil Karn]

On 5/16/18 22:08, Bruce Perens via Ground-Station wrote:

> So, cubesat folks have been creative in finding radiation-tolerant parts
> in consumer or industrial grades.

Radiation really isn't a serious problem in LEO. It *is* a problem in
higher orbits, especially the elliptical orbits for the Phase III
program. That's where the 1802s you mention were largely used.

You are correct that non-rad-hard RAM with scrubbing was used. This
worked for about 3 years on AMSAT-Oscar-10 until the error rate
overwhelmed the ECC.

You are also correct that latchup need not be avoided if 1) it is rare
and 2) a mechanism is provided to keep it from doing any damage. This
can be as simple as a self-resetting (solid state) circuit breaker.

> The Cortex M0 is more than enough for housekeeping and might be enough
> for some signal and image processing tasks. But the housekeeping CPU
> need not be the signal-processing CPU as well.

Housekeeping and even attitude control doesn't require much CPU at all.
Remember how primitive the 1802 was...

> set it running. Nobody's told me, but if there was any cryptography on
> that it wasn't much more than exclusive-OR of a secret word.

They were simple LFSR sequences, plus a hefty dose of
security-through-obscurity. Go look at the Phase III schematics. The
pages with white-on-black title boxes were considered "sensitive"
because they depicted command receivers or decoders. The kick motor was
also armed and fired with its own LFSR sequences, decoded in hardware.
This was not a bad way to protect against an insane computer if not
deliberate attack by a sentient party. (The LFSR sequences were not
present anywhere in computer memory at launch, which is when safety was
most important. Remember these were secondary payloads.

I still consider it incredible that mere hams were allowed to build
their own satellites with built-in rocket engines using hypergolic
propellants and tack them on the sides of real space-going rockets
carrying some seriously expensive commercial and scientific payloads.
IIRC, the delta-V capability of a Phase III was on the order of 1500
m/s. The muzzle velocity of an AR-15 bullet is about 1000 m/s, and it's
a lot lighter.

> They have
> their own FORTH-like language, first written in the '70's, which does
> concurrent but not parallel threads. Most housekeeping is written in
> this language.

That's IPS, Interpreter for Processor Structures, the brainchild of Karl
Meinzer, DJ4ZC. Threading was fully cooperative, with each thread polled
sequentially to see if it had something to do. It wasn't bad for its
time: highly compact, interactive, and easily modified on the fly.


Phil