[Tig] Spirit maintenance costs

Martin Euredjian ecinema
Sat Feb 22 07:23:27 GMT 2003

> Let's not be too hard on Mr. Frey. He does have a point.

Just to be clear.  The intention wasn't to come down on John Frey for the
opinion he voiced.  Not at all.  I apologize if it could have been read that

> The CCD's used in the Marconi's, the FDL's and the ADS were all made in 
> the US by Fairchild.
> They were, in fact, remarkably reliable solid state devices.

Let's talk CCD's for a moment.  I know the guy who probably designed those.
I've taken classes from him.  In fact, he's probably had his hands on most
space-borne imagers.  He designed the Hubble's cameras, for example.  CCD's
are remarcably crude devices and, at the same time, remarcably beautiful for
the simplicity with which can do what they do.  CCD's have different failure
modes when compared to I.C. because they are made differently.  I'm speaking
primarily of CMOS technology when referring to I.C.s, which is the prevalent
technology for integrated circuits.  You could almost literally make CCD's
in your garage, while, if you attempted to make CMOS circuits you'd risk
killing everyone in your neighborhood.

CCD's achieve their magic by shoving electrons around by using carefully
crafted phased signals that move a potential wave across the surface of the
device.  The "wells" everyone talks about are like "force fields" created by
thes waves, the biasing of the photodiode, the physics of the material
(different dopings will produce different results) and, ultimately the
desired geometry.

Now, a Marconi, 15 years ago, to output NTSC or PAL did not require a CCD
that broke significant performance barriers.  CCD's have not fundamentally
changed at all in 20 years, maybe even since they were invented back in the
60's if I remember correctly. In great contrast to that, a Spirit's line
arrays are operating at the edges of what is attainable with CCD technology.
The one problem that CCD's have is that there's real limit to how fast you
can go before the biasing problems become very significant.  The rise times
required and the raw power that needs to go into the device can and will
affect longevity and reliability.  Noise is another factor.  While we know
how to manufacture CCD arrays that are, for most practial purposes, perfect
charge transfer devices (in excess of 99.9999% charge transfer efficiency)
the bummer is that the noise performance for all CCD's come down to one and
only one transistor (a FET) that buffers the 0 to, say, 60,000 electron
signal into a couple of volts to drive the output pin.  Virtually all of the
noise comes from this FET.  That's Nature keeping us in check!

Regarding the use of CCD's in space.  The great majority of space-borne
imagers, by far, have no need to operate at television rates.  All of them
are custom designed from a mission specification perspective, and very few
of them operate in the visible spectrum.  In most cases the guiding
parameters are things like noise performance and dynamic range, all of which
benefit from operating at seconds-per-frame as oppossed to
frames-per-second.  The great problem in space-borne CCD's is, of course,
pixel damage.  It is impossible to radiation-harden CCD's.  That's why
there's a huge push to get CMOS imagers up to speed, CMOS can be
radiation-hardened.  It also offers the amazing possiblitiy of integrating
various pixel-level processing circuits right on the imager.  My
aquaintances in the aerospace imaging field tell me that we (as in the
government/military, not consumers) may not be too far away from a viable
solution in CMOS ... which I translate into: we are already flying some
pretty cool stuff, but we can't tell you about it!

Martin Euredjian
eCinema Systems, Inc.
voice: 661-305-9320
fax: 661-775-4876
martin at ecinemasys.com

More information about the Tig mailing list