[PLUG-TALK] Mars lander Curiousity, OMSI Sunday Night Aug 5

[Keith Lofstrom]

On Sat, Aug 4, 2012 at 10:27 AM, Keith Lofstrom <keithl at gate.kl-ic.com> wrote:
> Sunday is Astronomy Day at OMSI, http://www.omsi.edu/node/3420
>
> The Curiousity Mars Rover lands at 1030pm Sunday, and OMSI is
> opening the doors at 9pm for live coverage from JPL.  Or as
> live as it gets with a 15 minute ping time to Mars.
(oops .................. 28 minute .........)
> The landing is one of the most elaborate kludges I've ever seen.
> It will be a miracle if it works.

On Sat, Aug 04, 2012 at 10:56:29AM -0700, Ronald Chmara wrote:
> They lost me at the "sky-crane" portion of the landing. Too many
> points of untested technology, let alone autonomous untested
> technology.
> 
> I'm going with "it'll be a miracle if we learn anything new from this".
> 
> For those who haven't been following:
> http://www.youtube.com/watch?v=Ki_Af_o9Q9s

Agreed.

I'm more worried about the 500,000 lines of code.  Imagine a 
coder finding a bug, 30 minutes before the landing ...

A former boss used to say, "too many things that oughta work is
a not-oughta work".  TJ Rodgers, founder of Cypress Semiconductor,
taught that you make the fastest progress by taking your current
product and changing one thing, then measuring what happened.  If
you change two things and fail, then you spend more time pointing
fingers and dodging blame than you spend on debug.  When the test-
and-evaluate loop is optimized, you can repeat it much more quickly,
much faster than the evaluate-two-things loop, even if both succeed.

For me, the important question is whether the system has Really
Good telemetry, high bandwidth, multimeasurement, and continuous,
all the way down.  When one of the many steps doesn't work, will
we even find out which step failed, much less exactly why?  It
doesn't matter too much if this attempt fails, as long as we
learn enough to make every attempt after this one succeed.  If
we can't learn anything, there should never be a second one.
Doing the same thing blind, twice, is insane.

If I was designing this system, I would have the Ka band (32GHz)
receiver and the 3 meter dish on the Mars Reconnaisance Orbiter
collecting telemetry overhead.  Small rocket launchers on the
reentry system would fire kilogram-sized "data transmitter pods"
vertically, out of the reentry ionization trail, every 20 seconds
or so.  Each pod could broadcast a 10 watt beam, 100 megabits or
so of recorded telemetry data during the few minutes before it
hits the ground.  It might take a while to send all the bits
back to Earth from the orbiter, but we would learn a heck of
a lot about what went wrong.

If the MRO can't be made to do that, I would co-orbit some
upgrade packages that can, before I attempted a crazy complicated
landing.  The MRO does operate as a relay for landers, so it
should be able to do that for small relay satellites optimized
for listening to "son of Curiosity".  You could test out the
relay system and telemetry pods in advance, with a few hundred
kilograms of stuff, before you send the one ton kludge.

Keith
-- 
Keith Lofstrom          keithl at keithl.com         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs