On 7/9/07, Chuck Green <greencl@mindspring.com> wrote:

Hi Guys,

Juan has done a lot of outstanding work which resulted in some
substantial critiquing of the CAN-Do!  (Affectionately called a
"widget.")  It is unfortunate that it has taken several years since the
CAN-Do! was designed and then 100 units built before an application of
sufficient sensitivity used it to discover it's shortcomings.  History
can provide lessons that I hope we can learn from, but it seldom
provides solutions to the problems encountered.  Lyle and I have
exchanged a few thoughts privately and it seems it may now be time to
consider solutions to the problems found.

The only practical way to accomplish this is to develop the next
generation CAN interface device.  Dare I call it the CAN-Do-Too!  ?

All technical specifications should remain the same.  What this really
means is that a next-generation controller must run *exactly* the same
firmware currently running the CAN-Do! .

All specifications added or redefined should be carefully defined and be
measurable.

General specifications that we worked from before were that the widget
should use as little power as possible and consume as little of a
module's volume as possible.  The first of these should remain the same,
"use as little power as possible."

But the second should be changed to "consume as little of the connector
panel space as possible" even if it means consuming a little more of the
module volume.  This means the widget PCB and components should not
extend beyond the dimensions of the DA-15P connector in either
dimension.  A possible compromise to this would be to let the PCB run
past one end of the DA-15P but not more than the DA-15P is forced away
from the side of the box by the box design.

The power supply could be completely redesigned.  Or the inductor of the
existing supply could be exchanged for one that is a toroid (the
existing one is not).  If someone wants to step up and design a new
power supply, great!  If not, then we would simply change the inductor.
I'd sure like to see someone take this on.  With so many of these in the
satellite, only a few milliwatts is important.  And the noise issue Juan
uncovered is *very* important.

It may be that some, or maybe all, of the widget should be enclosed in a
metal box.  It may be that just changing the inductor would allow a new
widget to meet the yet-to-be-defined noise specifications.

The input power filter for module power should be separate from the
widget power supply input filter.  The module power filter is a filter
that will not meet all module requirements, but would likely meet the
requirements of a digital module.  Some modules, such as receivers, may
need additional power conditioning.  But in any case, the widget power
supply should not add to the module power supply noise.

There should be a simple way to disconnect the filter capacitors on the
widget from the data lines when the widget is in Byte mode.  Most people
are not aware of this problem which was uncovered by another module
builder.  It only effects those using the CAN-Do! in Byte mode.

Using a synchronizing signal does not seem practical to us.  It would
complicate the design of the widget power supply so that it would
function with or without the presence of the synchronizing signal (we
don't want to introduce a single-point-of-failure).  It would
dramatically increase the satellite wiring harness complexity, something
the widget was intended to a simplify.  And it would inhibit the widget
power supply from going into various power-saving modes.

Recruit some new people into this project.  Lyle simply doesn't have any
time for doing new designs right now.  We need a power supply designer
as stated above.  And my time is limited.  I'm willing to lay out the
new design and build a few prototypes, but I need others to do parts
procurement and volume building of widgets.  I'll prepare flight units
if desired.  We've talked about having someone skilled in parts
procurement before but I don't know of anything having come of it.  The
bottom line, if this is going to happen, Lyle and I need others to be
involved.

If you think this is a good idea, or bad, please express yourself.  And
if you have other comments to add to the above, or would like to
modify/expand on above comments, please do so.

Looking forward to your comments,

Chuck and Lyle
_______________________________________________
Via the Eagle mailing list courtesy of AMSAT-NA
Eagle@amsat.org
http://amsat.org/mailman/listinfo/eagle